mimecast

Mimecast introduces AI agent governance and managed threat response Mimecast has unveiled Agent Risk Center, a beta capability for discovering, monitoring, and governing AI agents, alongside Managed Threat Response, a redesigned 24/7 service that combines AI-assisted triage with analyst-confirmed remediation. According to Mimecast’s analysis, 98% of organizations already have unsanctioned AI tools in use, and by 2029 more than a billion agents will take some 217 billion actions a day1 on employees’ behalf with limited visibility for security tools. The Mimecast Agent Risk Center gives security and risk teams one place to see who is behind that activity and act on it, applying the same risk lens Mimecast already applies to human insiders. “AI agents are being deployed faster than security teams can see them,” said Rob Juncker, Chief Technology & Product Officer, Mimecast. “Most organizations don’t know what agents are running in their environment, what data they’re touching, or who’s accountable when something goes wrong. Mimecast Agent Risk Center changes that — same risk model, same controls, extended to every agent acting on behalf of your people.” Mimecast Agent Risk Center: Beta now open As enterprises deploy AI agents across email, collaboration, and business workflows, security teams face a growing risk surface with limited visibility into agent behavior, permissions, and data access. Mimecast Agent Risk Center gives security and risk teams a single point of control to discover, monitor, and govern AI agents operating in their environment. Key capabilities include: Unified Agent Dashboard — a single, real-time inventory of every AI tool and connection, sanctioned or not AI Rulebook — governance for classifying and enforcing policy by department, with no new deployment required Response Controls — desktop app blocking, browser upload and paste blocking, and in-context user nudges The Mimecast Agent Risk Center is in beta as a free, opt-in capability for active Incydr subscription customers, with early access in September 2026 and general availability for those customers planned for January 2027. Managed Threat Response: AI-accelerated triage, analyst-confirmed remediation Mimecast also introduced Managed Threat Response, a redesigned 24/7 managed service that combines AI and human expertise to address one of security’s most persistent problems: uninvestigated alerts. Research shows that 42% of security alerts go entirely uninvestigated.2 Mimecast closes that gap for emails flagged by users, taking on 100% of the queue and remediating confirmed threats. Mimecast AI triages every reported email. Mimecast Security Operations Center analysts then action confirmed threats — blocking senders or domains, removing campaigns tenant-wide, and updating detection logic based on confirmed outcomes. Customers see the outcome, not the work: no new console, no model to tune, no analyst headcount to fund. Detection logic improves continuously as confirmed threats are identified across Mimecast’s 42,000-customer base.

Stolen Meta and Google ad accounts are worth more than the money they hold Ad account theft, the systematic hijacking of Meta Business Manager and Google Ads accounts, has grown into a commodity-driven cybercrime economy complete with tiered pricing, escrow services, and money-back warranties for stolen accounts. Public reporting on this topic tends to focus on drained ad budgets, but according to Mimecast, that’s often a short-lived gain for attackers. “Both platforms operate on a pre-loaded credit or card-on-file model. Campaigns spend continuously until a budget cap, or the account holder intervenes. A newly compromised account with a $5,000 monthly budget can be drained in hours,” Mimecast explained. “Legitimate ad spend history raises account trust scores, meaning aged, active accounts can serve ads that pass platform safety checks that would reject a new attacker-created account. This is what makes the older, higher-history account worth a 2 to 4 times premium in the underground market,” they added. Age, spending history, verification status and daily spending limits determine the price, with Zscaler reporting stolen Meta Business Manager accounts selling for anywhere from around $15 to $340, while some Google Ads accounts for high-risk sectors have been listed for $200 to $270 on Telegram. A record high, right after the biggest crackdown The company’s Threat Research Team has tracked 6.4 million detections of Meta Business Manager and Google Ads account theft over four years, and the trend line is not moving in the direction anyone would want. The second half of 2025 produced approximately 1.86 million detections, a new high above every prior period. That volume came after some of the largest enforcement actions against this ecosystem to date had already concluded. The pattern holds throughout the four-year dataset. Takedowns and arrests produce a dip, then activity returns to equal or higher levels. Mimecast’s four-year detection telemetry for Meta Business Manager and Google Ads account theft campaigns (Source: Mimecast) Mimecast ties activity to Vietnam-linked malware families including DuckTail, NodeStealer, VietCredCare, and PXA Stealer, alongside separate operations traced to Brazil and Portugal and to China and Hong Kong. When the PXA Stealer ring was dismantled in March 2026, with 14 people prosecuted, detection volume dropped for a period before beginning to recover. How the phishing gets past the filters Mimecast looked at how these campaigns land in inboxes in the first place, beyond the phishing templates and credential harvesting pages themselves, and found a deliberate pattern of abusing trusted sending infrastructure. Threat actors have moved away from purpose-built malicious infrastructure and toward legitimate, high-reputation sending platforms that email security tools are configured to trust. Mimecast’s telemetry shows that among these campaigns, about one in three detections arrived through Salesforce infrastructure, with another quarter delivered through Google Workspace mail-merge tools and SharePoint-hosted links. “This distribution is not accidental,” researchers noted. “Salesforce and Google carry established sender reputations that bypass reputation-based filtering. Delivery through these platforms means the sending IP, domain, and authentication records (SPF, DKIM) all pass checks that would reject an unknown domain. The attacker’s only task is to make the content convincing.” Losing the account costs more than losing the budget Fraudulent ad spend can typically be shut down within hours, once a business notices unfamiliar campaigns and cancels the linked payment method. Getting the account itself back takes much longer. Attackers who gain access often add their own admins and downgrade the legitimate owner to a limited role, something platform permission structures don’t always let a newly added admin reverse. Reclaiming ownership and rebuilding standing with the platform’s review systems can take months, all while the account may still be locked in appeal queues. “Unlike card fraud, where chargeback and zero-liability protections exist, platforms offer no equivalent, and have a structural financial incentive not to act quickly,” the researchers said. Platforms still profit When a compromised ad account keeps running campaigns, the platform earns revenue on every impression served, whether the advertiser is the legitimate owner or an attacker spending stolen budget. A 2026 class-action complaint from the Consumer Federation of America cites Meta’s own estimates putting scam advertising at 15 billion impressions a day, worth an annualized $7 billion in revenue, and alleges Meta prioritized that revenue over user safety. Meta has taken some enforcement steps of its own, suing scam advertisers in February 2026 and working with law enforcement on Southeast Asian criminal networks the next month. The incentive stays in place regardless, an account under review can keep serving ads while the review plays out. Malwarebytes documented one advertiser that stayed active after 30 reports. “Until advertisers manage administrative access, partner relationships, and account activity with the same rigor applied to other high-value digital assets, and platforms provide recovery processes that reflect the value of these accounts, compromised advertising accounts will continue to circulate within the criminal ecosystem,” Mimecast concluded.

How Mimecast brings enterprise-grade email protection to API deployment In this Help Net Security video, Andrew Williams, Senior Product Manager at Mimecast, walks through the company’s API-based email security protection for Microsoft 365 and Google Workspace environments. The video covers a core problem: AI-generated phishing and business email compromise are slipping past native Microsoft 365 controls. According to Mimecast’s State of Human Risk report, 64% of organizations know their built-in email security has gaps. Mimecast’s API solution connects via Microsoft Graph API in minutes, with no MX record changes or mail flow disruption. It covers the threat spectrum from spam and commodity malware through to targeted phishing and BEC, across both E3 and E5 licensing tiers. Detection runs across 40+ inspection layers, including URL analysis, sandboxing, and behavioral modeling in 20+ languages. Learn more: Mimecast makes enterprise email security deployable in minutes Mimecast expands Incydr with runtime data security for AI and human risk

New infosec products of the month: March 2026 Here’s a look at the most interesting products from the past month, featuring releases from Beazley, Bonfy.AI, Mend.io, Mimecast, NinjaOne, Novee, Intel 471, Singulr AI, Stellar Cyber, Teleport, and Vicarius. Beazley Exposure Management platform identifies external exposures and prioritizes cyber risk Beazley Security has announced its Exposure Management product, which delivers continuous, automated discovery and intelligence-driven exposure notifications to help security teams accelerate risk mitigation in an era where AI-assisted attackers have compressed the time between vulnerability disclosure, weaponization, and exploitation. Singulr AI’s Agent Pulse delivers enforceable runtime governance and visibility for AI agents Singulr AI has announced the launch of Agent Pulse, extending its Unified AI Control Plane to autonomous AI agents and model context protocol (MCP) servers. Agent Pulse delivers enforceable runtime governance, contextual discovery, and measurable oversight for the agentic enterprise. Mend.io eliminates AI prompt weaknesses before production Mend.io has launched System Prompt Hardening within Mend AI to detect, score, and automatically remediate weaknesses in AI system prompts. Hidden instructions in system prompts have emerged as a growing security concern that traditional AppSec tools do not fully address. Vicarius vIntelligence brings continuous risk validation and AI-driven security automation Vicarius has announced the launch of vIntelligence, a new product that introduces agentic intelligence and continuous validation to the company’s security portfolio. Unlike intelligence layers that rely solely on third-party findings, vIntelligence includes its own validation engine. It can create and execute custom scripts, identify gaps in detection coverage, and revalidate exposures as environments change. NinjaOne Vulnerability Management enables real-time detection and autonomous patching NinjaOne has unveiled NinjaOne Vulnerability Management, a new solution that helps IT teams identify, prioritize, and remediate vulnerabilities faster, without relying on periodic scans from security teams that often lack context and connection to remediation workflows. Intel 471 debuts integrated platform for external threat management Intel 471 has announced its latest product offering, the Cyber Threat Exposure Bundle, on its Verity471 platform. This new bundle combines Attack Surface Exposure, Third-Party Exposure, and Brand Exposure into a single solution, providing users with a standardized view to identify, prioritize, and remediate external risks. Bonfy ACS 2.0 helps organizations control data use in AI environments Bonfy.AI announced Bonfy Adaptive Content Security (Bonfy ACS) 2.0, a platform built to secure enterprise content across all systems, applications, and AI agents – anywhere data moves, resides, or is processed. Bonfy delivers real-time, contextual protection across email, SaaS apps, collaboration tools, browsers, cloud and on‑prem file stores, AI systems, and agent frameworks, so enterprises can safely accelerate AI adoption without flying blind. Mimecast expands Incydr with runtime data security for AI and human risk Mimecast has announced a major expansion of its Incydr offering with new data security capabilities and a preview of the Agent Risk Center. These enhancements deliver runtime data security through a unified approach to detect, govern, and remediate data exposure in real time, whether driven by employees or agents acting on their behalf. Novee introduces autonomous AI red teaming to hunt LLM vulnerabilities Novee introduced AI Red Teaming for LLM Applications for its AI penetration testing platform, designed to uncover security vulnerabilities in LLM-powered applications before attackers can exploit them. Stellar Cyber advances agentic AI for SOC to reduce noise and speed investigations Stellar Cyber has introduced agentic AI to cut alert noise and speed investigations, changing how SOC teams operate. The enhancements advance its autonomous SOC platform, shifting teams from reactive alert handling to outcome-driven security operations. Teleport announced Beams, wants to remove major barrier to agentic AI Teleport announced Beams, a trusted runtime designed to solve the security and IAM challenges blocking teams from designing and running AI agents in production infrastructure. Beams runs each agent in an isolated Firecracker VM with built-in identity. Each Beam is connected to infrastructure and inference services without secrets, with audit and access control.

Mimecast makes enterprise email security deployable in minutes Most organizations running Microsoft 365 rely on native email controls as their primary line of defense. According to Mimecast research, 38% of organizations depend exclusively on those native controls for collaboration security, and 64% say those controls are insufficient against the threat landscape. Ranjan Singh, Chief Product and Technology Officer at Mimecast, outlines how the company’s API-based approach delivers protection on par with a traditional Secure Email Gateway without requiring infrastructure changes, and why that matters for stretched security teams trying to close detection gaps on BEC and credential phishing. Closing the gap between deployment speed and protection depth Organizations have long treated deployment speed and protection depth as competing priorities, accepting weaker email protection in exchange for faster rollout. Singh said that tradeoff is no longer necessary. Mimecast’s API-based solution connects directly into Microsoft 365 environments and begins protecting messages within minutes, with no MX record changes or mail flow disruptions required. The detection engines running through the API are the same ones that power Mimecast’s gateway product, covering deep URL inspection, malware sandboxing, behavioral AI, and advanced BEC detection. “Security teams get real-time visibility into threats that slipped past first-line defenses,” Singh said, “and the deployment path, whether API or SEG, is determined by infrastructure needs, not by what the technology can deliver.” Where ICES point solutions fall short Integrated Cloud Email Security vendors typically sit on top of native controls and inherit the blind spots those controls carry. Mimecast’s architecture takes a different approach, built on nearly 25 years of enterprise email security experience, detection engines trained across trillions of data points, and threat intelligence drawn from 42,000 customers globally. Singh said the differentiation lies in the intelligence behind the architecture. The solution applies multi-vector threat protection, analyzing sender authentication, domain reputation, URLs, and content simultaneously, delivering what Singh described as “consistent, high-efficacy protection without fragmented detection layers.” Catching what traditional tools miss on BEC and credential phishing BEC and credential phishing have evolved specifically to evade tools built for an earlier threat environment. Singh said Mimecast’s detection data shows the system catches three times more BEC and credential phishing attacks than traditional methods, which translates directly into fewer high-risk alerts reaching analysts and less remediation burden. When a threat is identified, automated response acts without waiting on human intervention. It pulls back malicious emails from every affected inbox, blocks compromised URLs, and quarantines suspicious attachments. “The shift from firefighting to proactive posture is where our customers see the most meaningful operational impact,” Singh said. Human risk signals beyond the inbox Email is the entry point into an organization, but Singh said human behavior is the real attack surface. Mimecast extends beyond inbox protection through its Human Risk Command Center, which connects email threat signals with user behavior, identity data, insider risk indicators, and generative AI activity. That includes surfacing when users are granting excessive permissions to AI agents, sharing sensitive data through AI-assisted tools, or exhibiting behavioral patterns that indicate compromise. By correlating those signals, the platform enables organizations to identify compromised, negligent, and malicious users before damage occurs. Singh said point solutions operating in isolation cannot provide that level of context, and the Human Risk Command Center delivers a unified risk view that aligns security, legal, and compliance teams around a shared picture of organizational exposure. Validating results before committing Because the API deployment does not disrupt existing mail flow, security teams can run Mimecast alongside current tools, observe effectiveness in real time, and measure return on investment before making any long-term commitment. Organizations begin scanning for threats that slipped past existing defenses immediately after connection, surfacing gaps in native controls that were previously invisible. Singh said that structure provides faster time to value, minimal operational friction, and a data-driven case for stronger email security. Mimecast offers a proof-of-value program for organizations that want to see what is slipping through their current defenses.

Breaches of employee data reported to the UK regulator have hit their highest level in at least seven years, according to new analysis from law firm Nockolds. The company said that reports to the Information Commissioner’s Office (ICO) had increased 5% over the past year to reach 3872 breach incidents in 2025. This is nearly 29% higher than the total number of reported breaches recorded in 2019 (3010), when these records began. However, cyber-related breaches actually fell by 6% over the past year to 1568, while non-cyber incidents jumped 15% to 2304. Nockolds principal associate, Joanna Sutton, blamed hybrid working for the evolving picture. “Organizations have strengthened their digital defenses, but many have not adapted their physical and procedural safeguards to match. The flow of devices and documents between homes and offices creates vulnerabilities that cyber tools alone cannot fix,” she claimed. “These incidents have become more common as employees split their time between multiple locations. Sensitive and highly confidential information including HR and payroll documents, disciplinary records, medical information and identity documents are now routinely handled outside controlled office environments.” HR and Security Must Work Together Nockolds said that non-cyber incidents could include: Lost or stolen laptops, phones or USB drives Paperwork left on trains, in cars or shared home environments Post or emails sent to the wrong recipient Printed documents not disposed of securely Files transported between home and office without proper controls Sutton warned that even if breaches are accidental, employees have the right to bring claims if the incident has caused them stress or anxiety. That puts the onus on employers to ensure they safeguard the large volumes of sensitive personally identifiable information (PII) held on staff. “Even if an employee accidentally causes a breach, organizations may still be liable if policies are outdated or staff have not been properly trained. HR teams therefore play a critical role in ensuring that both the human and technical elements of data protection are aligned,” she concluded. “Effective data security depends as much on employee awareness as on robust IT systems. The rise in non‑cyber incidents shows that organizations need to invest in regular, practical training and ensure that policies reflect the realities of hybrid working.” A Mimecast report earlier this month claimed that rising use of AI in the workplace is putting sensitive data at risk of misuse and abuse. It said that 42% of global organizations reported a rise in cybersecurity incidents because of employee negligence, and the same share (42%) from malicious insiders.

La recente analisi pubblicata da Huntress mette in luce una campagna di attacco sofisticata e su larga scala che ha colpito centinaia di organizzazioni a livello globale. Il dato più rilevante non è soltanto il numero delle vittime, ma il modello operativo adottato dagli attaccanti: non si tratta di una classica operazione di phishing finalizzata al furto di credenziali, bensì di un attacco strutturato il cui vero obiettivo strategico sono i token di accesso. In particolare, il monitoraggio avrebbe individuato un’attività di phishing tramite codice di accesso ai dispositivi, che ha preso di mira le identità di Microsoft 365 di oltre 340 organizzazioni negli Stati Uniti, in Canada, Australia, Nuova Zelanda e Germania. I ricercatori ritengono, inoltre, che gli aggressori abbiano sfruttato Railway, un provider PaaS (Platform-as-a-Service) pensato per il vibe coding e gli sviluppatori, attribuendo l’attacco al gruppo EvilTokens autore dell’omonima piattaforma (Phishing as a Service). Indice degli argomenti Il ruolo centrale del device code phishing Alla base della campagna si trova una tecnica nota come device code phishing, che sfrutta un flusso di autenticazione OAuth perfettamente legittimo. L’attaccante genera un codice dispositivo e induce la vittima a inserirlo su una pagina ufficiale Microsoft. L’utente completa l’accesso senza percepire anomalie, inclusa l’eventuale autenticazione a più fattori. A quel punto, il sistema rilascia un set di token validi che vengono intercettati dall’attaccante. Questo approccio consente di aggirare completamente i meccanismi tradizionali di difesa, poiché non avviene alcuna sottrazione diretta di password. “Il phishing tramite codice dispositivo si differenzia notevolmente da un attacco “adversary-in-the-middle” perché l’attaccante non si trova realmente “nel mezzo, commenta Huntrees, “Piuttosto, utilizza la legittima funzionalità di autenticazione tramite codice dispositivo Microsoft e richiede all’utente di autenticarsi in circostanze che controlla. L’autenticazione a più fattori non è una contromisura efficace contro questo tipo di attacco, poiché la vittima inserisce il codice, poi il nome utente e la password, e infine il codice MFA, se richiesto. L’autenticazione risultante genera un token che l’attaccante può recuperare, annullando di fatto la protezione dell’MFA”. Token replay e persistenza dell’accesso Il vero elemento critico infatti emerge nella fase successiva, quando i token ottenuti vengono riutilizzati in un attacco di tipo replay. I token OAuth funzionano come credenziali temporanee che non richiedono ulteriori verifiche una volta emessi e chiunque li possieda può accedere alle risorse autorizzate. Questo implica che l’attaccante può operare indisturbato all’interno dell’ambiente Microsoft 365 della vittima, senza attivare controlli di sicurezza basati su password o MFA. Inoltre, la presenza dei refresh token consente di mantenere l’accesso nel tempo, anche in caso di modifica delle credenziali dell’utente compromesso. L’abuso delle piattaforme cloud legittime Un aspetto particolarmente innovativo della campagna è, come accennato prima, l’utilizzo della piattaforma Railway per ospitare l’infrastruttura malevola. Questo servizio offre ambienti affidabili, scalabili e difficilmente classificabili come sospetti e gli attaccanti sfruttano tali caratteristiche per gestire la raccolta e il riutilizzo dei token, beneficiando di indirizzi IP con ottima reputazione. Questo rende estremamente complesso distinguere il traffico legittimo da quello malevolo, riducendo l’efficacia dei sistemi di rilevamento tradizionali basati su blacklist. “La cosa interessante non è che abbiamo colto un hacker mentre digitava i prompt in Railway in tempo reale. Il punto è che le stesse funzionalità del prodotto, progettate per aiutare gli sviluppatori legittimi a lavorare più velocemente, possono anche aiutare gli avversari a industrializzare le infrastrutture di phishing”, spiegano i ricercatori di Huntress. Una catena di attacco distribuita e resiliente La campagna si distingue anche per la complessità della catena di distribuzione. Le vittime vengono raggiunte tramite e-mail di phishing altamente personalizzate che simulano comunicazioni aziendali credibili. I link contenuti nei messaggi attivano una sequenza di reindirizzamenti attraverso servizi di fornitori di sicurezza legittimi (come Cisco, Trend Micro e Mimecast, in modo da eludere i filtri antispam) e siti compromessi fino a condurre l’utente alle landing page finali dell’attacco (ospitate solitamente su istanze Cloudflare workers.dev, ciò permetterebbe agli autori delle minacce di aggirare i filtri dei contenuti web, in quanto Cloudflare risulta una piattaforma affidabile a livello globale). In pratica, questo approccio multi-hop rende difficile bloccare l’operazione nelle fasi iniziali e dimostra un elevato livello di adattabilità. Un altro elemento distintivo osservato dai ricercatori è la forte variabilità delle esche utilizzate nella campagna. I contenuti risultano contestualizzati e coerenti con il settore della vittima, suggerendo l’impiego di strumenti avanzati di automazione e, con ogni probabilità, l’integrazione di tecniche di intelligenza artificiale generativa. La campagna analizzata rappresenta dunque un chiaro esempio dell’evoluzione delle minacce nel contesto cloud. Gli attaccanti non cercano più di violare i sistemi attraverso vulnerabilità tecniche evidenti, ma puntano piuttosto a sfruttare gli stessi meccanismi su cui si basa la sicurezza dei sistemi. In questo scenario, la capacità di monitorare e proteggere i flussi di autenticazione diventa un elemento essenziale per garantire la sicurezza delle organizzazioni. L’allarme di ACN nel contesto italiano A conferma della rilevanza della minaccia, l’Agenzia per la Cybersicurezza Nazionale ha pubblicato un bollettino che segnala una campagna attiva basata sull’abuso del flusso Device Code Grant per l’ottenimento di token OAuth. L’analisi evidenzia come gli attaccanti inducano le vittime a completare un processo di autenticazione legittimo, ottenendo così token di accesso validi senza necessità di compromettere direttamente le credenziali. Questo approccio consente di aggirare i controlli di sicurezza tradizionali, inclusi i meccanismi di autenticazione multifattore, e rafforza la necessità di adottare misure difensive orientate al monitoraggio delle sessioni, alla limitazione dei flussi di autenticazione meno controllati e alla revoca tempestiva dei token in caso di attività sospette. Anche il CERT-AgID ha diramato un’allerta per l’impatto che questa nuova minaccia ha sulle PA: gli analisti, infatti, hanno già raccolto prove che confermano come questa minaccia stia prendendo di mira anche la Pubblica Amministrazione italiana. Tutte le informazioni sulla catena di compromissione, sulle raccomandazioni per le PA e le azioni di contrasto sono disponibili nel relativo bollettino di sicurezza.

Top product launches at RSAC 2026 RSAC 2026 showcased a wave of innovation, with vendors unveiling technologies poised to redefine cybersecurity. From AI-powered defense to breakthroughs in identity protection, this year’s conference delivered a glimpse into the future. Here are the most interesting products that caught our attention, and could shape what’s next. Astrix Security has revealed a major expansion of its AI agent security platform, covering every layer where AI agents operate in the enterprise: from managed AI platforms to shadow deployments running on managed devices, detecting both agent existence and unauthorized access to enterprise resources, and enforcing policy over what agents are allowed to do. Bonfy.AI announced Bonfy Adaptive Content Security (Bonfy ACS) 2.0, a platform built to secure enterprise content across all systems, applications, and AI agents – anywhere data moves, resides, or is processed. Bonfy delivers real-time, contextual protection across email, SaaS apps, collaboration tools, browsers, cloud and on‑prem file stores, AI systems, and agent frameworks, so enterprises can safely accelerate AI adoption without flying blind. The Vellox suite showcases how AI-native cyber defense can counter growing threats to U.S. national security and critical infrastructure. The product suite is fueled by more than 30 years of technology, tradecraft, and adversarial insights. Booz Allen’s cyber operators are engaged at the center of nearly all major commercial and federal cyber missions, enabling singular insight when developing and deploying military-grade offensive and defensive products for U.S. federal, defense, and intelligence customers as well as Fortune 500 and Forbes Global 2000 companies. Black Duck has announced the general availability of Black Duck Signal, an agentic AI application security solution purpose-built to secure AI-generated code in autonomous development workflows. Signal introduces a new model for application security: agentic AI augmented by decades of human-curated security context. Delivered as an agentic AI solution, Signal enlists a coordinated system of specialized AI security agents that draw on ContextAI, Black Duck’s application security model, to analyze code, assess impact and guide remediation actions in real time. Cisco has introduced solutions to address AI security issues and remove a top barrier to agent adoption. By establishing trusted identities, enforcing strict zero trust Access controls, hardening agents before deployment, enforcing guardrails at runtime, and giving SOC teams the tools to stop threats at machine speed, Cisco is building security into the foundation of the emerging AI economy. Mimecast has announced a major expansion of its Incydr offering with new data security capabilities and a preview of the Agent Risk Center. These enhancements deliver runtime data security through a unified approach to detect, govern, and remediate data exposure in real time, whether driven by employees or agents acting on their behalf. Novee introduced AI Red Teaming for LLM Applications for its AI penetration testing platform, designed to uncover security vulnerabilities in LLM-powered applications before attackers can exploit them. Unlike conventional application security tools built for web and infrastructure testing, Novee’s AI pentesting agent is specifically designed to continuously probe AI-enabled applications. The agent autonomously simulates real-world, sophisticated attack scenarios and chaining techniques together to identify vulnerabilities that manual testing or static scanners often miss. Dashlane has unveiled Omnix AI Advisor, a natural-language AI security assistant embedded into the Dashlane Omnix platform. Built upon Omnix’s advanced credential protection and visibility capabilities, Omnix AI Advisor accelerates enterprises’ transition to a proactive security posture by turning real-time credential risk data, such as dark web exposure and phishing logs, into contextual, actionable intelligence. Palo Alto Networks has advanced its AI security platform with Prisma AIRS 3.0, securing the agentic AI lifecycle and enabling enterprises to move from observation to safe autonomous execution. Prisma AIRS replaces fragmented point solutions with a single platform to manage the primary threats and risks of AI apps and autonomous agents. The new capabilities allow teams to future-proof their operations as agent ecosystems evolve. Stellar Cyber has introduced agentic AI to cut alert noise and speed investigations, changing how SOC teams operate. The enhancements advance its autonomous SOC platform, shifting teams from reactive alert handling to outcome-driven security operations. Straiker has launched Discover AI and expanded Defend AI to secure coding agents, productivity agents, and custom-built agent platforms. Agents are operating across enterprise systems with broad access, growing autonomy, and zero security oversight. That’s why Straiker built Discover AI and Defend AI: to give security teams visibility into what agents are running and protection against what they might do. Teleport announced Beams, a trusted runtime designed to solve the security and IAM challenges blocking teams from designing and running AI agents in production infrastructure. Beams runs each agent in an isolated Firecracker VM with built-in identity. Each Beam is connected to infrastructure and inference services without secrets, with audit and access control.

Cybersecurity researchers are calling attention to an active device code phishing campaign that's targeting Microsoft 365 identities across more than 340 organizations in the U.S., Canada, Australia, New Zealand, and Germany. The activity, per Huntress, was first spotted on February 19, 2026, with subsequent cases appearing at an accelerated pace since then. Notably, the campaign leverages Cloudflare Workers redirects with captured sessions redirected to infrastructure hosted on a platform-as-a-service (PaaS) offering called Railway, effectively turning it into a credential harvesting engine. Construction, non-profits, real estate, manufacturing, financial services, healthcare, legal, and government are some of the prominent sectors targeted as part of the campaign. "What also makes this campaign unusual is not just the device code phishing techniques involved, but the variety of techniques observed," the company said. "Construction bid lures, landing page code generation, DocuSign impersonation, voicemail notifications, and abuse of Microsoft Forms pages are all hitting the same victim pool through the same Railway.com IP infrastructure." Device code phishing refers to a technique that exploits the OAuth device authorization flow to grant the attacker persistent access tokens, which can then be used to seize control of victim accounts. What's significant about this attack method is that the tokens remain valid even after the account's password is reset. At a high level, the attack works as follows - Threat actor requests a device code from the identity provider (e.g, Microsoft Entra ID) via the legitimate device code API. The service responds with a device code. Threat actor creates a persuasive email and sends it to the victim, urging them to visit a sign-in page ("microsoft[.]com/devicelogin") and enter the device code. After the victim enters the provided code, along with their credentials and two-factor authentication (2FA) code, the service creates an access token and a refresh token for the user. "Once the user has fallen victim to the phish, their authentication generates a set of tokens that now live at the OAuth token API endpoint and can be retrieved by providing the correct device code," Huntress explained. "The attacker, of course, knows the device code because it was generated by the initial cURL request to the device code login API." "And while that code is useless by itself, once the victim has been tricked into authenticating, the resulting tokens now belong to anyone who knows which device code was used in the original request." The use of device code phishing was first observed by Microsoft and Volexity in February 2025, with subsequent waves documented by Amazon Threat Intelligence and Proofpoint. Multiple Russia-aligned groups tracked as Storm-2372, APT29, UTA0304, UTA0307, and UNK_AcademicFlare, have been attributed to these attacks. The technique is insidious, not least because it leverages legitimate Microsoft infrastructure to perform the device code authentication flow, thereby giving users no reason to suspect anything could be amiss. In the campaign detected by Huntress, the authentication abuse originates from a small cluster of Railway.com IP addresses, with three of them accounting for roughly 84% of observed events - 162.220.234[.]41 162.220.234[.]66 162.220.232[.]57 162.220.232[.]99 162.220.232[.]235 The starting point of the attack is a phishing email that wraps malicious URLs within legitimate security vendor redirect services from Cisco, Trend Micro, and Mimecast so as to bypass spam filters and trigger a multi-hop redirect chain featuring a combination of compromised sites, Cloudflare Workers, and Vercel as intermediaries before taking the victim to the final destination. "The observed landing sites prompt the victim to proceed to the legitimate Microsoft device code authentication endpoint and input a provided code in order to read some files," Huntress said. "The code is rendered directly on the page when the victim arrives." "This is an interesting iteration of the tactic, as, normally, the adversary must produce and then provide the code to the victim. By rendering the code directly on the page, likely by some code generation automation, the victim is immediately provided with the code and pretext for the attack." The landing page also comes with a "Continue to Microsoft" that, when clicked, spews a pop-up window rendering the legitimate Microsoft authentication endpoint ("microsoft[.]com/devicelogin"). Almost every device code phishing site has been hosted on a Cloudflare workers[.]dev instance, illustrating how the threat actors are weaponizing the trust associated with the service in enterprise environments to sidestep web content filters. To combat the threat, users are advised to scan sign-in logs to hunt for Railway IP logins, revoke all refresh tokens for affected users, and block authentication attempts from Railway infrastructure if possible. Huntress has since attributed the Railway attack to a new phishing-as-a-service (PhaaS) platform known as EvilTokens, which made its debut last month on Telegram. Besides advertising tools to send phishing emails and bypass spam filters, the EvilTokens dashboard provides customers with open redirect links to vulnerable domains to obscure the phishing links. "In addition to rapid growth in tool functionality, the EvilTokens team has spun up a full 24/7 support team and a support feedback channel," the company said. "They also have customer feedback." The disclosure comes as Palo Alto Networks Unit 42 also warned of a similar device code phishing campaign, highlighting the attack's use of anti-bot and anti-analysis techniques to fly under the radar, while exfiltrating browser cookies to the threat actor on page load. The earliest observation of the campaign dates back to February 18, 2026. The phishing page "disables right-click functionality, text selection, and drag operations," the company said, adding it "blocks keyboard shortcuts for developer tools (F12, Ctrl+Shift+I/C/J) and source viewing (Ctrl+U)" and "detects active developer tools by utilizing a window size heuristic, which subsequently initiates an infinite debugger loop." Update In a follow-up analysis published on March 30, 2026, Sekoia described EvilTokens as a new turnkey Microsoft device code phishing kit that's sold under a PhaaS model since mid-February 2026. It also offers customers self-hosted phishing templates that include - A decoy page that impersonates a Microsoft service or trusted application (e.g., Adobe Acrobat, DocuSign, or SharePoint) as part of phishing emails bearing financial, meeting, logistics, or payroll-related lures. A verification code for the user to copy. Instructions to complete identity verification via Microsoft. A "Continue to Microsoft" button that redirects to the legitimate Microsoft device login page. Thus, when a user copies the verification code and clicks on the "Continue to Microsoft" button, the phishing page launches the legitimate Microsoft device sign-in page in a pop-up window. Once the verification code is entered, they are redirected to the standard Microsoft authentication login page. As soon as the sign-in is complete, the attacker gains access to the victim's account for the targeted service. Campaigns leveraging the PhaaS service have targeted organizations across North, Central, and South America, Europe, the Middle-East, Asia, and Oceania. Among the most affected are the U.S., Australia, Canada, France, India, Switzerland, and the U.A.E. The extensive reach and diversity of these campaigns reflect the rapid adoption of EvilTokens, Sekoia added. "EvilTokens provides a turnkey Microsoft device code phishing kit and a range of advanced features to conduct BEC attacks, including access weaponisation, email harvesting, reconnaissance capabilities, a built-in webmail interface, and AI-powered automation," the French cybersecurity company said. "The EvilTokens PhaaS operates through fully featured bots on Telegram and continuously improves its phishing kit with new capabilities. In the near future, the operator intends to extend support to Gmail and Okta phishing pages." (The story was updated after publication on March 31, 2026, to include additional details of EvilTokens.)

Mimecast expands Incydr with runtime data security for AI and human risk Mimecast has announced a major expansion of its Incydr offering with new data security capabilities and a preview of the Agent Risk Center. These enhancements deliver runtime data security through a unified approach to detect, govern, and remediate data exposure in real time, whether driven by employees or agents acting on their behalf. Eighty percent of Fortune 500 companies now run active AI agents, yet only 14% have full security approval for them1. Enterprise data loss is no longer just a people problem, AI agents have introduced an entirely new attack surface. Agents are accessing and sharing sensitive data through pathways which traditional security tools were never designed to monitor, MCP-connected workflows, commercial agents, user-built automations, and shadow AI tools. “Intent-based detection treats all agents equally. We don’t, because the human behind the agent is the signal that changes everything,” said Rob Juncker, Chief Product Officer, Mimecast. “Who deployed the agent? What do we already know about them? How is data moving across email, collaboration tools, browsers, SaaS apps, endpoints, and AI-driven workflows — and what intervention is required right now? That’s a runtime data security problem, not a model problem.” Adaptive data security Mimecast’s Incydr technology has long helped organizations prevent insider-driven data loss through out-of-the-box visibility, intelligent detection via its PRISM risk engine (250+ risk indicators), and adaptive response ranging from in-context education to real-time blocking. The new capabilities extend Incydr technology from insider-led data security into broader runtime data security for both human and AI-driven risk. This expansion takes a new approach, combining Incydr endpoint and browser intelligence with Mimecast’s email and collaboration security, delivering complete ingress-to-egress data visibility, covering the full path of enterprise data movement across endpoints, browsers, SaaS applications, AI tools, MCP connections, and email. New and expanded capabilities are engineered to include: Unified human and agent visibility – A single view into data loss risk across employees and autonomous agents, spanning endpoints, cloud and SaaS applications, email, browser activity, commercial AI tools, MCP server connections, and user-developed agents. Shadow AI and unsanctioned agent detection – Purpose-built detection for unsanctioned AI usage, out-of-policy commercial agents, unauthorized MCP connections to production databases and critical SaaS platforms, and user-built agents operating on unapproved LLM providers or accessing production environments without security review. Adaptive risk scoring for people and AI agents – The Incydr risk engine now continuously scores both human users and AI agents based on behavioral anomalies, policy violations, high-risk data access, unsanctioned application usage, agent compliance posture, and exposure to critical systems and data sources (e.g., Snowflake, Stripe, PostgreSQL, AWS, Salesforce, GitHub). Granular data-to-agent access mapping – A view of which agents and tools access which categories of sensitive data, including customer PII, source code, financial records, internal communications, HR data, and infrastructure configurations, enabling security teams to understand and control the agent-to-data blast radius. Policy-driven governance – A comprehensive governance framework for classifying and enforcing policy across all AI tools, commercial agents, MCP servers, and user-developed agents, with sanctioned, unsanctioned, and uncategorized classifications, department-level enforcement, and AI acceptable use policy management. Introducing the Mimecast Agent Risk Center A single data loss investigation might involve an employee sharing a file through an unsanctioned tool, such as DeepSeek, OpenClaw, Ollama, ChatGPT, a commercial AI agent, summarizing confidential records, and a user-built agent pulling from a production database it was never meant to access. These events show up in different systems, follow different detection logic, and require different response playbooks, if they show up at all. The Mimecast Agent Risk Center is designed to consolidate that fragmented picture into one experience. Critically, the Agent Risk Center is built to connect every finding directly to action. Built-in agentic workflows automate the response chain, notifying users, escalating to managers, enforcing controls, and generating compliance reports, so teams act at machine speed, not human speed. As engineered, the capabilities will include: Anomaly detection engine for risky agent behavior – Can automatically surface high-risk patterns, unsanctioned tools with production database access, finance users connected to payment MCP servers, user-developed agents using non-sanctioned LLM providers, and executives with overly broad MCP configurations. Governance scorecards – A continuous assessment of organizational posture across four dimensions: policy coverage, review currency, human-in-the-loop enforcement, and LLM compliance, giving CISOs a measure of their agentic governance maturity. Department-level risk heatmaps – Visual analytics showing risk distribution, department-level exposure, risk factor breakdowns, and trend patterns, enabling targeted intervention rather than blanket policy. Integrated remediation workflows – Every risk finding connects directly to action, block access, notify users, escalate to managers, create tickets, classify uncategorized tools, schedule agent reviews, and generate compliance reports, all without leaving the unified interface. Mimecast is previewing the Agent Risk Center at RSAC 2026, with Early Access expected in September 2026. Join Chief Product Officer, Rob Juncker, for a presentation and live demo — Mimecast booth N-5245 — at 10:30 AM on Tuesday, March 24th and Thursday, March 26th.

Top must-visit companies at RSAC 2026 RSAC 2026 Conference is taking place at the Moscone Center in San Francisco March 23 – 26. With hundreds of booths, countless product demos, and nonstop buzz, navigating RSAC can be overwhelming. That’s why we’ve done the legwork to highlight the standout companies you won’t want to miss. Whether you’re looking for cutting-edge innovation, industry veterans with new offerings, or rising stars shaking things up, these exhibitors are bringing something special to the floor this year. Be sure to carve out time in your schedule to stop by, as you might just discover your next big opportunity. Booth S-3316 | Book a demo Apiiro is an application security company with offices in New York and Tel Aviv. Its agentic Application Security Posture Management (ASPM) platform helps security and development teams detect, prioritize, and fix risks across the software development lifecycle, from design through code to deployment. Powered by patented Deep Code Analysis technology, the platform provides code-to-runtime context, automated threat modeling, and AI-driven remediation. Apiiro has raised over $135 million in funding from investors including General Catalyst, Kleiner Perkins, and Greylock. Gartner, IDC, and Frost and Sullivan have all recognized Apiiro as a leader in ASPM, and its customers include USAA, BlackRock, Shell, and TIAA. Booth ESE-19 Cline is an open-source AI coding agent that runs inside Visual Studio Code, JetBrains IDEs, and the command line. It goes beyond code completion by reading codebases, creating and editing files, executing terminal commands, automating browser interactions, and connecting to external tools via the Model Context Protocol, all with user approval at each step. Developers can bring their own API keys and connect to any major AI provider, including Anthropic, OpenAI, Google Gemini, AWS Bedrock, and local models. Cline has surpassed 5 million installs and nearly 60,000 GitHub stars, and is trusted by developers at companies including Samsung, Microsoft, Salesforce, Amazon, and Visa. Booth N-5181 | Book a strategy session GlobalSign by GMO is a leading Certificate Authority and digital identity provider founded in Belgium in 1996 and now a subsidiary of Japan’s GMO Internet Group. The company issues SSL/TLS certificates, S/MIME email security certificates, code signing certificates, and document signing solutions to businesses, enterprises, cloud providers, and IoT manufacturers worldwide. Its Atlas platform enables automated certificate lifecycle management at scale. GlobalSign by GMO is a founding member of the CA/Browser Forum and became a Qualified Trust Service Provider under the eIDAS regulation in both the EU and the UK. With over 600 employees across more than a dozen countries, the company serves clients including Microsoft, Cisco, and Johnson and Johnson. Booth S-2452 IDEMIA is a global leader in biometrics and cryptography, providing identity and security solutions to governments and enterprises in more than 180 countries. The company operates through two main divisions: IDEMIA Secure Transactions, which delivers payment cards, eSIM connectivity, and cryptographic security including hardware security modules and post-quantum cryptographic libraries; and IDEMIA Public Security, which provides biometric solutions for border control, law enforcement, access control, and travel. IDEMIA is trusted by more than 600 governmental organizations and 2,400 enterprises. Booth N-5245 | Book a demo Mimecast is a leading cybersecurity company focused on managing and mitigating human risk for organizations worldwide. Its AI-powered, API-enabled platform is built to protect businesses from a broad spectrum of cyber threats by integrating advanced technology with human-centric security pathways. The platform enhances visibility, delivers strategic insight, and enables decisive action to safeguard critical data and collaborative environments. It also actively engages employees in reducing risk and improving productivity. Booth ESE-28 MyCISO is a SaaS cybersecurity platform designed to help organizations assess, improve, and manage their security posture without relying on spreadsheets or fragmented tools. Its Security Operating System centralizes assessments, risk management, compliance, supplier security, incident response, security awareness, and board-ready reporting into a single platform supporting over 65 frameworks. AI-powered insights and automated external and internal vulnerability scans help security leaders prioritize risks, track maturity over time, and demonstrate progress to executives. Booth S-0262 | Book a demo Novee is an AI-powered penetration testing platform that continuously simulates real-world cyberattacks to help organizations find and fix vulnerabilities before hackers do. Unlike traditional annual pentests or generic scanners, Novee deploys a hive-mind of AI agents trained on offensive security tradecraft to map environments, uncover exploit chains, and identify business logic flaws. It can begin with zero knowledge, mirroring how real attackers operate, then expand into deeper coverage. For every issue discovered, Novee validates the finding and delivers personalized, step-by-step remediation guidance. Booth S-3111 | Book a demo Teleport is an infrastructure identity company that provides a unified platform for securing access across classic and AI infrastructure. Its platform consolidates identity for humans, machines, workloads, and AI agents using cryptographic identity and short-lived certificates, eliminating static credentials and standing privileges. Key capabilities include zero trust access, identity governance, privileged access management, machine and workload identity, and security for agentic AI and Model Context Protocol tooling. Teleport has raised over $169 million in funding, and is valued at $1.1 billion. Customers include Nasdaq, DoorDash, Accenture, Discord, and GitLab. Booth ESE-09 Unbound AI is a cybersecurity company that created the Agent Access Security Broker (AASB) category, a governance layer purpose-built for AI coding agents. Its platform helps enterprises discover every AI coding agent in use across their organization, including tools such as Cursor, Claude Code, GitHub Copilot, and Cline, assess their risk, and enforce granular policies over terminal commands, MCP server connections, and sensitive data flows. The platform processes over one million agent tool calls per month and deploys via MDM with no code changes required. Customers include THG Ingenuity, WeWork, Siemens, and Exterro. Booth ESE-11 12Port is a cybersecurity company providing an agentless Privileged Access Management (PAM) platform for enterprises and managed service providers. Its platform secures, monitors, and audits privileged sessions across physical, virtual, and cloud environments without requiring software agents on target endpoints. Core capabilities include a credential vault with FIPS 140-3 validated encryption, automated credential rotation, just-in-time access controls, session recording, MFA enforcement, and AI-powered session intelligence that detects policy violations and anomalies in real time. The platform supports hybrid, multi-cloud, and air-gapped environments, integrating with Active Directory, Entra ID, SSO, SIEM tools, and cloud platforms.

Microsoft has warned of fresh campaigns that are capitalizing on the upcoming tax season in the U.S. to harvest credentials and deliver malware. The email campaigns take advantage of the urgency and time-sensitive nature of emails to send phishing messages masquerading as refund notices, payroll forms, filing reminders, and requests from tax professionals to deceive recipients into opening malicious attachments, scanning QR code, or interacting with suspicious links. "Many campaigns target individuals for personal and financial data theft, but others specifically target accountants and other professionals who handle sensitive documents, have access to financial data, and are accustomed to receiving tax-related emails during this period," the Microsoft Threat Intelligence and Microsoft Defender Security Research teams said in a report published last week. While some of these efforts direct users to sketchy pages designed through Phishing-as-a-service (PhaaS) platforms, others result in the deployment of legitimate remote monitoring and management tools (RMMs), such as ConnectWise ScreenConnect, Datto, and SimpleHelp, enabling the attackers to gain persistent access to compromised devices. The details of some of the campaigns are below - Using Certified Public Accountant (CPA) lures to deliver phishing pages associated with the Energy365 PhaaS kit to capture victims' email and password. The Energy365 phishing kit is estimated to be sending hundreds of thousands of malicious emails on a daily basis. Using QR code and W2 lures to target approximately 100 organizations, mainly in the manufacturing, retail, and healthcare industries located in the U.S., to direct users to phishing pages mimicking the Microsoft 365 sign-in pages and built using the SneakyLog (aka Kratos) PhaaS platform to siphon their credentials and two-factor authentication (2FA) codes. Using tax-themed domains for use in phishing campaigns that trick users into clicking on bogus links under the pretext of accessing updated tax forms, only to distribute ScreenConnect. Impersonating the Internal Revenue Service (IRS) with a cryptocurrency lure that specifically targeted the higher education sector in the U.S., instructing recipients to download a "Cryptocurrency Tax Form 1099" by accessing a malicious domain ("irs-doc[.]com" or "gov-irs216[.]net") to deliver ScreenConnect or SimpleHelp. Targeting accountants and related organizations, asking for help to file their taxes by sending a malicious link that leads to the installation of Datto. Microsoft said it also observed a large-scale phishing campaign on February 10, 2026, in which more than 29,000 users across 10,000 organizations were affected. About 95% of the targets were located in the U.S., spanning industries like financial services (19%), technology and software (18%), and retail and consumer goods (15%). "The emails impersonated the IRS, claiming that potentially irregular tax returns had been filed under the recipient's Electronic Filing Identification Number (EFIN). Recipients were instructed to review these returns by downloading a purportedly legitimate 'IRS Transcript Viewer,'" the tech giant said. The emails, which were sent through Amazon Simple Email Service (SES), contained a "Download IRS Transcript View 5.1" button that, when clicked, redirected users to smartvault[.]im, a domain masquerading as SmartVault, a well-known document management and sharing platform. The phishing site relied on Cloudflare to keep bots and automated scanners at bay, thus ensuring that only human users are served the main payload: a maliciously packaged ScreenConnect that grants the attackers remote access to their systems and facilitates data theft, credential harvesting, and further post‑exploitation activity. To stay safe against these attacks, organizations are recommended to enforce 2FA on all users, implement conditional access policies, monitor and scan incoming emails and visited websites, and prevent users from accessing the malicious domains. The development coincides with the discovery of several campaigns that have been found to drop remote access malware or conduct data theft - Using fake Google Meet and Zoom pages to lure users into fraudulent video calls that ultimately deliver remote-access software like Teramind, a legitimate employee monitoring platform, by means of a bogus software update. Using a fraudulent website that leverages the Avast branding to trick French-speaking users into handing over their full credit card details as part of a refund scam. Using a typosquatted website impersonating the official Telegram download portal ("telegrgam[.]com") to distribute trojanized installers that, in addition to dropping a legitimate Telegram installer, execute a DLL responsible for launching an in-memory payload. The malware then initiates communication with its command-and-control infrastructure to receive instructions, download updated components, and maintain persistent access. Abusing Microsoft Azure Monitor alert notifications to deliver callback phishing emails that use invoice and unauthorized-payment lures. "Attackers create malicious Azure Monitor alert rules, embedding scam content in the alert description, including fake billing details and attacker-controlled support phone numbers," LevelBlue said. "Victims are then added to the Action Group linked to the alert rule, causing Azure to send the phishing message from the legitimate sender address [email protected]." Using quotation-themed lures in phishing emails to deliver a JavaScript dropper that connects to an external server to download a PowerShell script, which launches the trusted Microsoft application "Aspnet_compiler.exe" and injects into it an XWorm 7.1 payload via reflective DLL injection. The updated malware comes with a .NET-developed component engineered for stealth and persistence. Similar requests for quotation lures have also been used to trigger a fileless Remcos RAT infection chain. Using phishing emails and ClickFix ploys to deliver NetSupport RAT and gain unauthorized system access, exfiltrate data, and deploy additional malware. Using Microsoft Application Registration Redirect URI's ("login.microsoftonline[.]com") in phishing emails to abuse trust relationships and bypass email spam filters to redirect users to phishing websites that capture victims' credentials and 2FA codes. Abusing legitimate URL rewriting services from Avanan, Barracuda, Bitdefender, Cisco, INKY, Mimecast, Proofpoint, Sophos, and Trend Micro to conceal malicious URLs in phishing emails evades detection. "Threat actors have increasingly adopted multi-vendor chained redirection in their phishing campaigns," LevelBlue said. "Earlier activity typically relied on a single rewriting service, but newer campaigns stack multiple layers of already‑rewritten links. This nesting makes it significantly harder for security platforms to reconstruct the full redirect path and identify the final malicious destination." Using malicious ZIP files impersonating a wide range of software, including artificial intelligence (AI) image generators, voice-changing tools, stock-market trading utilities, game mods, VPNs, and emulators, to deliver Salat Stealer or MeshAgent, along with a cryptocurrency miner. The campaign has specifically targeted users in the U.S., the U.K., India, Brazil, France, Canada, and Australia. Using digital invitation lures sent via phishing emails to divert users to a fake Cloudflare CAPTCHA page that delivers a VBScript, which then runs PowerShell code to fetch an evasive .NET loader dubbed SILENTCONNECT from Google Drive to eventually deliver ScreenConnect. The findings follow an uptick in RMM adoption by threat actors, with the abuse of such tools surging 277% year-over-year, according to a recent report published by Huntress. One notable tactic involves the daisy-chaining of distinct RMM tools to fragment telemetry, distribute persistence, and complicate attribution and containment efforts, the company added. "As these tools are used by legitimate IT departments, they are typically overlooked and considered 'trusted' in most corporate environments," Elastic Security Labs researchers Daniel Stepanic and Salim Bitam said. "Organizations must stay vigilant, auditing their environments for unauthorized RMM usage."

Microsoft Azure Monitor alerts are being abused to send callback phishing emails that impersonate warnings from the Microsoft Security Team about unauthorized charges on your account. Azure Monitor is Microsoft's cloud-based monitoring service that collects and analyzes data from Azure resources, applications, and infrastructure. It enables users to track performance, notify about billing changes, detect issues, and trigger alerts based on various conditions. Over the past month, numerous people have reported receiving Azure Monitor alerts warning of suspicious charges or invoice activity on their accounts, urging them to call an enclosed phone number. "Alert rule description MICROSOFT CORPORATION BILLING AND ACCOUNT SECURITY NOTICE (REF: MS-FRA-6673829-KP). Our system has detected a potentially unauthorized charge on your account. Transaction Details: Merchant: Windows Defender. Transaction ID: PP456-887A-22B. Amount: 389.90 USD. Date: 03/05/2026l," reads the fake billing alert. "For your protection, this transaction has been temporarily placed on hold by our Fraud Detection Team. To prevent possible account suspension or additional fees, please verify this transaction immediately. If you did NOT authorize this payment, contact our 24/7 Microsoft Account Security Support at +1 (864) 347-2494 or +1 (864) 347-4846." "We apologize for any inconvenience and appreciate your prompt response. Microsoft Account Security Team." Unlike other phishing campaigns, these messages are not spoofed, but are sent directly by the Microsoft Azure Monitor platform using the legitimate [email protected] email address. As the emails are sent through Microsoft's legitimate email platforms, they pass SPF, DKIM, and DMARC email security checks, making them appear more trustworthy. Authentication-Results: relay.mimecast.com; dkim=pass header.d=microsoft.com header.s=s1024-meo header.b=CKfQ8iOB; arc=pass ("microsoft.com:s=arcselector10001:i=1"); dmarc=pass (policy=reject) header.from=microsoft.com; spf=pass (relay.mimecast.com: domain of [email protected] designates 40.107.200.103 as permitted sender) [email protected] The threat actors are conducting this campaign by creating alerts in Azure Monitor for easily triggered conditions, such as new orders, payments, generated invoices, and other billing events. When creating alerts, you can enter any message you want in the description field, which the attackers use to put their callback phishing message. These alerts are then configured to send emails to what is believed to be a mailing list under the attacker's control, which forwards the email to all the targeted people in the attack. This also preserves the original Microsoft headers and authentication results, helping the emails bypass spam filters and user suspicion. BleepingComputer has seen multiple alert categories used in this campaign, mostly using invoice and payment-themed rules designed to resemble automated billing notifications: Azure monitor alert rule order-22455340 was resolved for invoice22455340 Azure monitor alert rule Invoice Paid INV-d39f76ef94 was resolved for invd39f76ef94 Azure monitor alert rule Payment Reference INV-22073494 was resolved for purchase22073494 Azure monitor alert rule Funds Successfully Received-ec5c7acb41 was triggered for subec5c7acb41 Azure monitor alert rule MemorySpike-9242403-A4 was triggered Azure monitor alert rule DiskFull-3426456-A6 was triggered for locker3426456 The campaign relies on creating a sense of urgency, which in this case is the unusual $389 Windows Defender charge, to trick the users into calling the listed phone number. While BleepingComputer did not call the number in this scam, previous callback phishing campaigns led to credential theft, payment fraud, or the installation of remote access software. As these emails use a more enterprise or corporate theme, they may be intended to gain initial access to corporate networks for follow-on attacks. Users should treat any Azure or Microsoft alert that includes a phone number or urgent request to resolve billing issues with suspicion. Overall prevention scores can hide what happens after initial access. Once attackers are using valid credentials, prevention drops sharply. The Blue Report 2026 measures defenses technique by technique across 338 million simulations run in customer production environments. Get the report

Mimecast adds AI investigation and adaptive controls to manage human risk Mimecast has announced major platform capabilities designed for a new enterprise reality as AI agents and automated workflows scale across the business and establish the human layer as the new security control plane. According to Mimecast’s The State of Human Risk 2026, 98% of organizations now use AI to defend against threats, yet 80% are concerned about sensitive data exposure through generative AI tools and 60% still lack strategies to address AI-driven threats. At the same time, the number of actively deployed AI agents worldwide is set to grow significantly over the coming years. The proliferation of AI across businesses is exposing new pathways for data leakage, impersonation, insider risk, and policy drift. Mimecast’s latest platform enhancements help security teams meet that challenge with adaptive controls, AI-powered investigation, and open ecosystem integrations that secure how employees and AI tools interact in real time. “Every AI agent, every automated workflow ultimately depends on a human to direct it, approve its actions, and govern its behavior,” said Ranjan Singh, Chief Product and Technology Officer at Mimecast. “That means the enterprise AI era will not be secured by isolated tools or static policies. It requires a connected platform that can understand behavior, detect risk, protect sensitive data, and adapt controls in real time. That is exactly where Mimecast is focused.” Mimecast helps enterprises move beyond disconnected point products and protect the spectrum of human risk across email, collaboration, identity, insider activity, and generative AI usage. The following enhancements deliver on that vision with measurable improvements in detection accuracy, data protection, and policy enforcement. Investigate at machine speed and respond using AI and adaptive controls Mimecast data shows that just 8% of employees account for 80% of security incidents, yet only 28% of organizations coordinate security training with continuous monitoring. These capabilities ensure simplicity in the workflows that close that gap. Automated, adaptive security policies: The Mimecast platform now allows more flexibility to automatically adjust security controls based on individual user risk levels, leveraging insight from the Human Risk Command Center. High-risk users receive tighter controls instantly, low-risk users maintain productivity. Security policies evolve with behavior rather than waiting for manual intervention. AI-powered investigation and response: The new Mimecast Mihra Investigation Agent saves security teams valuable time by synthesizing events, summarizing findings, and recommending actions to resolve incidents and improve response times. It’s the latest in a growing suite of Mimecast Mihra Agents designed to help customers proactively identify, assess, and prevent threats. Streamlined incident response workflows: Redesigned response capabilities deliver 78% faster resolution of reported messages through AI-powered campaign identification, simplifying incident response for organizations that outsource security operations. An open platform that fits how security teams work Security teams do not operate in a single vendor world. They work across multiple tools, data sources and workflows, yet too many security architectures still force teams into fragmented operations and manual response. Mimecast’s latest enhancements are built to fit the way security teams actually work: open, integrated, and centered on measurable user risk. Bring your own AI via the Mihra MCP Gateway: Organizations can now connect Mimecast investigation workflows directly into their existing AI platforms like Claude or Gemini, accelerating threat response through familiar environments without forcing teams onto another siloed interface. Expanded ecosystem intelligence across the human risk surface: The Human Risk Command Center now correlates signals across email, endpoint, identity, data, generative AI, and other third-party security tools, giving CISOs a unified view of organizational risk and enabling more coordinated action. Modernized API-based deployment: Mimecast’s full detection stack is now available through API-based deployment, with complete threat protection and no MX record changes required. Protection across the platform Mimecast is delivering measurable gains in detection and protection across its platform, with new capabilities designed to address the broad spectrum of AI-driven threats, from sensitive data leaking into unsanctioned AI tools to increasingly sophisticated, AI-generated attacks. Controls for shadow AI and sensitive data exposure: With Incydr Data Protection, organizations can better prevent sensitive data from reaching unsanctioned AI tools and help ensure that approved enterprise data is used only in compliant, authorized environments. Multi-Vector Threat Protection: Now enhanced for all Mimecast customers, Multi-Vector Threat Protection delivers 99.998% detection accuracy by correlating signals across multiple detection mechanisms to identify sophisticated, coordinated campaigns that isolated security engines miss, while simplifying policy management. Enhanced business email compromise protection: BEC protection now detects impersonation and social engineering attacks across 20 languages, with detection rates improving 5%+ quarterly as the system continuously learns from emerging attack patterns observed across Mimecast’s global threat intelligence network of more than 1.7 billion emails inspected per day.

Mimecast brings gateway-grade email security to API deployment Mimecast has announced that its complete email security protection stack is now available through API deployment, eliminating a fundamental trade-off in the market. Standalone integrated cloud email security (ICES) solutions offered fast deployment but came at a cost: they were built primarily for targeted, sophisticated attacks and relied on native Microsoft or Google controls to handle the volume of everyday threats. The just-released Mimecast State of Human Risk 2026 report quantifies the cost of that dependency: 64% of organizations acknowledge that native collaboration tool security controls are insufficient, yet many continue relying on them. The consequences are measurable: 53% report increased phishing volume and 48% see rising business email compromise attacks that native protections and current ICES solutions fail to stop at scale. Mimecast’s API deployment is engineered to deliver the full detection stack, from deep URL and malware inspection to advanced AI-powered engines, through direct Microsoft 365 integration that deploys in minutes without requiring mail exchange (MX) record changes or mail flow modifications. Threat detection capabilities delivered through API architecture Mimecast’s AI-driven detection engines have been trained across 24 trillion data points and hardened across 42,000 organizations globally. Customers using Mimecast’s new detection models are catching 3x more business email compromise and credential phishing attacks than traditional detection methods identify, these threats are specifically designed to bypass conventional security. That detection capability, previously available only through gateway deployment, is now delivered through an API architecture. The API deployment also includes behavioral AI that is engineered to identify patterns across email and identity, along with broad threat protection capabilities that pure-play ICES vendors can’t offer at scale: Multi-vector threat protection (MVTP) is built to correlate signals across sender authentication, domain reputation, URLs and content simultaneously, delivering the layered analysis that single-vector ICES engines cannot replicate Deep URL inspection is engineered to deliver time-of-click analysis that catches threats evading pre-delivery scanning Advanced BEC protection is designed to apply modern AI infrastructure across more than 20 languages, detecting impersonation and social engineering at a precision that requires real-world training volume to achieve Malware detection and active sandboxing is designed to analyze attachments in isolated environments, catching zero-day threats that signature-based approaches miss Account takeover protection is engineered to identify post-compromise behavior through identity signal correlation, containing breaches before data leaves the environment These capabilities are available today through an API architecture that is built to deploy rapidly. Choose your architecture, keep your full detection stack Organizations can deploy via API or maintain MX-based architecture, with identical detection capabilities across both options. The API deployment is designed to integrate directly with Microsoft 365 in minutes, enabling organizations to test and validate protection without infrastructure changes. Regardless of deployment model, Mimecast connects directly with more than 350 security vendors across the customer security stack, spanning endpoint, XDR, SIEM, SOAR, data protection, threat intelligence and identity. Threat signals captured at the email layer flow automatically into the tools security teams already rely on, eliminating alert silos and accelerating response. Organizations gain immediate value from their existing investments in platforms like CrowdStrike, Okta, Palo Alto Networks and many others. “Standalone ICES vendors secure email. Mimecast secures the human behind it,” said Ranjan Singh, Chief Technology and Product Officer at Mimecast. “We’ve invested in AI and detection engineering that the market said couldn’t be delivered through API architecture. The result is a fundamentally different approach to email security, one that gives organizations full protection without compromising deployment speed.” A connected suite of AI-driven human risk solutions Unlike standalone ICES solutions, Mimecast’s API deployment connects to the broader Mimecast Human Risk platform. Email threat signals feed directly into the Human Risk Command Center, where they are correlated with user behavior, identity data, insider risk indicators and generative AI activity, giving CISOs a unified view of organizational risk. From that single view, organizations can detect threats, change behavior, protect data and prove compliance across three integrated solution areas: Security behavior management translates email threat detections into targeted interventions. When a user clicks a malicious link or falls for a phishing simulation, in-the-moment behavioral nudges and adaptive training turn the threat event into a teachable moment that reduces repeat risk. Insider risk management & data protection extends protection from external threats to insider risk, correlating email activity with file movement, data exfiltration patterns and identity signals to identify when trusted users, whether negligent, compromised or malicious, pose a risk to sensitive data before damage occurs. Governance compliance & insights provides complete visibility into collaboration data across email and connected platforms. Legal, compliance and security teams gain search, discovery and audit capabilities, ensuring that the same environment being protected is also fully governed and audit ready. All Mimecast email security customers benefit from increased visibility, correlation and efficacy of the Mimecast advanced protection engines, which are consistent across both API and MX-based deployments.

The risk of insider threats is on the rise and businesses are concerned about the cybersecurity implications of intentionally malicious or negligent employees, research by Mimecast has warned. According to the company’s State of Human Risk Report 2026, internal cybersecurity risk has grown across the board, to the extent that it should be treated as a “critical business threat.” In many cases, the additional insider risk is because of employees mishandling or actively abusing AI tools. According to the report, cybersecurity leaders have concerns about the rise of AI in the workplace and the potential for large language models (LLMs) and other AI productivity tools to expand the potential attack surface which could be exploited by both external and internal threats. Over the past year, 42% of organizations have reported an increase in threats from malicious insiders, employees who want to actively cause harm to their employer by stealing, manipulating or destroying data. The same percentage (42%) reported a rise in cybersecurity incidents because of employee negligence. These are incidents which occur because of careless actions by the employee which could have easily been avoided, such as transferring data insecurely using personal cloud accounts, using weak passwords or opening malicious links in phishing emails. The report warns that that cyber attackers look to exploit this negligence – or indeed, actively malicious intent - to help gain access to accounts, files and systems and that the problem is growing. According to the paper, concerns about malicious insiders from information security leaders has grown by 10% in the last year and IT and cybersecurity leaders expect to face an average of six insider-driven threats a month. “Insider risk has become one of the most consequential and underestimated threats facing organizations today, not just because of the data loss it causes, but because attackers are increasingly exploiting insiders as a deliberate entry point to bypass perimeter defenses entirely,” said Mimecast CISO Leslie Nielsen. Attackers also deploy AI tools themselves, using them to help create more realistic, more effective phishing emails. Meanwhile, it’s possible for malicious insiders to deploy AI tools to help them achieve their goals, for example, by searching for and exfiltrating files and data. “As AI makes it easier for insiders to exfiltrate data at scale, security must meet users at the point of risk,” said Nielsen. The paper is based on research by Mimecast and Vanson Bourne which surveyed 2500 IT security anddecision makers across the world, including North America, Europe, Southeast Asia and Australia. Organization sizes ranged from 250 to over 10,000 employees.

Today, we are excited to announce Varonis Interceptor, AI-native email security. AI has enabled a new breed of email threats that are more deceptive than ever. By mimicking tone, language, branding, and even voice and video, attackers are unleashing automated phishing campaigns that are virtually impossible for humans — or traditional email security tools — to detect. Organizations need AI to combat malicious AI. Varonis Interceptor uses several integrated layers of AI analysis to address known tactics and predict never-before-seen threats. This proven approach produces the highest rate of detection on the market, catching advanced business email compromise, social engineering, and phishing attacks that other products simply can’t see. The landscape has changed. Has your defense? The first generation of AI-powered email security tools relies too heavily on NLP as a catch-all, and other solutions fail to incorporate behavioral analysis or fully analyze the visual elements of an email for anomalies using AI computer vision. Sophisticated phishing and social engineering threats require a layered approach. Interceptor replaces ineffective API-based solutions and augments legacy SEGs with a true multimodal AI strategy. It combines numerous sources of AI-powered analysis to achieve a higher rate of detection and deliver superior protection. Multimodal AI for email defense in depth Robust email security solutions must adopt a multimodal approach to effectively detect modern threats. Attackers increasingly use varied techniques — like text manipulation, image-based phishing, and embedded malicious links — to evade traditional filters. By combining signals from visual, linguistic, and behavioral analysis, Interceptor can accurately identify and block AI-generated threats. As a result, security teams experience fewer false negatives and positives, eliminating busywork and noise. Our team analyzed data from 1,000 real-world IT environments and found that no organization was breach-proof. In fact, 99% of organizations have exposed sensitive data that can easily be surfaced by AI. Read the report Vision Model Interceptor’s vision model simulates how users visually experience each email or website to detect subtle signs of deception in the layout, images, and other visual elements. This can include: Authenticity of the logo in an email signature or website Attackers' attempting to hide text by dropping it in as an image QR codes embedded in a PDF or obfuscated as text to avoid image processing Language Model Interceptor’s model is grounded on the tone and linguistic patterns of each sender, including external parties such as vendors and internal communications. Interceptor also analyzes the topic, tone, and intent of messages to detect manipulation triggers common in social engineering attacks. The model is specifically tuned for scenarios where other NLPs struggle: Short messages with very little body or context Mail containing several incoherent topics aimed at deceiving AI Link-less BEC attempts Behavior Model Varonis’ behavior model uses the sender and recipient’s relationship graph and known communication patterns and conversation styles to detect deviations from their baseline. Interceptor can understand a full spectrum of social dynamics: Relationship graph and social patterns for internal and external parties Organizational structure and context Conversational styles and common topics Applying this layered approach to identifying threats in real time has been proven to outperform other leading providers, such as Abnormal Security and Mimecast. Security teams prioritize detection quality above all else. Varonis Interceptor gives leaders confidence in their defenses and users trust in the inbox. Detecting zero-hour threats — Interceptor Phishing Sandbox In addition to AI analysis conducted on messages, Varonis Interceptor proactively scans nearly every newly registered domain and published URL in its phishing sandbox, providing swift and in-depth analysis. Interceptor’s Phishing Sandbox follows the user’s path to the very end, going past every form fill, CAPTCHA, Cloudflare authentication, and more. As a result, Varonis uncovers any potential attempt to trick users and steal secrets or credentials. Interceptor is often the first AI to see and analyze the site before any other agent or human in the world. As a result, Varonis blocks malicious domains and URLs 12-24 hours before other solutions in the market and automatically removes mail containing them. Multichannel protection beyond the inbox — Interceptor Browser Security Hackers are also flooding users with social engineering attacks across email, messaging, collaboration, and social apps. This requires a solution that spans multiple channels beyond email alone. Interceptor extends to the browser and blocks malicious websites, no matter where the link-based attack originates. Most malicious links will slip past controls relying on known threat feeds. Interceptor receives advanced notice from its phishing sandbox and live threat database to stop threats as fast as they start. Interceptor Browser Security shields employees from live phishing sites with a lightweight browser extension available for all major browsers. End-to-end approach to stop breaches By offering Varonis Interceptor’s unmatched threat detection with the Varonis Data Security Platform and MDDR service, we are speeding up our ability to stop data breach attempts earlier in the attack chain. Varonis integrates directly with email services like Microsoft Exchange Online to classify inbound and outbound traffic containing sensitive information, remediate posture issues on exposed mailboxes, and monitor anomalous email traffic for insider risks using industry-leading behavior analytics. The addition of Varonis Interceptor represents a significant advancement in end-to-end email and browser security. By harnessing the power of multimodal AI, it more effectively identifies and mitigates phishing threats than current solutions in the market. Varonis Interceptor allows businesses to confidently protect their inboxes and, subsequently, the sensitive data in their digital estate. Join us on Thursday, October 16, to see Interceptor in action live and explore how Varonis enables true AI-powered security outcomes. Want to see what Interceptor can do faster? Request a demo today. Sponsored and written by Varonis.

An active phishing campaign is impersonating the Home Office to compromise UK organizations licensed to sponsor foreign workers and students. The sophisticated campaign, which closely mimics official UK Home Office communications and web pages, aims to compromise sponsor license holders’ Sponsorship Management System (SMS) credentials. The compromised credentials are used to facilitate a range of elaborate immigration fraud schemes, extortion attempts and other monetization schemes, according to an investigation by cybersecurity firm Mimecast. The most elaborate of these involves creating fake job offers and visa sponsorship schemes, with threat actors observed charging victims between £15,000-£20,000 ($20,186-$26,914) for non-existent employment opportunities. The attacks target UK organizations holding sponsor licences across all industries and sectors, with particular focus on companies actively managing visa sponsorship programs and regular SMS system users. “The threat actors demonstrate advanced understanding of government communication patterns and user expectations within the UK immigration system,” the researchers noted. Samantha Clarke, threat research engineer at Mimecast, told Infosecurity that around 8000 emails related to this campaign were observed in the first half of July 2025. The campaign ramped up in early August, with around 2500 emails sent in the first six days of the month. On July 10, the Home Office issued a notification on the Sponsorship Management System (SMS) as well as direct communications to sponsors' key contacts and authorizing officers, warning of phishing scams that could compromise SMS account security. Organizations Sent Fake Home Office Warnings The campaign begins with target organizations being sent emails containing urgent alerts around SMS notifications or system alerts requiring immediate attention. SMS is the online tool used by sponsors to manage their license and meet their duties to notify the Home Office of changes in circumstances. These emails contain a link that direct users to fraudulent login pages designed to prompt them into entering SMS authentication credentials. The Mimecast report, published on August 12, highlighted common subject lines used in the initial phishing email. These include ‘A new message has been posted to your Sponsorship Management System’ and ‘Message Notification from SMS’. When the link on the initial email is clicked, the user is first sent to a CAPTCHA-gated URL, which acts as a filtering mechanism. They are then redirected to a phishing page that closely replicates the authentic SMS interface. The researchers said this replication is achieved through direct copying of the official SMS login page HTML, hotlinking of official assets and minimal but critical changes to the form submission process. The user credentials, once inputted, are sent to an attacker-controlled script rather than the legitimate authentication system. Follow-on Immigration Fraud and Extortion Schemes Once the attackers have captured the SMS credentials, they engage in a range of monetization schemes. These include: Selling access to compromised accounts on dark web forums Conducting extortion schemes against affected organizations Facilitating fraudulent Certificate of Sponsorship (CoS) issuance Creating fake job offers and visa sponsorship schemes via seemingly legitimate visa documents Mimecast advised UK organizations holding sponsor licenses to deploy anti-phishing tools that can detect government impersonation attempts and suspicious URL patterns. Additionally, firms should implement URL rewriting and sandboxing to analyze links before user interaction takes place. Image credit: James Copeland / Shutterstock.com