samsung

Innovative Tech Meta Ray-Banで向かいの席から撮影→指の動きからスマホのパスワードを解読する盗み見攻撃 イスラエルのテルアビブ大学に所属する研究者らが発表した論文「Watch and Crack: Password Inference from Smart-Glasses Video」は、スマートグラスで撮影された指の動きから、パスワードを解読する手法を実証した研究報告だ。 カフェなどの公共の場でスマートフォンにパスワードを入力する行為は、スマートグラスの普及によって脅威に晒されている。のぞき見防止フィルムやプライバシー表示機能で画面を隠していても、指先の動きさえ録画されればパスワードなどは突破される可能性がある。 また、普通の眼鏡に見える「Meta Ray-Ban」などのスマートグラスだと撮影されていることに気が付きにくい。しかも、前方から撮影されても指を解析されるため、ターゲットは油断してしまうだろう。後ろからのぞき見されるショルダーハッキングに注意しているだけではいけないことになる。 従来の映像解析は、キーが大きいタブレットや4桁のPINコードなど単純な条件に限られていた。今回実証した手法の場合は、相手の機種やキーボードアプリ（Gboard、iOS、Samsung）の事前知識を必要とせず、両親指や人差し指での自然なタイピングから、大文字・小文字・数字・記号の4つの配列すべてにまたがるパスワードを推定できるという。 まず画像認識AI（SAM 3）と手の追跡モデル（MediaPipe Hands）で指先をサブピクセル精度で追跡し、3Dニューラルネットワークでキーを押した瞬間を特定する。バックスペースキーによる入力修正も自動で検出・除外される。 さらに、大文字や数字を入力する際に必ず押されるシフトキーや配列切り替えキーの位置を手がかりに、画面上のキーボード領域を逆算する。そして、映像から算出した各キーの打鍵確率と、約22億件の漏えいパスワードの統計データを数学的手法で統合し、正解の候補を絞り込む。16人の参加者による実験では、全体の77.8%で打鍵数の特定に成功した。 撮影距離は最大1.8mまで有効で、Pixel 10、iPhone 16、Galaxy A53といったスマートフォンの全てで効果が確認された。一方、指の移動が極端に小さく高速に打つ若年層や、迷いながら打つ人、両親指でほぼ同時に打つ人などの場合は精度が低下する限界も確認された。 Source and Image Credits: Orenbach, Yoav, and Avishai Wool. “Watch and Crack: Password Inference from Smart-Glasses Video.” arXiv preprint arXiv:2609.06539 (2026). Copyright © ITmedia, Inc. All Rights Reserved. Innovative Tech 2019年にスタートした本連載「Innovative Tech」は、世界中の幅広い分野から最先端の研究論文を独自視点で厳選、解説している。執筆は研究論文メディア「Seamless」（シームレス）を主宰し、日課として数多くの論文に目を通す山下氏が担当。イラストや漫画は、同メディア所属のアーティスト・おね氏が手掛けている。

A reverse engineer claims that their Samsung smart TV has been collecting audio without their consent. The researcher, who calls themselves “log4j,” said in an X thread that their 2025 Samsung TV was delivering ads that seemed suspiciously relevant to them. Investigating, they claim to have found a package called “always on voice,” which uses the TV’s built-in microphone to collect sounds from the room. The software logged events such as dogs barking and babies crying, log4j said, adding that these events are counted and reported back to Samsung. Samsung advertises similar features on some of its AI-enabled TVs. The company says its Pet Care and Family Care features use on-device AI to detect a dog barking or a baby crying and alert users. This doesn’t confirm log4j’s claims about the package or the data it reports. However, log4j says that on TVs without microphones, the software searches for connected devices such as soundbars or Bluetooth microphones and uses those to listen instead. The TV was also collecting geolocation data and mapping local networks, they said, adding that it ran a service on the TV that analyzes traffic on the local network. log4j also says that the system uses a neural network tailored to a particular region to analyze your viewing habits. A neural network is a type of machine-learning model. log4j said : “…in youtube it’s looking for videos you watch, ads you skip, progress in a video, etc.” The tracking that’s on the record While log4j’s claims haven’t been independently verified, Samsung has used automatic content recognition (ACR), a system that identifies what’s on screen. Earlier this year, Texas Attorney General Ken Paxton settled with Samsung, getting the company to agree that it would get Texans’ express consent before collecting or processing ACR viewing data. A class action complaint filed in New York in January also accused Samsung of using ACR to record video and audio passing through the TV. The complaint accused Samsung of using machine-learning algorithms to analyze and link this data back to consumers. This includes information displayed when the TV is used as a computer monitor, it said, adding that the information can be linked to different people using the TV. The case was voluntarily dismissed in July. The New York complaint added: “Samsung exploits this Sensitive Information by (a) selling targeted advertisements to its corporate advertising clients, and (b) providing it to third-party advertising companies such as Google and X (f/k/a Twitter) for targeted advertising.” A separate class action complaint filed in California this month accuses Samsung Electronics America of not being transparent about its ACR data collection. This isn’t the first time that Samsung has been criticized for allegedly picking up in-room audio. In 2015, privacy group EPIC filed a complaint with the FTC, accusing Samsung of recording conversations and relaying them elsewhere. The nonprofit said at the time: “Everything a user says in front of the Samsung SmartTV is recorded and transmitted over the internet to a third party regardless of whether it is related to the provision of the service.” EPIC identified text-to-speech company Nuance as the third party. What you can do We can’t confirm all of log4j’s allegations, and Samsung doesn’t appear to have commented. But there have been multiple complaints about the data smart TVs collect. You can turn off features such as ACR, internet-based advertising, and voice recognition where possible. log4j has also raised concerns about whether monitoring services are truly off. They said : “ACR off via your preferences menu on samsung TVs should not be assumed to actually disable ACR telemetry. i’ve found several instances where the checks (whether deliberate or not) are either incorrectly implemented or not checked at all.” log4j says disconnecting your TV from Wi-Fi is the best option for immediate peace of mind, though that also means losing internet-based TV features. They also suggest using an Apple TV or a modifiable Android TV box with open-source software. What do cybercriminals know about you? Use Malwarebytes’ free Digital Footprint scan to see whether your personal information has been exposed online. SCAN NOW

A malicious release of LiteLLM pushed to PyPI and Docker registries on Wednesday carried malware built to harvest cloud logins, developer secrets and API keys for AI services from anyone who installed it. “LiteLLM got compromised… again,” security researcher Kevin Beaumont wrote on Mastodon. In August, Hudson Rock analyzed a 153GB archive stolen during an earlier LiteLLM supply chain attack, in March, and linked it to 2,488 corporate domains, including AWS, Samsung, Cisco and Salesforce, … More → The post LiteLLM hit by second supply chain attack this year, users urged to rotate credentials appeared first on Help Net Security .

On the second day of Pwn2Own Ireland 2026, security researchers collected $232,500 in cash awards after exploiting 45 unique zero-day vulnerabilities. The day's highlight was the Samsung Galaxy S26 flagship getting hacked three times by KAIST Hacking Lab's Kyeongmin Kim, PetoWorks, and Mobile Hacking Lab's Dimitrios Valsamaras and Ken Gannon. Jack Dates of RET2 Systems demoed a Sonos Era 300 exploit chain in under a minute, and Out of Bounds team's HaeJung Yang was awarded $40,000 for hacking Dynamo in the AI Infrastructure category. PetoWorks, Yves Bieri of Xint, Kyeongmin Kim, _McCaulay, and Doyensec's Yassine Bengana and Maxence Schmitt also hacked the Home Assistant Green smart home hub, while Ikotas Labs breached the Oracle Autonomous AI Database using a seven-chain zero-day exploit. Before day two began, Kyeongmin Kim withdrew his attempt at a USB-based attack targeting the Google Pixel 10. Trend Micro's Zero Day Initiative (ZDI) organizes the competition to identify zero-day flaws in fully patched devices before attackers exploit them in the wild. According to Pwn2Own rules, all devices run the latest firmware versions, and contestants must compromise the target and demonstrate arbitrary code execution. After zero-days are exploited and disclosed at Pwn2Own, vendors have 90 days to patch their software before ZDI publicly discloses them. Throughout the Pwn2Own Ireland 2026 contest, competitors target products in seven categories, including mobile phones (Samsung Galaxy S26 and Google Pixel 10), messaging apps, smart home devices, printers, AI infrastructure, AI coding apps, and a new category where hackers will try to exploit wellness healthcare devices. While Apple's iPhone 17 was also a potential target with a maximum award of $300,000 for a remote hack, no contestant registered for an attempt. Interrupt Labs, Ikotas Labs, and Nguyen Thanh Dat of Viettel Cyber Security also hacked Samsung's Galaxy S26 flagship on day one, but some of the bugs exploited were already known to the vendor. Vũ Chí Thành and Huỳnh Đức Tin of VinSOC, who topped the leaderboard on the first day, won $40,000 for a five-zero-day exploit chain targeting the Oracle Autonomous AI Database, plus an additional $40,000 for chaining seven zero-daysto exploit a Philips Hue Bridge Pro smart lighting hub. On the third day, security researchers will attempt to hack multiple smart home, AI infrastructure, and printer devices, as well as the Samsung Galaxy S26 and Google Pixel 10 smartphones again. During the Pwn2Own Ireland 2025 competition, hackers demoed 73 zero-day flaws to earn $1,024,750. Summoning Team won the contest and collected $187,500 after hacking the Samsung Galaxy S25, the Home Assistant Green, the QNAP TS-453E NAS, and multiple Synology devices. Join Mikko Hyppönen and security leaders from the NFL, CHANEL, and Atlassian for a two-hour digital summit on what AI-speed attacks change, what defenders should stop doing, and how to validate, decide, fix, and re-validate at machine speed. Save your seat

Google heeft kwetsbaarheden verholpen in Android. Samsung heeft de voor Samsung Mobile relevante kwetsbaarheden verholpen in hun producten. De kwetsbaarheden betreffen onder andere onjuiste bounds checking, ontbrekende permissiecontroles, integer overflows, onjuist geïnitialiseerde pointers, type confusion, race conditions en onjuiste exception handling. Deze fouten kunnen leiden tot out-of-bounds memory reads en writes, geheugen corruptie, privilege escalatie, informatielekken, denial-of-service situaties en controle over het systeem door een aanvaller. Exploitatie kan lokaal of op afstand plaatsvinden, vaak zonder dat er gebruikersinteractie of verhoogde rechten nodig zijn. De kwetsbaarheden zijn aanwezig in diverse functies en modules die gebruikt worden in verschillende softwareomgevingen, waaronder Bluetooth subsystemen, netwerk parsing, device management en audio controllers.

Google has published security fixes for Android . The October updates are available for Android operating system versions 14, 15, 16, 16-qpr2, and 17 where “qpr” stands for Quarterly Platform Release. QPRs are interim updates Google pushes out between major yearly cycles. These Android security bulletins contain valuable information for Android users. The updates are important as they fix several vulnerabilities rated as Critical, which can be exploited without users even doing anything. However, we don’t write about them very often. The main reason for that lack of attention is that many Android users aren’t able to get the updates immediately and lots of others will never see them. Google’s October fixes list Android 14 and newer, but a substantial share of the Android ecosystem still runs older software. In the distribution figures reported about last year, roughly four in ten devices were running Android 12 or earlier . This was while Android 13 still received updates which ended early March 2026. And then there is the patch gap caused by device vendors meaning that even phones that remain supported may have to wait until the vendors test and sometimes modify the patches, so they work with their devices. And other vendors are on a different stride. For example, Samsung lists devices including the Galaxy S22 series and several Galaxy A models on a quarterly rather than a monthly security-update schedule. You can imagine that cybercriminals love to get their hands on known vulnerabilities that many users have not yet been able to patch. Looking at the number of users that are unable to patch immediately that makes for an enormous target group. How to stay safe Get the updates as soon as they are made available to you is the most important word of advice here. For most Android phones this is the general idea: open Settings , find Software update or System update (sometimes under System or About phone ), check for updates , and follow the instructions to download, install, and restart. The menu names vary between vendors and even models, but the process is broadly the same. If that doesn’t work for you or you can’t find instructions for your specific device, post what you’re looking for in the comments. We have lots of helpful readers. Other things you can do: Use up-to-date real-time anti-malware protection . Refrain from sideloading unless it’s absolutely necessary. Make sure Google Play Protect is enabled. If you are a Malwarebytes Mobile Security user, you can check this using Trusted Advisor. Should you decide to get a new Android phone, read our blog Upgrading your Android device? Read this first . Scammers know more about you than you think.   Malwarebytes Mobile Security protects you from phishing , scam texts, malicious sites, and more. With real-time AI-powered Scam Guard built right in.  .wp-block-kadence-advancedbtn.kb-btns97520_6de029-2f{gap:var(--global-kb-gap-xs, 0.5rem );justify-content:flex-start;align-items:center;}.kt-btns97520_6de029-2f .kt-button{font-weight:normal;font-style:normal;}.kt-btns97520_6de029-2f .kt-btn-wrap-0{margin-right:5px;}.wp-block-kadence-advancedbtn.kt-btns97520_6de029-2f .kt-btn-wrap-0 .kt-button{color:#555555;border-color:#555555;}.wp-block-kadence-advancedbtn.kt-btns97520_6de029-2f .kt-btn-wrap-0 .kt-button:hover, .wp-block-kadence-advancedbtn.kt-btns97520_6de029-2f .kt-btn-wrap-0 .kt-button:focus{color:#ffffff;border-color:#444444;}.wp-block-kadence-advancedbtn.kt-btns97520_6de029-2f .kt-btn-wrap-0 .kt-button::before{display:none;}.wp-block-kadence-advancedbtn.kt-btns97520_6de029-2f .kt-btn-wrap-0 .kt-button:hover, .wp-block-kadence-advancedbtn.kt-btns97520_6de029-2f .kt-btn-wrap-0 .kt-button:focus{background:#444444;}ul.menu .wp-block-kadence-advancedbtn .kb-btn97520_074a67-a8.kb-button{width:initial;}.wp-block-kadence-advancedbtn .kb-btn97520_074a67-a8.kb-button{color:var(--color-neutral-10);background:var(--color-brand-blue-60);border-top-left-radius:40px;border-top-right-radius:40px;border-bottom-right-radius:40px;border-bottom-left-radius:40px;box-shadow:1px 1px 2px 0px rgba(0, 0, 0, 0.2);}.wp-block-kadence-advancedbtn .kb-btn97520_074a67-a8.kb-button:hover, .wp-block-kadence-advancedbtn .kb-btn97520_074a67-a8.kb-button:focus{box-shadow:0px 4px 0px 0px var(--color-brand-blue-90);} Try it now

The Japanese chip testing giant said hackers stole personal information from its servers in the February 2026 cyberattack. The post Advantest Discloses Data Breach Months After Ransomware Attack appeared first on SecurityWeek .

L’azienda: “ Prestazioni già in linea con alcune tecnologie cinesi e supera numerosi modelli concorrenti open-weight “. L’Europa si rilancia nel confronto tecnologico e geo-economico dell’AI con la Cina e gli Usa, grazie alla Francia e alla startup Mistral che ha svelato Large 4 , il suo ultimo modello. Si è interrotto così un intervallo di circa cinque mesi dall’ultima release di rilievo. Secondo l’azienda, “ il nuovo sistema supera numerosi modelli concorrenti open-weight “. Mistral non svelava un nuovo modello dal mese di maggio. Ad agosto aveva invece iniziato a offrire ai propri clienti anche modelli sviluppati da società terze, a partire dal GLM 5.3 di Z.ai , azienda con sede a Pechino. Noto internamente come Le Chonk e ufficialmente denominato Mistral Large 4 , sarà reso pubblicamente disponibile il 27 ottobre . Prima del lancio, esperti di cybersecurity e autorità governative avranno accesso a una versione del modello con meno restrizioni di sicurezza, con l’obiettivo di testarne le capacità . “ L’Europa può competere “ “ Il modello che annunciamo oggi è superiore ai modelli cinesi sotto alcuni aspetti, incluso il cyber “, ha dichiarato l’Amministratore delegato di Mistral , Artur Mensch , durante un evento di presentazione ad Abu Dhabi . “ La narrazione secondo cui l’Europa non può competere non è vera “. Artur Mensch, Amministratore delegato di Mistral Secondo Pierre Stock , Vice Presidente della ricerca scientifica di Mistral , Large 4 si colloca tra i sistemi open-weight più potenti attualmente disponibili. Il termine open-weight indica modelli i cui pesi possono essere scaricati dagli utenti e utilizzati direttamente sulla propria infrastruttura hardware. Molti modelli cinesi adottano questo approccio. I sistemi AI statunitensi più avanzati sono invece generalmente closed , ossia accessibili esclusivamente attraverso l’infrastruttura dei rispettivi sviluppatori. Con oltre 3 miliardi di euro raccolti nell’ultimo giro di finanziamenti, Mistral punta così a presentarsi come un’alternativa europea e più sicura ai grandi operatori globali dell’intelligenza artificiale. “ Stiamo colmando il divario con i modelli di frontiera “ In termini tecnici e di velocità il modello Large 4 sta “ colmando il divario con i modelli di frontiera ” in diversi ambiti operativi. Tra questi: programmazione. Finanza. Analisi geospaziale. Produzione industriale. Progettazione di prodotti. Il modello è stato addestrato utilizzando l’ infrastruttura proprietaria di Mistral . Stock ha spiegato inoltre che l’obiettivo della società è costruire sinergie “ molto profonde ” con i propri finanziatori, tra cui ASML , e con il nuovo investitore Samsung , lungo l’intera catena del valore tecnologica. Seguici anche sul nostro canale WhatsApp Vai al sito di Cybersecurity Italia. L'articolo Francia, Mistral lancia il suo nuovo modello AI. Così l’Europa vuole sfidare i giganti dell’intelligenza artificiale sembra essere il primo su CyberSecurity Italia .

On the first day of the Pwn2Own Ireland 2026 competition, security researchers hacked the Samsung Galaxy S26 twice and earned $388,500 after exploiting 32 zero-days. [...]

This week, the useful words are boring ones: inspect, cache, compile, store, trust. Each sounds harmless. Each can become an attack path when a system does a little more than people expect. A model check can run code. A cache can mix up requests. A public secret can stay useful for years. That is the lesson running through the list. Attackers do not always need a brilliant new trick. They can

Researchers show that file-change notification systems can leak keystroke timing, browsing activity, and WhatsApp media events. The post Windows, Linux, Android File Notification Systems Leak User Activity appeared first on SecurityWeek .

Un normale applicazione Android senza permessi richiesti può ottenere il controllo completo su uno smartphone di punta moderno. Il ricercatore di Calif Lucas Maar ha mostrato un’attacco su dispositivi Samsung, Xiaomi, Oppo e OnePlus con bootloader bloccato e firmware originale, trasformando un’applicazione non privilegiata in un processo con diritti massimi. Il metodo ha preso il nome di OEMpocalypse Now . Un dettaglio importante è che Maar non ha trovato un unico bug in tutti gli smartphone. Il ricercatore ha sviluppato una strategia unificata e ha costruito su di essa tre catene di sfruttamento separate per Samsung, Xiaomi e la famiglia Oppo, che include anche OnePlus e Realme . Questo approccio permette di utilizzare l’attacco su numerosi modelli di un singolo produttore con un volume di modifiche molto inferiore. Nelle dimostrazioni sono stati riportati attacchi contro Galaxy S26 Ultra e Galaxy S26 con Android 16, Galaxy S23 con Android 14, Xiaomi 17, Oppo Find X9 Ultra e OnePlus Ace 6 Ultra. Gli smartphone utilizzavano processori Qualcomm Snapdragon, Samsung Exynos e MediaTek Dimensity , mentre le versioni del kernel Linux variavano da 5.15 a 6.12. Galaxy S26, Xiaomi 17, Oppo Find X9 Ultra e OnePlus Ace 6 Ultra funzionavano con firmware originali con aggiornamenti di luglio 2026, che rimanevano gli ultimi disponibili al momento del test all’inizio di agosto. Maar ha deliberatamente evitato di cercare errori nella parte comune di Android o in un processore specifico. L’obiettivo principale è stato il codice che i produttori aggiungono sopra Android: servizi di sistema proprietari, applicazioni e driver di One UI, HyperOS e ColorOS. Un componente del genere viene spesso utilizzato contemporaneamente in decine di modelli con diversi processori, quindi una vulnerabilità trovata su un device ottiene una portata maggiore. L’attacco si struttura in due fasi. Inizialmente, un’applicazione dannosa, se necessario, esce dalla sandbox di sistema attraverso un errore nel meccanismo di interazione con un componente privilegiato del produttore. La causa può essere l’assenza di verifica del processo chiamante, un componente di sistema aperto in modo errato o una verifica insufficiente del percorso del file. L’accesso ottenuto permette di accedere a un driver del kernel, che un’applicazione normale non dovrebbe utilizzare direttamente. La seconda fase sfrutta un errore di classe use-after-free  nel driver del produttore. Il driver continua a mantenere un riferimento a una pagina di memoria fisica dopo la liberazione, e il sistema riesce ad allocare la stessa sezione per altri dati del kernel. L’ exploit ottiene la possibilità di leggere o modificare il contenuto della pagina riutilizzata e, in ultima analisi, raggiunge l’accesso root . L’approccio si è rivelato resistente a diversi meccanismi di protezione di Android.   Per Samsung, lo studio copre almeno i modelli di punta Galaxy S23, S24, S25 e S26, nonché i modelli pieghevoli moderni della serie Z. Per Xiaomi, i componenti vulnerabili, secondo la stima di Maar, sono presenti su una parte significativa degli smartphone di fascia media e alta. La stessa strategia si applica ai modelli di punta moderni di Oppo, OnePlus e Realme. Gli smartphone Google Pixel con un Android vicino allo standard non rientrano nello schema descritto, poiché l’attacco è mirato specificamente al codice aggiuntivo dei produttori. Il bollettino di settembre di Samsung elenca separatamente due vulnerabilità trovate da Maar nel componente DualDAR . Il CVE-2026-21101 è legato alla verifica errata dei dati di input da parte del driver, mentre CVE-2026-21102 rappresenta un use-after-free e permette a un processo locale privilegiato di eseguire codice con diritti root. Entrambi i problemi interessavano i dispositivi Samsung con Android 14, 15, 16 e 17 prima dell’installazione del pacchetto di correzioni di settembre. Il collegamento dettagliato delle vulnerabilità DualDAR con la catena dimostrata di OEMpocalypse l’autore non ha ancora reso pubblico. Non esiste un exploit pronto disponibile pubblicamente. Maar ha pubblicato video di attacchi riusciti, ma i dettagli tecnici specifici delle catene ha deciso di pubblicarli nelle parti successive dello studio, separatamente per Samsung, Xiaomi e la famiglia Oppo.  L'articolo Arriva OEMpocalypse Now: come diventare root su smartphone top di gamma proviene da Red Hot Cyber .

A OnePlus 15 running the latest OxygenOS can be rooted by a malicious app the owner installs, one that asks for no special permissions. A researcher, Rasmus Moorats, chained two flaws in OnePlus's own software to gain root access, the highest level of control over an Android phone. OnePlus told him the same flaws affect many more of its own devices and those of OPPO, though it has not

I proprietari dei frigoriferi “intelligenti” Samsung Bespoke AI hanno scoperto che, dopo l’installazione dell’ultimo aggiornamento del firmware, i dispositivi sono completamente fuori uso. Alcuni utenti hanno visto il loro elettrodomestico non funzionare più, costringendoli a buttare via i prodotti rovinati. I primi rapporti sul problema hanno iniziato ad arrivare il 22 settembre 2026 da utenti della Corea del Sud. Secondo quanto riportato dal media locale Star News Korea , il malfunzionamento ha principalmente colpito i frigoriferi a quattro porte Samsung del 2024 e modelli più recenti. I problemi sono sorti dopo aver tentato di aggiornare il firmware dei frigoriferi tramite la piattaforma SmartThings per la casa intelligente . I dispositivi si sono spenti improvvisamente e hanno smesso di funzionare, mentre nell’app SmartThings apparivano come non disponibile. Alcuni proprietari hanno anche segnalato che durante l’aggiornamento il display integrato del frigorifero si bloccava sul messaggio di verifica dell’app SmartThings. Sul forum Samsung coreano è apparso un gran numero di reclami , con alcuni utenti che scrivono di aver acquistato i propri frigoriferi solo un anno fa o meno . Un utente ha raccontato che, dopo il malfunzionamento, sia il vano frigorifero che quello freezer hanno smesso di funzionare , rovinando tutti i prodotti conservati, che poi ha dovuto buttare. Secondo quanto riportato da SBS Korea , i proprietari dei frigoriferi colpiti hanno presentato centinaia di reclami al supporto Samsung. I giornalisti osservano che, per ora, le lamentele arrivano solo dalla Corea del Sud, dove il problema è emerso in un momento particolarmente sfortunato : il 24 settembre nel paese inizia il Chuseok, uno dei più grandi festival nazionali. Inoltre, a causa del festival, parte del personale e degli specialisti del servizio Samsung è risultata irreperibile. Alcuni proprietari dei dispositivi sono stati addirittura che un tecnico potrà recarsi da loro solo a ottobre scrive uno dei colpiti. Un altro utente ha espresso la sua opinione sulla situazione in modo più conciso: “Mai aggiornare i vostri frigoriferi Samsung!” Il 23 settembre i rappresentanti di Samsung hanno interrotto il rilascio del firmware problematico, e, come riporta ZDNET Korea , gli specialisti dell’azienda prevedono di risolvere il problema in alcuni frigoriferi già oggi, 24 settembre. I rappresentanti di Samsung hanno confermato il problema sul forum ufficiale dell’azienda. Secondo loro, il 22 settembre, durante il test dell’aggiornamento del software, si è verificato un errore che ha causato problemi di alimentazione e funzionamento dello schermo in alcuni dispositivi. Si segnala che al momento i centri di assistenza Samsung stanno adottando misure di emergenza e si consiglia a tutti i colpiti di rivolgersi al supporto. L'articolo Casa smart, frigorifero offline! Questo è l’aggiornamento più “indigesto” di sempre proviene da Red Hot Cyber .

​The U.S. Cybersecurity and Infrastructure Security Agency (CISA) warned federal agencies on Wednesday that ransomware gangs are now also exploiting a critical JetBrains TeamCity vulnerability patched in July. [...]

A new Android banking trojan called RemControl tricks victims into installing a fake TV app, then takes control of their phones to steal banking PINs, Group-IB has found. Researchers confirmed that the malware targets customers of more than 30 banks in Italy, France, Spain, Poland, Portugal, Canada and some Gulf states. The first samples were submitted to VirusTotal on July 19, 2026. The domain used for its command and control (C2) server was registered on … More → The post New Android malware RemControl steals banking PINs and blocks removal attempts appeared first on Help Net Security .

A new Android malware-as-a-service (MaaS) platform called RemControl is targeting users through malvertising campaigns that impersonate the TVTap IPTV application. [...]

Academic publishing giant Elsevier confirmed a compromise this week after students found its platform redirecting users to a cybercriminal crew's leak page. One Reddit user, a self-described nursing student, highlighted the issue on September 22, posting a screenshot of LAPSUS$’s leak site after trying to access “homework and textbooks.” “Every time I try to open the Elsevier website, I am met with this,” they wrote. “Anyone know anything or have any explanation? Totally creepy.” Amsterdam-based Elsevier told The Register it was briefly compromised following an attack Monday, but played down the wider impact. “On September 21, Elsevier identified that visitors to select platforms were being redirected to a third-party page,” a spokesperson said. “Our cybersecurity team responded immediately, resolving the issue and restoring normal service. “Our investigation indicates that this was a narrowly scoped, limited-duration event involving the temporary redirection of traffic for certain web properties. There is no indication that core platforms, customer data, research content, or operational systems were compromised.” Elsevier did not respond to additional questions related to the specific platforms that were affected or for how long LAPSUS$’ redirect was in place. The company is best known for its ScienceDirect platform, which hosts scientific, technical, and medical journal articles. It is also behind ClinicalKey, an AI-powered platform designed to provide medical professionals fast answers to care queries, and LeapSpace – an AI-assisted workspace for academic researchers. LAPSUS$, meanwhile, is better known for its criminal enterprises, namely big-name cyberattacks on the likes of Rockstar Games, which led to the earliest high-profile Grand Theft Auto VI leaks, and more recently, attacks on Adidas and GitHub. The online assault on Rockstar Games was part of a wider spree of crimes carried out when the group was in its pomp between 2020 and 2022. Other victims included BT, Microsoft, Okta, Samsung, and Vodafone, which in turn stoked a concentrated law enforcement operation to disrupt the teenage criminals behind it. After a protracted break, the LAPSUS$ name returned in 2025, partnering with Scattered Spider and ShinyHunters in another string of cyberattacks affecting household names, before splitting up and activity dropping to a modest six attacks per month, according to SOCRadar. ®

Researchers from the i2CAT Foundation, the University of Murcia, and NEC Laboratories Europe built a low-cost tool called 5G-Shark that lures a target phone onto a fake base station and questions it, then used it to audit commercial 5G networks. On the standalone-5G networks they tested, operators concealed the phone’s permanent identity correctly in every case but one. The same networks still handed out temporary IDs in a pattern predictable enough that an observer can … More → The post A cheap fake base station can still track 5G subscribers appeared first on Help Net Security .

Sempre più organizzazioni devono scegliere tra soluzioni in cloud e modelli installati internamente. Ma il dibattito sulla sicurezza è impostato male. Il perché ce lo spiega Mirco Marchetti, professore associato di UniMoRe e direttore del Centro di Ricerca Interdipartimentale sulla Sicurezza L'articolo Commerciale o open source, l’AI in azienda tra rischi noti e trappole nascoste proviene da Cyber Security 360 .

I ricercatori cinesi hanno delineato un percorso per la produzione di transistor da 3 nm utilizzando la litografia a ultravioletti profondi (DUV) invece della più moderna EUV . Lo sviluppo è attualmente in una fase iniziale e non è pronto per la produzione di massa, ma dimostra che la Cina continua a cercare modi per produrre semiconduttori avanzati senza accesso alle attrezzature occidentali più recenti. L’Istituto di microelettronica dell’Accademia cinese delle scienze ( IMECAS ) ha sviluppato un processo tecnologico preliminare per la fabbricazione di transistor a griglia circolare (gate-all-around, GAA) utilizzando esclusivamente attrezzature DUV disponibili. La notizia è stata riportata dal South China Morning Post , citando un discorso dell’ingegnere capo dell’istituto Ye Tianchun. Il GAA non rappresenta una sostituzione della litografia EUV, ma una struttura di transistor in cui la griglia circonda il canale su tutti i lati. Per ottenere strutture corrispondenti alla classe da 3 nm con la DUV è necessario un multiplo pattern sulla piastra. Questo approccio richiede un maggior numero di esposizioni, maschere e operazioni tecnologiche, complicando così la produzione, aumentando i costi e creando ulteriori difficoltà con la precisione dell’allineamento degli strati e il rendimento dei cristalli buoni. Il lavoro viene svolto in un contesto di restrizioni sulle forniture di attrezzature alla Cina. L’azienda ASML non può vendere ai clienti cinesi le apparecchiature litografiche EUV, utilizzate dai produttori per una produzione economicamente efficiente dei chip più avanzati. Teoricamente, alcuni strati critici possono essere formati anche con la DUV, ma l’esposizione multipla rende il processo tecnologico significativamente più complesso e costoso. I transistor GAA sono già utilizzati nella produzione in serie: Samsung ha iniziato a produrre chip da 3 nm con questa struttura nel 2022, e TSMC la utilizza nel suo processo tecnologico da 2 nm N2. All’Accademia cinese delle scienze sottolineano che lo sviluppo è ancora lontano dalla produzione di massa. Secondo Ye Tianchun, le sue prospettive diventeranno più chiare solo dopo i test delle piastre sulle linee di produzione reali. Anche la produzione riuscita di singoli transistor GAA non significa che il processo tecnologico completo sia pronto: è necessario padroneggiare la formazione delle interconnessioni e di altri strati, ottenere un rendimento accettabile di cristalli buoni e garantire la stabilità della produzione. Nel 2025, i ricercatori cinesi hanno anche presentato una sorgente a stato solido di radiazione coerente con una lunghezza d’onda di 193 nm, corrispondente a quella utilizzata nella litografia DUV moderna. Lo sviluppo è considerato come una possibile base per i componenti dei futuri sistemi litografici, ma da solo non rappresenta una sostituzione immediata delle sorgenti di radiazione e degli scanner industriali ASML. Huawei Technologies sta sviluppando un proprio percorso per il ulteriore scaling dei chip. A maggio, l’azienda ha presentato il concetto di Tau Scaling Law, che prevede l’aumento della densità di calcolo attraverso modifiche architettoniche e il cosiddetto scaling temporale. Huawei prevede che entro il 2031 i suoi processori ad alte prestazioni potranno raggiungere una densità di transistor equivalente a quella prevista per il processo tecnologico da 1,4 nm. In questo caso, si parla di densità equivalente, non di padronanza della produzione fisica secondo gli standard da 1,4 nm. L'articolo La Cina sfida i limiti: vuole fare chip da 3 nm senza macchine EUV proviene da Red Hot Cyber .

La piattaforma italiana ha individuato tre 0-day di severità critica. Al centro della ricerca c’è la funzionalità White Box, che correla codice sorgente, documentazione e applicazione live per costruire scenari di attacco e validare la reale sfruttabilità dei finding. Nel mondo della cybersecurity, la maggior parte degli strumenti di vulnerability assessment lavora partendo da ciò che è già noto: firme, pattern conosciuti, configurazioni errate già catalogate, CVE pubbliche e comportamenti già documentati. È un approccio indispensabile, ma non sufficiente quando l’obiettivo è individuare vulnerabilità che non sono ancora presenti in alcun database e per le quali non esiste una firma da cercare. È proprio su questo terreno che Ares Cyber AI ha raggiunto uno dei risultati più significativi dalla nascita del progetto: la scoperta di tre vulnerabilità zero-day di severità critica in prodotti applicativi Samsung. Un risultato ottenuto attraverso un processo di analisi che non si limita a cercare debolezze note, ma prova a comprendere il funzionamento dell’applicazione, il suo contesto e le possibili catene di attacco. Tre zero-day critiche individuate nei prodotti Samsung Le tre vulnerabilità sono state individuate diversi mesi fa e sono state gestite seguendo un percorso di responsible disclosure. Al momento non è possibile rendere pubblici i nomi delle applicazioni coinvolte, i componenti interessati, i vettori di attacco o ulteriori dettagli che possano anticipare informazioni non ancora autorizzate alla divulgazione. Samsung ha riconosciuto il contributo inserendo Ares Cyber AI (Bitcube Security) nella propria Hall of Fame, pubblicata all’interno del programma Samsung Security for Smart TV, Audio and Displays. Il riconoscimento è consultabile direttamente sul sito ufficiale Samsung: Hall of Fame Samsung . Il punto più interessante, tuttavia, non è soltanto il numero delle vulnerabilità individuate o la loro severità. È il modo in cui sono emerse: non attraverso la semplice corrispondenza con vulnerabilità già conosciute, ma attraverso un modello di analisi progettato per correlare informazioni diverse e trasformarle in ipotesi di attacco verificabili. Perché trovare una zero-day richiede più di uno scanner Uno scanner tradizionale può essere estremamente efficace nel riconoscere una versione vulnerabile, un servizio esposto, una dipendenza affetta da una CVE o un pattern già noto. Ma una zero-day, per definizione, non offre necessariamente nessuno di questi riferimenti. Può nascere da una combinazione specifica di logica applicativa, controlli di autorizzazione, flussi di dati, assunzioni progettuali e comportamenti runtime che, presi singolarmente, non sembrano vulnerabili. È qui che il concetto di contesto diventa centrale. Una stessa porzione di codice può essere irrilevante in un’applicazione e diventare critica in un’altra, a seconda di come viene raggiunta, di quali dati gestisce, dei privilegi disponibili, dell’esposizione dell’endpoint e del ruolo che quella funzione ricopre nel processo di business. Per individuare vulnerabilità di questo tipo non basta quindi produrre un finding: bisogna ricostruire lo scenario nel quale quel finding può essere utilizzato da un attaccante e, soprattutto, capire se lo scenario è realmente percorribile. White Box: codice, documentazione e applicazione live in un unico modello di analisi La funzionalità White Box di Ares Cyber AI nasce esattamente con questo obiettivo. La piattaforma può acquisire contemporaneamente il codice sorgente dell’applicazione, la documentazione tecnica e funzionale e l’applicazione live da sottoporre a test. Queste fonti non vengono trattate come tre analisi indipendenti: i risultati vengono correlati per ottenere una visione più completa del sistema e delle sue possibili superfici di attacco. Il codice sorgente permette di osservare ciò che dall’esterno spesso non è visibile: controlli, flussi, funzioni, logiche di autorizzazione, trasformazioni dei dati e percorsi applicativi. La documentazione aiuta a comprendere come il sistema dovrebbe funzionare, quali ruoli esistono, quali operazioni sono previste e quali processi hanno maggiore rilevanza per il business. L’applicazione live, infine, consente di confrontare queste informazioni con il comportamento reale del sistema e con ciò che è effettivamente raggiungibile durante un test. Il valore sta nella correlazione. Un’informazione individuata nel codice può assumere un significato diverso quando viene collegata a un endpoint live; una funzione descritta nella documentazione può far emergere un confine di autorizzazione da verificare; un comportamento osservato dinamicamente può essere ricondotto a una specifica logica presente nel sorgente. È da queste relazioni che possono emergere scenari di attacco che un’analisi isolata rischierebbe di non vedere. Dal finding allo scenario di attacco: il ruolo del business Ares non considera tutti i progetti allo stesso modo. La fase di analisi tiene conto dello scenario nel quale l’applicazione viene utilizzata e delle caratteristiche del business che deve proteggere. Questo permette di valutare una vulnerabilità non soltanto in base alla sua categoria tecnica, ma anche in base a ciò che potrebbe consentire di ottenere in quello specifico ambiente. Un controllo apparentemente secondario può diventare il punto di partenza di una catena di attacco se consente di raggiungere una funzione sensibile. Allo stesso modo, una vulnerabilità teoricamente grave può avere un impatto molto inferiore se il percorso necessario per sfruttarla non è realmente accessibile. La piattaforma prova quindi a collegare i singoli segnali e a ragionare in termini di percorso offensivo, privilegi, condizioni necessarie e possibile impatto. Questa impostazione è particolarmente importante nella ricerca di zero-day, perché molte vulnerabilità inedite non si presentano come un singolo errore evidente. Possono emergere dalla combinazione di più condizioni corrette solo in apparenza, ma pericolose quando vengono concatenate dal punto di vista di un attaccante. Ares prova ciò che trova per misurare la reale sfruttabilità La seconda parte del processo è la validazione. Una delle differenze più rilevanti tra un finding teorico e una vulnerabilità realmente critica è la possibilità di dimostrarne la sfruttabilità. Per questo Ares non si ferma alla generazione dell’alert, ma prova in maniera controllata le ipotesi emerse durante l’analisi, verificando se le condizioni necessarie all’attacco sono effettivamente presenti. L’obiettivo non è semplicemente aumentare il numero di vulnerabilità riportate, ma ridurre la distanza tra ciò che uno strumento segnala e ciò che un attaccante potrebbe davvero utilizzare. La validazione consente di distinguere più rapidamente i falsi positivi, ridimensionare finding che non risultano sfruttabili e dare priorità a quelli che possono essere dimostrati sul sistema reale. In altre parole, l’AI non viene utilizzata soltanto per descrivere un possibile problema, ma per correlare le evidenze disponibili, costruire una possibile strategia di attacco e sottoporla a verifica. È un approccio che avvicina l’automazione al metodo di lavoro di un penetration tester: osservare, formulare un’ipotesi, cercare le condizioni che la rendono possibile e provare a validarla. Come può emergere una zero-day senza conoscere prima la vulnerabilità È proprio questa combinazione a spiegare come Ares possa individuare vulnerabilità zero-day. Se un sistema si limita a confrontare ciò che vede con un elenco di problemi già conosciuti, può trovare soltanto ciò che è stato descritto in precedenza. Un modello White Box, invece, può partire dal comportamento specifico di un’applicazione e cercare condizioni anomale o percorsi offensivi anche quando non esiste ancora una CVE o una regola dedicata. La piattaforma dispone infatti di informazioni che normalmente vivono in silos differenti: quello che il codice permette di fare, quello che la documentazione dichiara che dovrebbe essere possibile fare e quello che l’applicazione live consente realmente di fare. Quando questi tre livelli non coincidono, o quando la loro combinazione apre una strada inattesa, nasce un’ipotesi che può essere verificata. La scoperta di una zero-day non è quindi trattata come una funzione separata o come una semplice ricerca di “bug sconosciuti”. È una possibile conseguenza di un processo che tenta di comprendere il sistema nel suo insieme, di ragionare sul suo modello di sicurezza e di verificare concretamente le deviazioni più interessanti dal comportamento atteso. Un riconoscimento che nasce dalla ricerca applicata L’ingresso nella Hall of Fame Samsung rappresenta un risultato importante per Ares Cyber AI, ma soprattutto un banco di prova concreto per l’approccio tecnologico sviluppato dal team. La capacità di correlare codice, documentazione, applicazione live e scenari di business mostra come l’AI applicata alla cybersecurity possa andare oltre il semplice triage degli alert e diventare uno strumento operativo per la ricerca e la validazione delle vulnerabilità. In un mercato nel quale l’automazione viene spesso misurata sul numero di controlli eseguiti o di finding prodotti, la sfida più interessante è un’altra: riuscire a comprendere quali debolezze possono trasformarsi in un attacco reale, anche quando quelle debolezze non hanno ancora un nome, una firma o una CVE. Le tre zero-day individuate nei prodotti Samsung rappresentano, in questo senso, un primo risultato pubblico di questa impostazione. I dettagli tecnici arriveranno quando sarà possibile divulgarli; il riconoscimento, invece, è già visibile. Riferimento: Samsung Hall of Fame L'articolo Ares Cyber AI scopre tre vulnerabilità zero-day critiche nei prodotti Samsung ed entra nella Hall of Fame proviene da Red Hot Cyber .

Google announced patches for the exploited privilege escalation vulnerability (CVE-2026-58704) on September 15. The post Pixel Modem Zero-Day Exploited in Targeted Attacks appeared first on SecurityWeek .

Week in review: Linux rootkit deployed on F5 BIG-IP APM devices, Cisco FMC bugs exploited Here’s an overview of some of last week’s most interesting news, articles, interviews and videos: Zero trust AI agents demand a different kind of security In this interview, Chris Webber, VP, Product Marketing at Teleport, explains why zero trust principles need to change for AI agents. He covers how agents act fast, unpredictably, and continuously, and why old ideas like least privilege and point-in-time verification fall short. AI-Infra-Guard: Open-source security scanner for AI systems Tencent’s Zhuque Lab built AI-Infra-Guard, an open-source security scanner for AI systems. It fingerprints running services such as Ollama, vLLM and ComfyUI and checks them against more than 1,600 known CVEs, inspects MCP servers and agent skills across 14 categories of risk, and runs jailbreak evaluations against a target model. Kevin Mandia joins the Amazon board with 30-plus years in cybersecurity Amazon elected Kevin Mandia to its Board of Directors on September 8. Mandia founded Mandiant and served as its CEO before Google acquired the firm in September 2022, and he has worked against cyber threats in the public and private sectors for more than 30 years. Getting a stranger’s phone kicked off the cellular network costs a few dollars Researchers at Michigan State University and three partner schools bought a Samsung Galaxy Z Fold 7, copied the identification number printed on the sealed box, and reported the phone to its carrier as lost. Then they opened the box and set the phone up the way a launch-day buyer would. It would not connect. The phone was new, unopened, and sitting on a lab bench the entire time. The team found six weaknesses in the system carriers use to shut off lost and stolen phones, spanning the devices themselves, the carrier systems that take the reports, and the machinery carriers use to share block lists. Trezor customers hit with phishing calls and letters after shipping-partner breach Roughly 67,000 more customers of SatoshiLabs, the maker of hardware crypto-wallet Trezor, are at heightened risk of phishing attacks after their names, email addresses, phone numbers, and shipping addresses were exposed. “Zero-click” WeChat worm could hijack accounts and spread via a single call Researchers with security company Calif have discovered, weaponized, and privately reported to Tencent a critical vulnerability that allowed them to create “WeWorm”, a worm that spreads via WeChat calls without any user interaction. September 2026 Patch Tuesday: Record patch count, 2 zero-days, and a SigRed successor September 2026 Patch Tuesday is here, with Microsoft delivering another record-breaking number of patches, including those for two vulnerabilities that have been exploited as zero-days. Cisco FMC bugs exploited by nation-state and ransomware actors (CVE-2026-20079, CVE-2026-20316) State-sponsored and financially-motivated attackers are actively exploiting CVE-2026-20079, a critical authentication bypass vulnerability in Cisco Secure Firewall Management Center (FMC), which is used for centrally managing multiple Cisco Secure Firewall devices across a network. What breach and attack simulation needs to become in the AI era Breach and attack simulation (BAS) has always had a supply chain. Somebody has to read the threat report, pull out the techniques, and turn them into something that will actually run against your controls. That somebody has always been a human red team. Up until a few months ago, turning a major new threat into working simulation content within 24 hours counted as very fast. Gartner: 70% of SOCs will pilot AI agents. Only 15% will see results Gartner predicts that by 2028, 70% of large SOCs will pilot AI agents, but only 15% will see measurable gains without structured evaluation. The technology has already moved from Gartner’s Innovation Trigger to the Peak of Inflated Expectations. Prophet Security reports that 40% of security teams use AI daily, 56% are evaluating or piloting it, and just 4% have no plans to adopt. Ransomware negotiation tactics have turned into a business process In this Help Net Security video, Dave Ross, Senior Director of the Intelligence Fusion Team at Intel 471, explains what happens behind the scenes during ransomware negotiations. Building a ransomware decision tree before the call comes in In this Help Net Security video, Kerri Shafer-Page, VP of Incident Response at Arctic Wolf, walks through the ransomware decision tree in this video. She covers four areas where decisions need settling in advance, starting with containment. 18 ways to check whether data can be trusted for AI ETSI has published TR 104 180, a technical report that defines 18 metrics for measuring data quality, giving companies a way to check whether their data is good enough for AI before they use it. The report defines each metric and includes the formulas needed to calculate it. Attackers use rogue ScreenConnect clients to spread malware A file transfer flaw in ScreenConnect Remote Access Support and Access sessions affects both Cloud and On-Premise deployments, ConnectWise confirmed. Until a fix is available, ConnectWise recommends that partners disable file transfers for technicians. N-able patches critical N-central zero-day exploited in the wild (CVE-2026-86218) N-able released an emergency hotfix for CVE-2026-86218, a remote code execution (RCE) flaw affecting N-central, its remote monitoring and management (RMM) solution popular with managed service providers (MSPs). In its release notes, N-able described CVE-2026-86218 as a “critical-CVSS-rated vulnerability that could allow for pre-authenticated remote code execution on the N-central server.” Hackers exploit RouterOS flaws to hijack MikroTik devices without authentication Attackers are exploiting a chain of RouterOS vulnerabilities to hijack MikroTik devices with SSH open to the internet, CERT Polska found. They have discovered six vulnerabilities in RouterOS and coordinated their disclosure with MikroTik. Among the six, two combined let an attacker take full control of a device without authentication, provided the device has SSH accessible from the internet. They named this exploit chain MikroTrick. Mathspace breach exposes data on over a million students and parents Mathspace has confirmed that attackers broke into its internal reporting system through an unpatched Metabase vulnerability and stole data belonging to more than a million students, parents, and school staff. The Sydney-based maths education company wrote in a blog post that the vulnerability, in its self-hosted installation of Metabase, allowed attackers to obtain administrator access to the system without a legitimate login. IT help-desk vishing tricks executives into handing over Microsoft 365 access IT help-desk vishing calls, stolen session tokens, and sign-ins routed through residential proxies are behind a wave of data theft and extortion against Microsoft 365 and other SaaS accounts, according to Arctic Wolf. Threat actors are giving AI agents a bigger role in cyberattacks AI agents are automating parts of cyberattacks with less human involvement, including vulnerability scanning, credential harvesting, and troubleshooting, according to Google Threat Intelligence Group’s Q3 2026 AI Threat Tracker. Google fixes yet another actively exploited Chrome zero-day (CVE-2026-87491) Google has fixed 230 vulnerabilities in Chrome, including a zero-day flaw, CVE-2026-87491, with an in-the-wild exploit. The fix has been shipped in Chrome 153.0.8010.36 and .37 for Windows and macOS and Chrome 153.0.8010.36 for Linux. Hackers deploy Linux rootkit on F5 BIG-IP APM devices, hiding web shell in memory A rootkit found on hacked F5 BIG-IP APM devices skips the usual step of writing a web shell to disk, hiding it in memory instead, according to Sophos. F5 BIG-IP APM provides access policy enforcement to secure access to apps, APIs, and data. It’s primarily used by enterprises, financial institutions, government agencies, and public sector organizations. $245 million in stolen crypto funded racketeering crew’s lavish lifestyle A 22-year-old man built his fortune by breaking into strangers’ digital wallets, then spent it on nightclub tabs, private jets, and a fleet of cars worth millions. Malone Lam, a Singapore citizen, pleaded guilty this week in a Washington D.C. federal court to running a racketeering conspiracy that stole and laundered more than $245 million in cryptocurrency. OpenSSL’s new alpha build speeds up post-quantum crypto The OpenSSL project released the first alpha of OpenSSL 4.1.0, giving developers an early look at a version built for encrypted communication over unreliable connections and faster post-quantum cryptography. This marks the opening test build for a version still months from general availability. Cybercriminals are building phishing pages that exist only inside victims’ browsers A phishing campaign routes victims through genuine Microsoft OAuth and Teams infrastructure before showing them a fake login page built entirely inside their own browser, according to researchers at Barracuda. Fake GTA 6 download delivers malware-packed bundle to impatient gamers Grand Theft Auto VI (GTA 6) is still three months from release, but cybercriminals are not waiting for the launch date. Security firm Huntress found malware disguised as a leaked copy of the game, aimed at fans hoping to get their hands on it early. Attackers call employees’ personal phones to break into Microsoft 365 accounts Attackers are calling or texting employees on their personal phones, posing as internal IT staff, in a social engineering campaign that tricks them into handing over access to corporate cloud accounts. Once inside, they pull files and email from Microsoft 365 apps, SharePoint, OneDrive, and inboxes, for weeks at a time, according to Microsoft Security Research. IDScan confirms breach after 153 million driver’s licenses leak on dark web Days after reports linked IDScan to a dark web database holding more than 153 million driver’s license scans, the identity verification company has confirmed hackers accessed customer data stored on its cloud platform. AI agents exploited PaperCut flaws to breach 395 organizations A threat actor built a working exploit for PaperCut print management software, then handed the job of breaking into hundreds of organizations to AI agents that did most of the work on their own, according to GreyNoise. ToolHive: The open-source way to run any MCP server securely ToolHive is an open-source platform that runs Model Context Protocol servers inside containers. An MCP server is the connector that lets an AI client like Cursor or Claude Code reach an outside tool, and Stacklok ships ToolHive under Apache 2.0, so the runtime, the Kubernetes operator, and the registry cost nothing to self-host. OpenAI just hit a milestone on the road to self-improving AI OpenAI has announced that it has reached a goal set last fall of having an automated research intern by September 2026. The milestone means a system can carry out well-defined research tasks under human direction, including work that would take a skilled researcher several days. The company is also working toward creating an automated AI researcher by March 2028. Microsoft’s Project Zenith puts large AI models directly on developer PCs Microsoft’s Project Zenith is a ready-to-code Windows 11 experience for developer-class PCs capable of running AI models with more than 30 billion parameters locally without relying on metered cloud tokens. Designed for systems with at least 64 GB of unified memory and 250 GB/s or more of memory bandwidth, it pairs powerful hardware with a preconfigured software environment for coding, testing, and experimentation. Jellyfin 12.0 security fixes arrive alongside the removal of legacy client logins Jellyfin shipped version 12.0 of its media server. Several of the security fixes in it block requests built to reach files outside the folders the server is supposed to hand out. The rest of the security work touches first-run setup, plugin installs, parental controls, and the web interface. BleachBit 6.0.4 fixes secure wiping that skipped clusters on Windows The open source cleaner BleachBit reached version 6.0.4 this week, erasing caches, browser traces, and files on Windows, Linux, and now macOS. If you shredded a sensitive file on Windows with an earlier build, parts of it may still sit on the disk where the wipe missed. Fragmentation is the ordinary case, since Windows scatters a file across noncontiguous clusters whenever it cannot find one open run large enough to hold it. AWS spent years rebuilding its routing control plane without taking the network down Every AWS API call, CloudFront video stream, and Route 53 lookup crosses the same infrastructure, which AWS calls its border network. It now runs on a routing system rebuilt from scratch over several years. Chinese AI firms are siphoning capabilities from American models, CISA warns China-based AI companies are using large-scale knowledge distillation campaigns to copy capabilities from leading U.S. AI models, according to a joint cybersecurity advisory from the CISA, NSA, and FBI. A new open standard locks AI weights to approved hardware OPAQUE, a confidential computing company that runs AI workloads inside hardware-isolated environments so operators cannot inspect them, released an open standard that lets AI model builders decide when and where their weights can be decrypted once those weights leave the builder’s own servers. The standard, called Weight Custody Manifest, ships as a developer-preview specification, a Python SDK, and a public test suite covering 91 cases. AI adoption brings new security headaches for already stretched CISOs CISOs are taking on AI governance without a matching increase in resources or expertise, adding to an already broad remit spanning data protection, identity, resilience and compliance, according to Proofpoint’s 2026 Voice of the CISO report. Apple is building photo verification for the people who need it most Apple has introduced Apple Reference Image, an opt-in feature designed to verify the authenticity of photos taken with iPhone 18 Pro models. The company will also add support for the SynthID standard in a software update later this year, helping identify images generated or edited using AI. WordPress adds automated security checks to block risky plugin releases WordPress’ automated security review will now assess every plugin release before it is distributed through the WordPress.org update API. Releases considered a potential security risk will be blocked automatically. Your passkeys can now move between password managers on Android Google turned on a transfer feature in Android that moves passwords and passkeys straight from one password manager to another, with no file to download along the way. You start it from inside the app you are switching to, and Google says the data moves between the apps in a few seconds. Ubuntu 24.04.5 LTS release patches security bugs across ten flavors Canonical shipped Ubuntu 24.04.5 LTS, bundling security updates and fixes for high-severity bugs into new installation media for the “Noble Numbat” release. Anyone installing fresh now gets those corrections baked in from the start, cutting the batch of updates that would normally follow setup. AI is changing what Salesforce security needs to govern Existing security and governance practices have largely focused on identities, permissions, access, configurations and controls. WithSecure’s Navigating Trust in the Modern Salesforce Ecosystem paper says Salesforce environments also require organizations to understand what information they rely on, how trust extends across connected systems, what actions are performed and what outcomes those actions produce. Companies may be measuring phishing resilience the wrong way Companies that judge phishing simulation programs by how often employees click simulated attack emails may be overlooking more important indicators of cyber resilience, according to Pistachio’s Phishing Behaviour Report 2026. Product showcase: Doppler secures secrets for humans, pipelines, and AI agents AI agents have expanded the secrets management challenge. Coding agents, automated workflows, and MCP servers create more identities that need credentials, and more places those credentials can leak. Doppler centralizes credentials for engineers, pipelines, and AI agents in one easy-to-use control plane. It gives security teams a critical secrets management tool developers use, available in the cloud or on-prem. Product showcase: GitGuardian Honeytoken catches credential theft as it happens GitGuardian turns credential harvesting into an immediate, high-confidence detection signal. It deploys honeytokens across developer fleets so that when an infostealer scans a machine and validates a decoy, defenders know within seconds. Cybersecurity jobs available right now: September 8, 2026 We’ve scoured the market to bring you a selection of roles that span various skill levels within the cybersecurity field. Check out this weekly selection of cybersecurity jobs available right now. New infosec products of the week: September 11, 2026 Here’s a look at the most interesting products from the past week, featuring releases from Akeyless, Orchid Security, Scytale, and Securin.

SECURITY Watch out: Apple timepiece can grab snippets of conversation without both speakers' consentWar is peace. Freedom is slavery. Privacy is surveillance SYSTEMS d-Matrix drinks the Nvidia Kool-Aid with NVLink Fusion and MGX rack designsAI infrastructure startup joins Qualcomm, Arm, Marvell, Amazon, Fujitsu, and MediaTek as NVLink true believers ai and ml Anthropic reveals fourth likely crime committed by its AIClaude's Felony Bench rap sheet is now as long as OpenAI's SYSTEMS Samsung to help fortify OpenAI's semiconductor supply chainSemiconductor supply chains are hard, but Samsung offers OpenAI relief in many forms spanning compute and memory AI+ML Google DeepMind rises above the AI scrum with genome atlasSee, AI can be used for good ... or at the very least, a useful distraction from the bad Security Russians are posing as Signal support to launch phishing attacksPLUS: US takes down Iranian propaganda sites; Marketing company asks 'Why Do We Have Your Information?' And more! Security Microsoft patches failed to fix on-prem SharePoint, which is now under zero-day attackPLUS: China upgrades smartphone surveillance tools; Ring eases anti-snooping stance; and more Black Hat and DEF CON DEF CON Franklin project enlists hackers to harden critical infrastructureVoting village reports have been so successful, says Jeff Moss, that the whole of DEF CON will now be included Security EQT buys majority share in Swiss cybersecurity biz AcronisWent at equivalent of $3.5B+ valuation for entire firm, though portion sold not specified Malware Month Ten years since the first corp ransomware, Mikko Hyppönen sees no end in sightOn the plus side, infosec's a good bet for a long, stable career Shopify extends lifeline to Tailwind as vibe coding erodes web dev platform's bottom lineAcquisition gives open source CSS framework 'a stable long-term home' Switzerland tests a FOSS escape route from Microsoft 365Swiss Army sticks a knife in American cloud apps with its own FOSS push Feel peak Windows was 7? You might like Kumander LinuxDebian and Xfce – solid, sensible choices – with a pretty skin Canonical shuttering some of its legacy chat channelsThe Ubuntu Pastebin went in June, IRC gets demoted next Audacity audio-editing app no longer looks like it's from the early 2000sThe FOSS tool for audio editing has a fresh coat of paint, and new features to boot Haiku OS rises / Beta 6 sails open web / Virtual winds fly fastA real alternative to running some kind of FOSS Unix clone

Getting a stranger’s phone kicked off the cellular network costs a few dollars Researchers at Michigan State University and three partner schools bought a Samsung Galaxy Z Fold 7, copied the identification number printed on the sealed box, and reported the phone to its carrier as lost. Then they opened the box and set the phone up the way a launch-day buyer would. It would not connect. The phone was new, unopened, and sitting on a lab bench the entire time. The team found six weaknesses in the system carriers use to shut off lost and stolen phones, spanning the devices themselves, the carrier systems that take the reports, and the machinery carriers use to share block lists. They ran the tests on three major U.S. carriers and their resellers. Blocking a device they did not own cost between $2.50 and $4 and took between about 20 and 80 seconds. The target does not have to be a phone. It can be the alarm panel in a house. End-to-end lost/stolen device reporting ecosystem across device, carrier, and cross-carrier domains, with vulnerabilities shown at their exposed points (Source: Research paper) What gets blocked and how Every device with a cellular radio carries a 15-digit serial number called an IMEI, burned in at the factory and separate from the SIM card, the phone number and the account. When you report a phone stolen, the carrier drops that number into a database called an Equipment Identity Register. After that, the network refuses to register the device. It does not matter who is holding it or which SIM is inside. That is the correct behavior for a stolen phone. What the researchers went looking for is what stops someone from typing in a number that is not theirs. Guan-Hua Tu, an associate professor in Michigan State’s College of Engineering, said an attacker can use the weaknesses to cut a device off remotely, “even though the device has not actually been lost, stolen or sold.” The identity checks are thin The three carriers accept reports only from people with active service, which is supposed to make reporters traceable. The team opened accounts on all three carriers to see how much traceability that buys. None of the three verified a Social Security number or a government ID when setting up a prepaid account, and prepaid customers could file lost-device reports the same as anyone else. Prepaid service can be paid for with an anonymous Visa gift card bought off a rack at Walmart. The second check is ownership, and carriers approximate it by looking at whether the device has been active on the account. The check does not establish who owns a device. It establishes that the device has spent some time on the network, which is a different thing, and a spoofed device can produce that history on demand. On one carrier, one second of attachment was enough. The other two wanted a minute. After that minimum, none of the three asked for any calls or any data at all before letting the researchers report a phone they had never used. The third gap is what counts as a reportable device. The team submitted reports for a smartwatch, for cellular development boards used in industrial equipment, and for a phone that physically cannot operate on the carrier’s frequencies. All of them went through, on all three carriers, from an ordinary smartphone plan with no IoT subscription attached. Once reported, the devices could not get back on. Pulling this off requires knowing the target’s IMEI, which is where the two demonstration attacks come in. The alarm panel goes quiet and the homeowner is not told Home security gateways usually run on Wi-Fi and keep a cellular link as backup, and to keep that backup cheap they use low-power cellular chips built for connected devices. The researchers tested four such boards alongside four phones, feeding each one a deliberately corrupted request for its identity. The phones ignored it, as the standard requires. Two of the boards answered with their IMEIs. Both were vulnerable because of the chipset inside, and those chipsets come from two vendors whose combined global share of that market is above 40 percent. The attack built on that runs in two parts, and the parts can be months apart. First, knock the gateway off Wi-Fi so it falls back to cellular, stand up a small rogue base station to lure it in, and ask for its IMEI. In the lab this took 17 seconds, after which the base station gets switched off and leaves nothing for the carrier to detect. Later, disrupt the Wi-Fi again and file the lost-device report. The gateway drops to cellular the way it is designed to, about 15 seconds after the Wi-Fi cuts out, and finds the network will not take it. Alarms stop reaching the homeowner and the monitoring center. The researchers bought a gateway from one of the two largest U.S. alarm providers, paid for its monitoring service, and confirmed all of this end to end. The other large provider uses the same vulnerable chipsets. Together the two hold more than 41 percent of the U.S. market. The same chips also sit inside water and electricity meters, industrial sensors and cardiac monitors, though the team did not test those devices. Blocking phones that have not been sold yet The second attack goes after new flagship handsets. IMEIs are assigned before a phone ships, and companies that verify devices for the supply chain can buy access to them. The team paid $600 for a full database from one such company, plus $30 a month for updates, and found the Z Fold 7’s number listed before the phone’s July 25, 2025 release. A single prepaid account reported ten devices without tripping anything. Per device, the block cost $2.50, $3.50 or $4 depending on the carrier. Blocking 100 Z Fold 7s would take ten prepaid accounts, $250 to $400, and somewhere between 33 minutes and a little over two hours, against $200,000 worth of phones. The team wrote a script to drive the carriers’ web forms and did not bother optimizing it. Victims get no notice that any of this has happened. Service simply stops. Getting it back means proving to the carrier that you own the device, usually with a receipt showing the IMEI plus identity verification, and if the fraudulent report came in through a different carrier the request can escalate to the GSMA, the industry body that runs the global block list, which adds more delay. One more finding cuts the other way, and it is not reassuring either. A phone reported lost on one carrier still worked fine on the other two a week later, which means at least two of the three are not syncing with the global list they tell customers about. The proposed fixes The team put forward four changes and built a working prototype of them. Device certification tests should check that a phone or module keeps its IMEI to itself when a request arrives unauthenticated, which is the gap that let those two boards leak. Reporting portals should run reporters through government-ID verification, using the same third-party services that federal agencies already use. Ownership checks should combine several signals, including whether the device matches the reporter’s plan, and send anything doubtful to an in-store visit. And the records carriers pass to each other should carry the security details they currently omit, such as whether the reporter was ever identity-checked, so a receiving carrier can decide whether to trust a block. The researchers notified the affected carriers, chipset vendors and manufacturers. The GSMA acknowledged the findings and passed them to its device security group. Tu has said the aim is to raise the cost of an attack while leaving the process alone for people who really have lost a phone. Download report: How security controls perform in practice

SECURITY Watch out: Apple timepiece can grab snippets of conversation without both speakers' consentWar is peace. Freedom is slavery. Privacy is surveillance SYSTEMS d-Matrix drinks the Nvidia Kool-Aid with NVLink Fusion and MGX rack designsAI infrastructure startup joins Qualcomm, Arm, Marvell, Amazon, Fujitsu, and MediaTek as NVLink true believers ai and ml Anthropic reveals fourth likely crime committed by its AIClaude's Felony Bench rap sheet is now as long as OpenAI's SYSTEMS Samsung to help fortify OpenAI's semiconductor supply chainSemiconductor supply chains are hard, but Samsung offers OpenAI relief in many forms spanning compute and memory AI+ML Google DeepMind rises above the AI scrum with genome atlasSee, AI can be used for good ... or at the very least, a useful distraction from the bad Security Russians are posing as Signal support to launch phishing attacksPLUS: US takes down Iranian propaganda sites; Marketing company asks 'Why Do We Have Your Information?' And more! Security Microsoft patches failed to fix on-prem SharePoint, which is now under zero-day attackPLUS: China upgrades smartphone surveillance tools; Ring eases anti-snooping stance; and more Black Hat and DEF CON DEF CON Franklin project enlists hackers to harden critical infrastructureVoting village reports have been so successful, says Jeff Moss, that the whole of DEF CON will now be included Security EQT buys majority share in Swiss cybersecurity biz AcronisWent at equivalent of $3.5B+ valuation for entire firm, though portion sold not specified Malware Month Ten years since the first corp ransomware, Mikko Hyppönen sees no end in sightOn the plus side, infosec's a good bet for a long, stable career Shopify extends lifeline to Tailwind as vibe coding erodes web dev platform's bottom lineAcquisition gives open source CSS framework 'a stable long-term home' Switzerland tests a FOSS escape route from Microsoft 365Swiss Army sticks a knife in American cloud apps with its own FOSS push Feel peak Windows was 7? You might like Kumander LinuxDebian and Xfce – solid, sensible choices – with a pretty skin Canonical shuttering some of its legacy chat channelsThe Ubuntu Pastebin went in June, IRC gets demoted next Audacity audio-editing app no longer looks like it's from the early 2000sThe FOSS tool for audio editing has a fresh coat of paint, and new features to boot Haiku OS rises / Beta 6 sails open web / Virtual winds fly fastA real alternative to running some kind of FOSS Unix clone

SECURITY Watch out: Apple timepiece can grab snippets of conversation without both speakers' consentWar is peace. Freedom is slavery. Privacy is surveillance SYSTEMS d-Matrix drinks the Nvidia Kool-Aid with NVLink Fusion and MGX rack designsAI infrastructure startup joins Qualcomm, Arm, Marvell, Amazon, Fujitsu, and MediaTek as NVLink true believers ai and ml Anthropic reveals fourth likely crime committed by its AIClaude's Felony Bench rap sheet is now as long as OpenAI's SYSTEMS Samsung to help fortify OpenAI's semiconductor supply chainSemiconductor supply chains are hard, but Samsung offers OpenAI relief in many forms spanning compute and memory AI+ML Google DeepMind rises above the AI scrum with genome atlasSee, AI can be used for good ... or at the very least, a useful distraction from the bad Security Russians are posing as Signal support to launch phishing attacksPLUS: US takes down Iranian propaganda sites; Marketing company asks 'Why Do We Have Your Information?' And more! Security Microsoft patches failed to fix on-prem SharePoint, which is now under zero-day attackPLUS: China upgrades smartphone surveillance tools; Ring eases anti-snooping stance; and more Black Hat and DEF CON DEF CON Franklin project enlists hackers to harden critical infrastructureVoting village reports have been so successful, says Jeff Moss, that the whole of DEF CON will now be included Security EQT buys majority share in Swiss cybersecurity biz AcronisWent at equivalent of $3.5B+ valuation for entire firm, though portion sold not specified Malware Month Ten years since the first corp ransomware, Mikko Hyppönen sees no end in sightOn the plus side, infosec's a good bet for a long, stable career Shopify extends lifeline to Tailwind as vibe coding erodes web dev platform's bottom lineAcquisition gives open source CSS framework 'a stable long-term home' Switzerland tests a FOSS escape route from Microsoft 365Swiss Army sticks a knife in American cloud apps with its own FOSS push Feel peak Windows was 7? You might like Kumander LinuxDebian and Xfce – solid, sensible choices – with a pretty skin Canonical shuttering some of its legacy chat channelsThe Ubuntu Pastebin went in June, IRC gets demoted next Audacity audio-editing app no longer looks like it's from the early 2000sThe FOSS tool for audio editing has a fresh coat of paint, and new features to boot Haiku OS rises / Beta 6 sails open web / Virtual winds fly fastA real alternative to running some kind of FOSS Unix clone

SYSTEMS d-Matrix drinks the Nvidia Kool-Aid with NVLink Fusion and MGX rack designsAI infrastructure startup joins Qualcomm, Arm, Marvell, Amazon, Fujitsu, and MediaTek as NVLink true believers ai and ml Anthropic reveals fourth likely crime committed by its AIClaude's Felony Bench rap sheet is now as long as OpenAI's SYSTEMS Samsung to help fortify OpenAI's semiconductor supply chainSemiconductor supply chains are hard, but Samsung offers OpenAI relief in many forms spanning compute and memory AI+ML Google DeepMind rises above the AI scrum with genome atlasSee, AI can be used for good ... or at the very least, a useful distraction from the bad AI and ML Amazon ropes Qualcomm into something, something AI, networking chipsMulti-generation chip collab is more buzzwords than compute Security Russians are posing as Signal support to launch phishing attacksPLUS: US takes down Iranian propaganda sites; Marketing company asks 'Why Do We Have Your Information?' And more! Security Microsoft patches failed to fix on-prem SharePoint, which is now under zero-day attackPLUS: China upgrades smartphone surveillance tools; Ring eases anti-snooping stance; and more Black Hat and DEF CON DEF CON Franklin project enlists hackers to harden critical infrastructureVoting village reports have been so successful, says Jeff Moss, that the whole of DEF CON will now be included Security EQT buys majority share in Swiss cybersecurity biz AcronisWent at equivalent of $3.5B+ valuation for entire firm, though portion sold not specified Malware Month Ten years since the first corp ransomware, Mikko Hyppönen sees no end in sightOn the plus side, infosec's a good bet for a long, stable career Switzerland tests a FOSS escape route from Microsoft 365Swiss Army sticks a knife in American cloud apps with its own FOSS push Feel peak Windows was 7? You might like Kumander LinuxDebian and Xfce – solid, sensible choices – with a pretty skin Canonical shuttering some of its legacy chat channelsThe Ubuntu Pastebin went in June, IRC gets demoted next Audacity audio-editing app no longer looks like it's from the early 2000sThe FOSS tool for audio editing has a fresh coat of paint, and new features to boot Haiku OS rises / Beta 6 sails open web / Virtual winds fly fastA real alternative to running some kind of FOSS Unix clone Offshoots of cancelled TrueNAS Core upgrade to FreeBSD 15Exeunt zVault stage right; enter FreeCORE and BSDnas

ai and ml Anthropic reveals fourth likely crime committed by its AIClaude's Felony Bench rap sheet is now as long as OpenAI's SYSTEMS Samsung to help fortify OpenAI's semiconductor supply chainSemiconductor supply chains are hard, but Samsung offers OpenAI relief in many forms spanning compute and memory AI+ML Google DeepMind rises above the AI scrum with genome atlasSee, AI can be used for good ... or at the very least, a useful distraction from the bad AI and ML Amazon ropes Qualcomm into something, something AI, networking chipsMulti-generation chip collab is more buzzwords than compute On-PREM AMD's Threadripper Halo is a local-AI workstation for researchers with deep pocketsAI workstation promises to put up to 576 GB of HBM3e and 16 TB/s of memory bandwidth on your desk Security Russians are posing as Signal support to launch phishing attacksPLUS: US takes down Iranian propaganda sites; Marketing company asks 'Why Do We Have Your Information?' And more! Security Microsoft patches failed to fix on-prem SharePoint, which is now under zero-day attackPLUS: China upgrades smartphone surveillance tools; Ring eases anti-snooping stance; and more Black Hat and DEF CON DEF CON Franklin project enlists hackers to harden critical infrastructureVoting village reports have been so successful, says Jeff Moss, that the whole of DEF CON will now be included Security EQT buys majority share in Swiss cybersecurity biz AcronisWent at equivalent of $3.5B+ valuation for entire firm, though portion sold not specified Malware Month Ten years since the first corp ransomware, Mikko Hyppönen sees no end in sightOn the plus side, infosec's a good bet for a long, stable career Switzerland tests a FOSS escape route from Microsoft 365Swiss Army sticks a knife in American cloud apps with its own FOSS push Feel peak Windows was 7? You might like Kumander LinuxDebian and Xfce – solid, sensible choices – with a pretty skin Canonical shuttering some of its legacy chat channelsThe Ubuntu Pastebin went in June, IRC gets demoted next Audacity audio-editing app no longer looks like it's from the early 2000sThe FOSS tool for audio editing has a fresh coat of paint, and new features to boot Haiku OS rises / Beta 6 sails open web / Virtual winds fly fastA real alternative to running some kind of FOSS Unix clone Offshoots of cancelled TrueNAS Core upgrade to FreeBSD 15Exeunt zVault stage right; enter FreeCORE and BSDnas

Smart TVs are internet-connected computers with microphones, app stores, advertising systems, and access to the same home networks used by your family’s phones, laptops, printers, and smart-home devices. In the past, we reported on Samsung settling a lawsuit with the Texas Attorney General over how its smart TVs collect and monetize viewing data using  Automated Content Recognition (ACR) .  ACR technology samples what appears on or is heard through a TV, creates a digital fingerprint, and compares that fingerprint against a reference database. It can be used to identify programs, ads, and viewing habits. Now, a new investigation by Gamers Nexus, carried out with Level1Techs and independent security researchers, has examined several LG TV models. The team says its found extensive device and network discovery, ACR tracking, and security weaknesses that could increase the consequences if a television were compromised. Some findings concern LG’s intended product behavior, while others rely on vulnerabilities that researchers say are still being disclosed responsibly. But the broader lesson is clear: A smart TV deserves the same privacy and security consideration as any other internet-connected computer. According to Gamers Nexus, packet captures and firmware analysis showed the tested LG TVs identifying devices on the local network, such as phones, PCs, printers, switches, and smart-home hardware. The investigation also says the TVs collected nearby Wi-Fi network names, signal information, and device-related identifiers. This network information could help build a picture of the other devices in a household. Combined with ACR data, advertising IDs, and other information, it could support detailed profiles of what people watch and the devices they use. The researchers also demonstrated how a compromised TV could capture audio through its microphone, including when the TV appeared to be off. They even showed how the TV stored audio when it was unplugged from the internet and retrieved it after the connection was restored. The researchers also reported remote-code-execution vulnerabilities to LG. They have not disclosed full details while the responsible disclosure process is ongoing. A compromised television could be more than a privacy issue. It might provide an attacker with a foothold on a home or business network, access to audio, or a route to probe other devices. How to stay safe The concerns are not limited to one brand. Smart TVs sit at the intersection of entertainment, advertising, and the home network. Treating them as security-sensitive devices—and demanding clear, meaningful privacy choices—is increasingly part of staying safe at home. There is no need to panic, but owners can take a few practical steps to limit what their TV collects and what it can access: Install firmware updates promptly, especially security updates. Check your model’s support page and the TV’s software-update settings. Review the privacy controls under Settings, Privacy & Terms, or User Agreements. Turn off ACR, viewing-information collection, personalized ads, voice recognition, and other features you don’t need. Don’t accept every agreement by default. Read each consent screen and decline optional advertising and voice-data features where possible. Use a separate IoT or guest network for televisions, cameras, speakers, and other smart-home devices. This limits what a compromised device can reach on your main network. Disable UPnP on your router unless it is genuinely needed and avoid exposing TV services directly to the internet. Our earlier guide to disabling ACR includes instructions for several popular TV brands. Browse like no one’s watching.   Malwarebytes Privacy VPN encrypts your connection and never logs what you do, so the next story you read doesn’t have to feel personal.  Try it free →  

Una nuova ondata di compromissioni sta interessando utenti iPhone in Italia. Account WhatsApp apparentemente integri vengono utilizzati per contattare amici e conoscenti e chiedere denaro, mentre la sezione “Dispositivi collegati” può non mostrare alcuna anomalia. In alcuni casi, la compromissione può inoltre verificarsi nonostante sia attiva la verifica in due passaggi , un ulteriore elemento che distingue questo scenario dalle tradizionali tecniche di acquisizione degli account. Le analisi forensi italiane, confrontate con la ricerca internazionale sulle vulnerabilità CVE-2025-55177 e CVE-2025-43300 , permettono di ricostruire uno scenario di attacco zero-click particolarmente sofisticato. La caratteristica più inquietante è che, nei casi analizzati, la vittima non sembra aver compiuto nessuna delle azioni normalmente associate alla compromissione di un account WhatsApp. Nessun QR code scansionato, nessun codice di verifica comunicato, nessun dispositivo sconosciuto apparentemente associato all’account. Eppure, a un certo punto, dal proprio account iniziano a partire messaggi verso persone conosciute, spesso accompagnati da una richiesta di denaro. È da qui che bisogna partire per comprendere cosa sta realmente accadendo. Il fenomeno osservato in Italia Le analisi forensi condotte in Italia hanno evidenziato una serie di casi con caratteristiche ricorrenti. Le prime segnalazioni risalgono alla primavera del 2026, seguite da ulteriori episodi nei mesi successivi e da una nuova, significativa ondata di compromissioni registrata negli ultimi giorni . Tra gli elementi osservati compaiono iPhone dalla generazione 8 fino a iPhone 14, con dispositivi sui quali risultavano installate versioni di iOS 16. Gli attaccanti scrivono ai contatti della vittima direttamente dal suo numero WhatsApp, spesso chiedendo di effettuare un bonifico. Un ulteriore elemento ricorrente è che l’attività sembra concentrarsi sulle conversazioni recenti: l’attaccante appare quindi in grado di utilizzare almeno una parte delle informazioni necessarie per individuare i destinatari più plausibili. Ma c’è un particolare ancora più significativo: nessun dispositivo estraneo risulta visibile nella sezione “Dispositivi collegati” . Le vittime, inoltre, non riferiscono di aver effettuato operazioni di pairing, fornito codici o inquadrato QR code. Questo esclude, nei casi analizzati, il classico schema del cosiddetto ghost pairing e porta verso un modello di compromissione zero-click. Il resync e la sessione WhatsApp L’analisi dei dispositivi ha evidenziato anche anomalie nei meccanismi di sincronizzazione, con ripetuti eventi di resync , cioè tentativi di riallineare la sessione e i relativi dati tra il dispositivo della vittima e i servizi di WhatsApp. Il fenomeno è particolarmente interessante perché aiuta a comprendere come l’account possa rimanere apparentemente funzionante sul dispositivo della vittima mentre viene utilizzato contemporaneamente dall’attaccante. In questo scenario, quindi, non è necessariamente presente il classico “nuovo dispositivo” che l’utente può individuare dall’interfaccia di WhatsApp. Un comportamento analogo è stato successivamente descritto anche in un advisory del Jamaica Cyber Incident Response Team, che ha documentato compromissioni zero-click di account WhatsApp su dispositivi iOS 16, con attività non autorizzata, richieste di denaro, anomalie di sincronizzazione e assenza di dispositivi sospetti nella sezione “Linked Devices”. Il parallelismo è significativo, anche se non permette di affermare che tutti gli episodi siano riconducibili allo stesso attore. Il doppio fattore non è sufficiente Un ulteriore elemento di rilievo è il rapporto con la verifica in due passaggi di WhatsApp. La protezione aggiunge normalmente un ulteriore livello di sicurezza contro la presa di controllo dell’account, ma la catena descritta in questi attacchi non segue il normale percorso di autenticazione previsto per l’accesso all’account. Il punto non è quindi il furto o l’indovinamento del PIN della verifica in due passaggi. La catena di exploit opera a un livello differente, sfruttando la vulnerabilità nella gestione della sincronizzazione per ottenere l’esecuzione delle operazioni necessarie alla compromissione senza passare dal normale processo di autenticazione dell’utente. CVE-2025-55177: la vulnerabilità di WhatsApp Per comprendere la possibile origine tecnica dell’attacco bisogna partire da CVE-2025-55177 . Il National Vulnerability Database di NIST descrive la vulnerabilità come una incomplete authorization of linked device synchronization messages in WhatsApp per iOS. In determinate condizioni, un soggetto non autorizzato poteva provocare sul dispositivo bersaglio l’elaborazione di contenuti provenienti da un URL arbitrario. Per WhatsApp per iOS risultano vulnerabili le versioni dalla 2.22.25.2 fino alle precedenti alla 2.25.21.73 ; per WhatsApp Business per iOS la correzione arriva con la versione 2.25.21.78. Il CSIRT ha classificato la vulnerabilità come ad alta gravità e ha spiegato che il problema riguarda la gestione dei messaggi di sincronizzazione tra dispositivi associati: nelle versioni vulnerabili il controllo delle fasi di autorizzazione risultava incompleto e poteva permettere l’inserimento di contenuto malevolo nel flusso di sincronizzazione. La stessa WhatsApp ha inoltre dichiarato che la vulnerabilità, combinata con CVE-2025-43300 , potrebbe essere stata sfruttata in un attacco sofisticato contro utenti specifici. CVE-2025-43300: la vulnerabilità di ImageIO Il secondo elemento della catena è CVE-2025-43300 , una vulnerabilità del componente ImageIO di Apple. Si tratta di una out-of-bounds write : durante l’elaborazione di un’immagine appositamente costruita era possibile provocare una corruzione della memoria. Apple ha dichiarato di essere a conoscenza di una segnalazione secondo cui la vulnerabilità poteva essere stata sfruttata in un attacco estremamente sofisticato contro individui specifici. La vulnerabilità è stata corretta in iOS 16.7.12 per i dispositivi che supportano quel ramo del sistema operativo. NIST indica inoltre come interessate versioni precedenti alla 15.8.5, 16.7.12 e 18.6.2, a seconda del ramo di iOS utilizzato. È quindi importante non ridurre il problema a una generica “vulnerabilità degli iPhone con iOS 16”: la condizione di esposizione dipende dalla versione effettivamente installata. Come viene concatenata la vulnerabilità È la combinazione delle due CVE a rendere particolarmente interessante il caso. In forma semplificata, la catena può essere rappresentata così: WhatsApp → CVE-2025-55177 → contenuto remoto → immagine appositamente costruita → ImageIO → CVE-2025-43300 La prima vulnerabilità permette di raggiungere il dispositivo attraverso il meccanismo di sincronizzazione di WhatsApp. La seconda riguarda invece l’elaborazione del contenuto grafico da parte del sistema operativo. Il risultato è una catena nella quale il contenuto può essere processato senza che l’utente debba necessariamente aprirlo volontariamente. È questo il punto centrale dello zero-click . La ricostruzione del 39C3 La possibilità di concatenare le due vulnerabilità è stata approfondita anche dalla ricerca internazionale. Al 39C3 – Chaos Communication Congress , Zhongrui Li, Yizhe Zhuang e Kira Chen hanno presentato DNGerousLINK: A Deep Dive into WhatsApp 0-Click Exploits on iOS and Samsung Devices . La ricerca analizza la catena CVE-2025-55177 / CVE-2025-43300 e il ruolo dei file DNG nel raggiungimento del componente di elaborazione delle immagini. I ricercatori hanno approfondito il comportamento del framework RawCamera durante il parsing dei file DNG e hanno ricostruito la condizione di out-of-bounds associata a CVE-2025-43300 . La ricerca ha inoltre portato alla realizzazione di una Proof of Concept, dimostrando la possibilità di riprodurre la catena su dispositivi Apple. Questo passaggio è importante perché permette di comprendere tecnicamente come una vulnerabilità applicativa e una vulnerabilità del sistema operativo possano essere utilizzate come anelli consecutivi della stessa catena. La conferma di CISA C’è poi un elemento istituzionale che rafforza ulteriormente il quadro. CVE-2025-55177 è stata inserita da CISA nel Known Exploited Vulnerabilities Catalog. NIST riporta inoltre la valutazione CISA secondo cui lo sfruttamento della vulnerabilità è classificato come active , con impatto tecnico total . Questo significa che esistono evidenze istituzionali di sfruttamento reale della vulnerabilità. Non significa però che ogni compromissione WhatsApp osservata oggi debba essere automaticamente attribuita a questa CVE. La presenza di una vulnerabilità sfruttata non costituisce, da sola, prova dell’origine di uno specifico incidente. Il quadro internazionale Il caso italiano trova un riscontro interessante anche all’estero. Il Jamaica Cyber Incident Response Team ha descritto un attacco di tipo Zero-Click WhatsApp Account Takeover contro dispositivi iOS 16, nel quale la vittima non deve cliccare link, scansionare QR code, aprire manualmente file o approvare richieste di accesso. Tra gli effetti indicati figurano messaggi non autorizzati inviati dall’account compromesso, comprese richieste di trasferimenti finanziari, anomalie di sincronizzazione e assenza di dispositivi sospetti nella sezione dei dispositivi collegati. L’organismo collega inoltre lo scenario alla combinazione tra CVE-2025-55177 e CVE-2025-43300 . Il confronto con i casi italiani è quindi particolarmente significativo. La ricerca internazionale ha inoltre evidenziato un secondo filone relativo ai dispositivi Samsung e alla vulnerabilità CVE-2025-21043 , anch’essa legata all’elaborazione di immagini. Il caso Samsung è tecnicamente distinto da quello Apple, ma conferma l’interesse degli attaccanti verso i parser multimediali come superficie d’attacco per tecniche zero-click. iPhone 14 e versioni di iOS La presenza di iPhone 14 nei casi italiani merita una precisazione. Non è corretto affermare che “la vulnerabilità riguarda gli iPhone fino al 14”. CVE-2025-55177 riguarda specifiche versioni di WhatsApp per iOS, mentre CVE-2025-43300 riguarda specifiche versioni dei sistemi operativi Apple. Il modello del dispositivo, quindi, non è sufficiente per stabilire l’esposizione. La verifica deve considerare: modello → versione iOS → versione WhatsApp → patch installate. Questo è particolarmente importante per i dispositivi più vecchi, che possono rimanere fermi a versioni del sistema operativo non più aggiornabili. Disattivare il download automatico protegge? La disattivazione del download automatico di foto, video e documenti può essere una precauzione utile, ma non costituisce una difesa specifica contro uno zero-click. Il semplice download del file e la sua elaborazione da parte di un componente software sono infatti operazioni differenti. Se il contenuto viene processato automaticamente, impedire il salvataggio del file non garantisce che il parser vulnerabile non venga raggiunto. La misura fondamentale rimane quindi l’aggiornamento di iOS e WhatsApp alle versioni disponibili e supportate. Quando la compromissione diventa una truffa La parte tecnica dell’attacco è sofisticata. L’utilizzo successivo dell’account, invece, è estremamente semplice. L’attaccante può rivolgersi alle persone che conoscono la vittima e utilizzare conversazioni reali per formulare una richiesta di denaro. Il destinatario non vede un numero sconosciuto. Vede il nome di una persona di cui si fida. Non è quindi necessario costruire una falsa identità: l’attaccante può utilizzare quella reale . È in questo passaggio che l’exploit diventa social engineering e può trasformarsi in una frode. Una richiesta di bonifico, di prestito urgente o di pagamento può risultare credibile proprio perché arriva dall’account autentico della persona conosciuta. Il rischio di sospensione o revoca dell’account Esiste anche un ulteriore effetto da considerare. Un account compromesso che viene utilizzato per inviare numerosi messaggi o richieste di denaro può generare un comportamento riconducibile ai sistemi automatici di contrasto allo spam. La vittima potrebbe quindi subire una sospensione o una revoca dell’account a causa di attività che non ha effettuato volontariamente. I due eventi devono però essere distinti: la compromissione è l’incidente di sicurezza; la sospensione o revoca è una conseguenza dell’attività rilevata dalla piattaforma. La seconda non dimostra quindi, da sola, quale tecnica sia stata utilizzata per ottenere l’accesso. Una campagna ancora attiva? Le informazioni disponibili indicano che il fenomeno continua a essere osservato e presenta caratteristiche riscontrate anche al di fuori dell’Italia. CISA continua a classificare CVE-2025-55177 come vulnerabilità sfruttata e il quadro internazionale mostra casi con caratteristiche analoghe. È quindi corretto considerare la minaccia concreta. È invece necessario mantenere prudenza nell’attribuzione: le informazioni pubbliche non consentono di stabilire che ogni caso recente sia stato realizzato dallo stesso attore, attraverso la stessa infrastruttura o con la medesima implementazione dell’exploit. Una minaccia che non richiede l’errore dell’utente Il caso mostra un’evoluzione importante rispetto alle campagne tradizionali. Per anni la principale raccomandazione è stata quella di non cliccare link sospetti, non aprire allegati e non comunicare codici di verifica. Queste precauzioni rimangono fondamentali, ma non possono proteggere da una catena zero-click nella quale l’utente non costituisce il punto debole. La superficie d’attacco può trovarsi nei componenti che lavorano automaticamente: sincronizzazione → contenuto remoto → parser → libreria multimediale → sistema operativo. È proprio questo l’aspetto più difficile da percepire per una vittima: il dispositivo può essere attaccato mentre, dal suo punto di vista, non sta facendo nulla di anomalo. Conclusioni La nuova ondata di compromissioni WhatsApp osservata in Italia rappresenta un caso significativo dell’evoluzione degli attacchi contro i dispositivi mobili. CVE-2025-55177 interessa la gestione dei messaggi di sincronizzazione dei dispositivi collegati di WhatsApp, mentre CVE-2025-43300 riguarda il componente ImageIO di Apple. WhatsApp ha confermato che la prima vulnerabilità, in combinazione con una vulnerabilità del sistema operativo Apple, potrebbe essere stata sfruttata in attacchi sofisticati contro utenti specifici. La presenza della CVE nel catalogo CISA delle vulnerabilità sfruttate e la successiva ricostruzione tecnica internazionale rendono il quadro particolarmente significativo. Ma l’aspetto più preoccupante emerge quando la componente tecnica lascia spazio a quella criminale. Una catena di exploit complessa può terminare con un messaggio semplicissimo, inviato dal vero account della vittima a una persona che si fida di lei: “Mi fai un bonifico?” Fonti primarie NIST NVD – CVE-2025-55177 https://nvd.nist.gov/vuln/detail/ CVE-2025-55177 NIST NVD – CVE-2025-43300 https://nvd.nist.gov/vuln/detail/ CVE-2025-43300 WhatsApp – Security Advisories https://www.whatsapp.com/security/advisories/2025/ Apple – iOS 16.7.12 Security Content https://support.apple.com/it-it/125141 CSIRT Italia – Vulnerabilità in WhatsApp per dispositivi Apple https://csirt.regione.toscana.it/vulnerabilita-in-whatsapp-per-dispositivi-apple-al01-250830-csirt-ita/ Analisi tecniche e fonti internazionali Forenser – Analisi forense dei casi WhatsApp su iPhone con iOS 16 https://www.forenser.it/account-whatsapp-compromessi-su-iphone-con-ios-16/ 39C3 – DNGerousLINK: A Deep Dive into WhatsApp 0-Click Exploits on iOS and Samsung Devices https://events.ccc.de/congress/2025/hub/en/event/detail/dngerouslink-a-deep-dive-into-whatsapp-0-click-exploits-on-ios-and-samsung-devices Jamaica Cyber Incident Response Team – Zero-Click WhatsApp Account Takeover Attack Targeting iOS 16 Devices https://www.cirt.gov.jm/advisory/zero-click-whatsapp-account-takeover-attack-targeting-ios-16-devices L'articolo WhatsApp sotto attacco su iPhone: la nuova ondata zero-click in Italia e la ricostruzione internazionale della minaccia proviene da Red Hot Cyber .

The ToxicPanda Android malware has evolved with new malicious functionality, expanding its targeting to 349 applications and adding support for 167 remote commands. The malware now requests VPN service permissions to create a local interface that allows it to control network traffic passing through it. The feature enables ToxicPanda 2.0 to block communication from Google Play and Google Play Services. Control at the network level permits the malware to interfere with various security checks and actions, such as app verifications, updates, Play Protect communication, or legitimate disruptions designed to protect users. After obtaining VPN service permissions, ToxicPanda 2.0 blocks communications to Google Play before extracting and installing its payload, then requests Accessibility Service permissions. Mobile security company Zimperium says that ToxicPanda 2.0 is being distributed through Amazon AWS-hosted buckets. Analysis of the malware revealed that it now includes functions to automate the Android Wireless Debugging Bridge (ADB), enabling shell-level access to infected devices. The latest version of the malware supports 167 remote commands and phishing overlays for 349 banking, financial, cryptocurrency, and e-wallet applications targeting 16 countries. It also includes a separate PIN-harvesting module that targets 140 financial and cryptocurrency apps and can dynamically update the target list. According to the researchers, the app overlays are invisible to the victim, allowing the malware to capture touch inputs on targeted apps. ToxicPanda also spoofs the Android lock screen to capture device PINs, unlocking patterns, and passwords. Some analyzed malware samples also used fake system update screens to hide ongoing malicious activity. One command, ‘autoBoot,’ identifies the host device manufacturer and launches the corresponding OEM-specific auto-start or power management settings to maintain persistence. Zimperium reports that this bypasses battery consumption protections that kill background processes on Xiaomi, OPPO, Vivo, Samsung, and Huawei devices. Abusing ADB One feature that stands out in the analyzed recent Toxic Panda version is its automatic abuse of the Android Debug Bridge (ADB) to gain shell access. ADB is the command-line tool for executing shell commands on Android devices. Wireless ADB, introduced in Android 11, provides this access over Wi-Fi without a USB connection. Using the Accessibility Services permission, the malware enables Developer Options, activates Wireless Debugging, extracts the six-digit ADB pairing code and port, and connects with the device’s local ADB service. “Once the malware gains shell user permissions, it starts executing high-privilege commands directly through the ADB daemon, the malware bypasses standard Android runtime consent prompts to grant itself broad permissions, neutralize OS background restrictions, silently enable critical components, and enforce persistence,” Zimperium explains. Wireless ADB abuse is a growing trend among Android malware, as other Android malware authors have implemented it in their malicious tools. Recently, Group-IB reported a similar mechanism implemented in the latest version of the RedHook malware. Zimperium has published a list of indicators of compromise (IoCs) associated with the latest ToxicPanda version in this GitHub repository. 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

It was discovered that some AMD processors did not properly clear data in the floating point divider unit during speculative execution. A local attacker could use this to expose sensitive information. (CVE-2025-54505) It was discovered that some AMD Zen 2 processors did not properly isolate shared resources in the operation cache. A local attacker could possibly use this issue to corrupt instructions executed at a higher privilege level, resulting in privilege escalation. (CVE-2025-54518) It was discovered that some AMD Zen 5 processors supporting RDSEED instruction did not properly handle entropy, potentially resulting in the consumption of insufficiently random values. A local attacker could possibly use this issue to influence the values returned by the RDSEED instruction causing loss of confidentiality and integrity. (CVE-2025-62626) Several security issues were discovered in the Linux kernel. An attacker could possibly use these to compromise the system. This update corrects flaws in the following subsystems: - ARM64 architecture; - RISC-V architecture; - S390 architecture; - x86 architecture; - Block layer subsystem; - Cryptographic API; - Compute Acceleration Framework; - ACPI drivers; - Serial ATA and Parallel ATA drivers; - Drivers core; - Power management core; - DRBD Distributed Replicated Block Device drivers; - Rados block device (RBD) driver; - Compressed RAM block device driver; - Bluetooth drivers; - Bus devices; - Character device driver; - Clock framework and drivers; - Data acquisition framework and drivers; - Counter interface drivers; - CPU frequency scaling framework; - Hardware crypto device drivers; - CXL (Compute Express Link) drivers; - DMA engine subsystem; - EDAC drivers; - EFI core; - GPU drivers; - Greybus drivers; - HID subsystem; - Hardware monitoring drivers; - I2C subsystem; - IIO ADC drivers; - IIO subsystem; - InfiniBand drivers; - Input Device (Miscellaneous) drivers; - IRQ chip drivers; - LED subsystem; - Mailbox framework; - Multiple devices driver; - Media drivers; - MediaTek SMI driver; - NVIDIA Tegra memory controller driver; - Fastrpc Driver; - IBM Advanced System Management driver; - MMC subsystem; - MTD block device drivers; - Network drivers; - Ethernet bonding driver; - Mellanox network drivers; - Microsoft Azure Network Adapter (MANA) driver; - STMicroelectronics network drivers; - Ethernet team driver; - MediaTek network drivers; - Near Field Communication (NFC) drivers; - NTB driver; - NVDIMM (Non-Volatile Memory Device) drivers; - NVME drivers; - Device tree and open firmware driver; - PCI subsystem; - Pin controllers subsystem; - x86 platform drivers; - Broadcom BCM2835 power domain driver; - Generic PM domains; - i.MX PM domains; - Remote Processor subsystem; - S/390 drivers; - SCSI subsystem; - SLIMbus drivers; - Freescale SoC drivers; - Microchip PolarFire SoC system controller driver; - SPI subsystem; - Media staging drivers; - Realtek RTL8723BS SDIO drivers; - SM750 framebuffer staging driver; - TCM subsystem; - Thermal drivers; - TTY drivers; - UFS subsystem; - Cadence USB3 driver; - USB Device Class drivers; - ULPI bus; - USB core drivers; - DesignWare USB2 driver; - USB Gadget drivers; - USB Host Controller drivers; - Mustek MDC800 USB digital camera driver; - USB YUREX driver; - Renesas USBHS Controller drivers; - USB Type-C Connector System Software Interface driver; - VFIO drivers; - Framebuffer layer; - TSM TDX Guest driver; - Xen hypervisor drivers; - File systems infrastructure; - BTRFS file system; - Ceph distributed file system; - EROFS file system; - Ext4 file system; - F2FS file system; - FUSE (File system in Userspace); - GFS2 file system; - HFS+ file system; - Journaling layer for block devices (JBD2); - Network file systems library; - Network file system (NFS) server daemon; - NILFS2 file system; - File system notification infrastructure; - NTFS3 file system; - OCFS2 file system; - Diskquota system; - SMB network file system; - SquashFS file system; - Tracing file system; - UDF file system; - XFS file system; - Kernel CPU control infrastructure; - QorIQ DPAA2 FSL-MC bus driver; - Memory Management; - Integrity Measurement Architecture(IMA) framework; - KVM subsystem; - Memory management; - Networking core; - padata parallel execution mechanism; - PPP protocol drivers and compressors; - Linux Security Modules (LSM) Framework; - Tracing infrastructure; - Network traffic control; - Distributed Switch Architecture; - IPv4 networking; - IP tunnels definitions; - MAC80211 subsystem; - Netfilter; - User-space API (UAPI); - io_uring subsystem; - Audit subsystem; - BPF subsystem; - Control group (cgroup); - Perf events; - Kernel exit() syscall; - Kernel fork() syscall; - Kernel futex primitives; - KProbes tracing; - Locking primitives; - Kernel module support; - Padata parallel execution mechanism; - Cryptographic library; - Heterogeneous memory management; - KASAN memory debugging framework; - Asynchronous Transfer Mode (ATM) subsystem; - B.A.T.M.A.N. meshing protocol; - Bluetooth subsystem; - Ethernet bridge; - CAIF protocol; - CAN network layer; - Ceph Core library; - IPv6 networking; - XFRM subsystem; - L2TP protocol; - Management Component Transport Protocol (MCTP); - Multipath TCP; - NCSI (Network Controller Sideband Interface) driver; - NFC subsystem; - Open vSwitch; - Packet sockets; - Qualcomm IPC Router (QRTR); - RDS protocol; - RF switch subsystem; - Rose network layer; - RxRPC session sockets; - SCTP protocol; - SMC sockets; - Stream parser; - Sun RPC protocol; - TIPC protocol; - TLS protocol; - Unix domain sockets; - VMware vSockets driver; - Wireless networking; - X.25 network layer; - eXpress Data Path; - Landlock security; - ALSA framework; - Generic PCM loopback sound driver; - FireWire sound drivers; - HD-audio driver; - Creative Sound Blaster X-Fi driver; - AMD SoC Alsa drivers; - QCOM ASoC drivers; - Renesas ASoC drivers; - Samsung ASoC drivers; - SoC audio core drivers; - SOF drivers; - STI ASoC drivers; - USB sound devices; - Objtool; (CVE-2025-21709, CVE-2025-22116, CVE-2025-38426, CVE-2025-39764, CVE-2025-40135, CVE-2025-40150, CVE-2025-68175, CVE-2025-68239, CVE-2025-68334, CVE-2025-68736, CVE-2025-71152, CVE-2025-71161, CVE-2025-71203, CVE-2025-71221, CVE-2025-71269, CVE-2025-71287, CVE-2025-71288, CVE-2026-22981, CVE-2026-22985, CVE-2026-22993, CVE-2026-23004, CVE-2026-23066, CVE-2026-23104, CVE-2026-23118, CVE-2026-23138, CVE-2026-23154, CVE-2026-23157, CVE-2026-23171, CVE-2026-23207, CVE-2026-23226, CVE-2026-23227, CVE-2026-23244, CVE-2026-23245, CVE-2026-23246, CVE-2026-23253, CVE-2026-23255, CVE-2026-23270, CVE-2026-23271, CVE-2026-23276, CVE-2026-23277, CVE-2026-23279, CVE-2026-23281, CVE-2026-23284, CVE-2026-23285, CVE-2026-23286, CVE-2026-23287, CVE-2026-23289, CVE-2026-23290, CVE-2026-23291, CVE-2026-23292, CVE-2026-23293, CVE-2026-23296, CVE-2026-23298, CVE-2026-23300, CVE-2026-23302, CVE-2026-23303, CVE-2026-23304, CVE-2026-23306, CVE-2026-23307, CVE-2026-23308, CVE-2026-23310, CVE-2026-23312, CVE-2026-23313, CVE-2026-23315, CVE-2026-23317, CVE-2026-23318, CVE-2026-23319, CVE-2026-23321, CVE-2026-23324, CVE-2026-23325, CVE-2026-23330, CVE-2026-23334, CVE-2026-23335, CVE-2026-23336, CVE-2026-23339, CVE-2026-23340, CVE-2026-23343, CVE-2026-23347, CVE-2026-23352, CVE-2026-23356, CVE-2026-23357, CVE-2026-23359, CVE-2026-23360, CVE-2026-23361, CVE-2026-23362, CVE-2026-23363, CVE-2026-23364, CVE-2026-23365, CVE-2026-23367, CVE-2026-23368, CVE-2026-23369, CVE-2026-23370, CVE-2026-23372, CVE-2026-23374, CVE-2026-23375, CVE-2026-23378, CVE-2026-23379, CVE-2026-23381, CVE-2026-23382, CVE-2026-23383, CVE-2026-23386, CVE-2026-23387, CVE-2026-23388, CVE-2026-23389, CVE-2026-23391, CVE-2026-23395, CVE-2026-23396, CVE-2026-23397, CVE-2026-23398, CVE-2026-23399, CVE-2026-23401, CVE-2026-23412, CVE-2026-23413, CVE-2026-23414, CVE-2026-23418, CVE-2026-23419, CVE-2026-23420, CVE-2026-23426, CVE-2026-23434, CVE-2026-23438, CVE-2026-23439, CVE-2026-23440, CVE-2026-23441, CVE-2026-23442, CVE-2026-23444, CVE-2026-23446, CVE-2026-23447, CVE-2026-23448, CVE-2026-23449, CVE-2026-23452, CVE-2026-23454, CVE-2026-23456, CVE-2026-23457, CVE-2026-23458, CVE-2026-23460, CVE-2026-23461, CVE-2026-23462, CVE-2026-23463, CVE-2026-23464, CVE-2026-23465, CVE-2026-23468, CVE-2026-23470, CVE-2026-23474, CVE-2026-23475, CVE-2026-31389, CVE-2026-31391, CVE-2026-31392, CVE-2026-31393, CVE-2026-31394, CVE-2026-31396, CVE-2026-31399, CVE-2026-31400, CVE-2026-31403, CVE-2026-31405, CVE-2026-31407, CVE-2026-31408, CVE-2026-31409, CVE-2026-31412, CVE-2026-31413, CVE-2026-31414, CVE-2026-31415, CVE-2026-31416, CVE-2026-31417, CVE-2026-31421, CVE-2026-31422, CVE-2026-31423, CVE-2026-31424, CVE-2026-31425, CVE-2026-31426, CVE-2026-31427, CVE-2026-31428, CVE-2026-31429, CVE-2026-31430, CVE-2026-31432, CVE-2026-31433, CVE-2026-31434, CVE-2026-31438, CVE-2026-31439, CVE-2026-31440, CVE-2026-31441, CVE-2026-31446, CVE-2026-31447, CVE-2026-31449, CVE-2026-31450, CVE-2026-31451, CVE-2026-31452, CVE-2026-31453, CVE-2026-31454, CVE-2026-31455, CVE-2026-31458, CVE-2026-31464, CVE-2026-31466, CVE-2026-31467, CVE-2026-31469, CVE-2026-31470, CVE-2026-31473, CVE-2026-31474, CVE-2026-31476, CVE-2026-31477, CVE-2026-31480, CVE-2026-31482, CVE-2026-31483, CVE-2026-31485, CVE-2026-31487, CVE-2026-31488, CVE-2026-31489, CVE-2026-31492, CVE-2026-31494, CVE-2026-31495, CVE-2026-31496, CVE-2026-31497, CVE-2026-31498, CVE-2026-31499, CVE-2026-31500, CVE-2026-31502, CVE-2026-31503, CVE-2026-31505, CVE-2026-31507, CVE-2026-31508, CVE-2026-31509, CVE-2026-31510, CVE-2026-31511, CVE-2026-31512, CVE-2026-31515, CVE-2026-31516, CVE-2026-31518, CVE-2026-31519, CVE-2026-31520, CVE-2026-31521, CVE-2026-31522, CVE-2026-31523, CVE-2026-31524, CVE-2026-31525, CVE-2026-31527, CVE-2026-31528, CVE-2026-31530, CVE-2026-31532, CVE-2026-31540, CVE-2026-31542, CVE-2026-31545, CVE-2026-31546, CVE-2026-31548, CVE-2026-31549, CVE-2026-31550, CVE-2026-31551, CVE-2026-31552, CVE-2026-31554, CVE-2026-31555, CVE-2026-31556, CVE-2026-31557, CVE-2026-31563, CVE-2026-31565, CVE-2026-31566, CVE-2026-31570, CVE-2026-31575, CVE-2026-31576, CVE-2026-31577, CVE-2026-31578, CVE-2026-31580, CVE-2026-31581, CVE-2026-31582, CVE-2026-31583, CVE-2026-31584, CVE-2026-31585, CVE-2026-31586, CVE-2026-31587, CVE-2026-31588, CVE-2026-31590, CVE-2026-31594, CVE-2026-31595, CVE-2026-31596, CVE-2026-31597, CVE-2026-31598, CVE-2026-31599, CVE-2026-31602, CVE-2026-31603, CVE-2026-31604, CVE-2026-31605, CVE-2026-31606, CVE-2026-31610, CVE-2026-31611, CVE-2026-31612, CVE-2026-31613, CVE-2026-31615, CVE-2026-31616, CVE-2026-31617, CVE-2026-31618, CVE-2026-31619, CVE-2026-31622, CVE-2026-31623, CVE-2026-31624, CVE-2026-31625, CVE-2026-31626, CVE-2026-31627, CVE-2026-31628, CVE-2026-31629, CVE-2026-31634, CVE-2026-31638, CVE-2026-31639, CVE-2026-31642, CVE-2026-31645, CVE-2026-31646, CVE-2026-31648, CVE-2026-31651, CVE-2026-31655, CVE-2026-31656, CVE-2026-31658, CVE-2026-31660, CVE-2026-31661, CVE-2026-31662, CVE-2026-31664, CVE-2026-31665, CVE-2026-31667, CVE-2026-31670, CVE-2026-31671, CVE-2026-31672, CVE-2026-31673, CVE-2026-31674, CVE-2026-31675, CVE-2026-31677, CVE-2026-31678, CVE-2026-31679, CVE-2026-31680, CVE-2026-31681, CVE-2026-31683, CVE-2026-31684, CVE-2026-31686, CVE-2026-31689, CVE-2026-31694, CVE-2026-31695, CVE-2026-31696, CVE-2026-31697, CVE-2026-31698, CVE-2026-31699, CVE-2026-31700, CVE-2026-31701, CVE-2026-31702, CVE-2026-31704, CVE-2026-31705, CVE-2026-31706, CVE-2026-31707, CVE-2026-31708, CVE-2026-31709, CVE-2026-31711, CVE-2026-31712, CVE-2026-31714, CVE-2026-31715, CVE-2026-31716, CVE-2026-31720, CVE-2026-31721, CVE-2026-31722, CVE-2026-31723, CVE-2026-31724, CVE-2026-31725, CVE-2026-31726, CVE-2026-31728, CVE-2026-31729, CVE-2026-31730, CVE-2026-31731, CVE-2026-31737, CVE-2026-31738, CVE-2026-31740, CVE-2026-31741, CVE-2026-31747, CVE-2026-31748, CVE-2026-31749, CVE-2026-31751, CVE-2026-31752, CVE-2026-31754, CVE-2026-31755, CVE-2026-31756, CVE-2026-31758, CVE-2026-31759, CVE-2026-31761, CVE-2026-31762, CVE-2026-31763, CVE-2026-31767, CVE-2026-31768, CVE-2026-31770, CVE-2026-31772, CVE-2026-31773, CVE-2026-31778, CVE-2026-31779, CVE-2026-31780, CVE-2026-31781, CVE-2026-31788, CVE-2026-43007, CVE-2026-43012, CVE-2026-43013, CVE-2026-43014, CVE-2026-43015, CVE-2026-43016, CVE-2026-43017, CVE-2026-43018, CVE-2026-43019, CVE-2026-43020, CVE-2026-43023, CVE-2026-43024, CVE-2026-43025, CVE-2026-43026, CVE-2026-43027, CVE-2026-43028, CVE-2026-43030, CVE-2026-43032, CVE-2026-43035, CVE-2026-43036, CVE-2026-43040, CVE-2026-43041, CVE-2026-43043, CVE-2026-43044, CVE-2026-43046, CVE-2026-43047, CVE-2026-43049, CVE-2026-43050, CVE-2026-43051, CVE-2026-43052, CVE-2026-43054, CVE-2026-43056, CVE-2026-43057, CVE-2026-43058, CVE-2026-43059, CVE-2026-43060, CVE-2026-43061, CVE-2026-43062, CVE-2026-43064, CVE-2026-43065, CVE-2026-43066, CVE-2026-43068, CVE-2026-43069, CVE-2026-43072, CVE-2026-43073, CVE-2026-43074, CVE-2026-43075, CVE-2026-43076, CVE-2026-43079, CVE-2026-43080, CVE-2026-43081, CVE-2026-43082, CVE-2026-43084, CVE-2026-43085, CVE-2026-43086, CVE-2026-43088, CVE-2026-43089, CVE-2026-43091, CVE-2026-43092, CVE-2026-43093, CVE-2026-43094, CVE-2026-43098, CVE-2026-43099, CVE-2026-43103, CVE-2026-43104, CVE-2026-43105, CVE-2026-43107, CVE-2026-43109, CVE-2026-43110, CVE-2026-43111, CVE-2026-43112, CVE-2026-43113, CVE-2026-43119, CVE-2026-43120, CVE-2026-43129, CVE-2026-43162, CVE-2026-43245, CVE-2026-43252, CVE-2026-43265, CVE-2026-43281, CVE-2026-43324, CVE-2026-43327, CVE-2026-43328, CVE-2026-43329, CVE-2026-43330, CVE-2026-43332, CVE-2026-43333, CVE-2026-43334, CVE-2026-43336, CVE-2026-43338, CVE-2026-43339, CVE-2026-43340, CVE-2026-43342, CVE-2026-43343, CVE-2026-43345, CVE-2026-43350, CVE-2026-43355, CVE-2026-43357, CVE-2026-43359, CVE-2026-43360, CVE-2026-43361, CVE-2026-43362, CVE-2026-43363, CVE-2026-43365, CVE-2026-43366, CVE-2026-43368, CVE-2026-43370, CVE-2026-43371, CVE-2026-43372, CVE-2026-43373, CVE-2026-43377, CVE-2026-43380, CVE-2026-43381, CVE-2026-43382, CVE-2026-43386, CVE-2026-43387, CVE-2026-43395, CVE-2026-43397, CVE-2026-43405, CVE-2026-43408, CVE-2026-43409, CVE-2026-43411, CVE-2026-43412, CVE-2026-43413, CVE-2026-43415, CVE-2026-43419, CVE-2026-43420, CVE-2026-43421, CVE-2026-43424, CVE-2026-43425, CVE-2026-43426, CVE-2026-43427, CVE-2026-43428, CVE-2026-43429, CVE-2026-43430, CVE-2026-43432, CVE-2026-43436, CVE-2026-43437, CVE-2026-43439, CVE-2026-43441, CVE-2026-43445, CVE-2026-43448, CVE-2026-43449, CVE-2026-43450, CVE-2026-43451, CVE-2026-43452, CVE-2026-43453, CVE-2026-43455, CVE-2026-43456, CVE-2026-43457, CVE-2026-43458, CVE-2026-43459, CVE-2026-43466, CVE-2026-43467, CVE-2026-43468, CVE-2026-43469, CVE-2026-43471, CVE-2026-43472, CVE-2026-43473, CVE-2026-43475, CVE-2026-43476, CVE-2026-43480, CVE-2026-43483, CVE-2026-43484, CVE-2026-43488, CVE-2026-43490, CVE-2026-43491, CVE-2026-43492, CVE-2026-43495, CVE-2026-43496, CVE-2026-43497, CVE-2026-43499, CVE-2026-43502, CVE-2026-45834, CVE-2026-45835, CVE-2026-45836, CVE-2026-45838, CVE-2026-45839, CVE-2026-45840, CVE-2026-45841, CVE-2026-45842, CVE-2026-45843, CVE-2026-45844, CVE-2026-45845, CVE-2026-45846, CVE-2026-45855, CVE-2026-45858, CVE-2026-45899, CVE-2026-45911, CVE-2026-45920, CVE-2026-45924, CVE-2026-45942, CVE-2026-45943, CVE-2026-45956, CVE-2026-45958, CVE-2026-45985, CVE-2026-45986, CVE-2026-45987, CVE-2026-45989, CVE-2026-45991, CVE-2026-45994, CVE-2026-45996, CVE-2026-45997, CVE-2026-45999, CVE-2026-46002, CVE-2026-46003, CVE-2026-46004, CVE-2026-46005, CVE-2026-46006, CVE-2026-46007, CVE-2026-46009, CVE-2026-46011, CVE-2026-46012, CVE-2026-46015, CVE-2026-46016, CVE-2026-46018, CVE-2026-46019, CVE-2026-46021, CVE-2026-46022, CVE-2026-46023, CVE-2026-46024, CVE-2026-46026, CVE-2026-46027, CVE-2026-46031, CVE-2026-46033, CVE-2026-46037, CVE-2026-46038, CVE-2026-46040, CVE-2026-46041, CVE-2026-46044, CVE-2026-46046, CVE-2026-46047, CVE-2026-46049, CVE-2026-46050, CVE-2026-46051, CVE-2026-46052, CVE-2026-46053, CVE-2026-46056, CVE-2026-46058, CVE-2026-46061, CVE-2026-46062, CVE-2026-46063, CVE-2026-46064, CVE-2026-46065, CVE-2026-46068, CVE-2026-46069, CVE-2026-46070, CVE-2026-46072, CVE-2026-46073, CVE-2026-46075, CVE-2026-46076, CVE-2026-46077, CVE-2026-46078, CVE-2026-46079, CVE-2026-46080, CVE-2026-46082, CVE-2026-46083, CVE-2026-46084, CVE-2026-46086, CVE-2026-46088, CVE-2026-46089, CVE-2026-46090, CVE-2026-46091, CVE-2026-46092, CVE-2026-46094, CVE-2026-46098, CVE-2026-46099, CVE-2026-46101, CVE-2026-46102, CVE-2026-46103, CVE-2026-46106, CVE-2026-46107, CVE-2026-46108, CVE-2026-46110, CVE-2026-46111, CVE-2026-46112, CVE-2026-46113, CVE-2026-46114, CVE-2026-46116, CVE-2026-46117, CVE-2026-46120, CVE-2026-46121, CVE-2026-46122, CVE-2026-46123, CVE-2026-46124, CVE-2026-46125, CVE-2026-46126, CVE-2026-46127, CVE-2026-46128, CVE-2026-46129, CVE-2026-46131, CVE-2026-46132, CVE-2026-46133, CVE-2026-46136, CVE-2026-46137, CVE-2026-46138, CVE-2026-46139, CVE-2026-46142, CVE-2026-46143, CVE-2026-46144, CVE-2026-46145, CVE-2026-46146, CVE-2026-46149, CVE-2026-46150, CVE-2026-46151, CVE-2026-46152, CVE-2026-46157, CVE-2026-46159, CVE-2026-46160, CVE-2026-46161, CVE-2026-46163, CVE-2026-46164, CVE-2026-46167, CVE-2026-46168, CVE-2026-46169, CVE-2026-46172, CVE-2026-46173, CVE-2026-46174, CVE-2026-46176, CVE-2026-46177, CVE-2026-46178, CVE-2026-46179, CVE-2026-46180, CVE-2026-46184, CVE-2026-46186, CVE-2026-46187, CVE-2026-46189, CVE-2026-46190, CVE-2026-46191, CVE-2026-46193, CVE-2026-46196, CVE-2026-46197, CVE-2026-46198, CVE-2026-46199, CVE-2026-46200, CVE-2026-46201, CVE-2026-46204, CVE-2026-46205, CVE-2026-46206, CVE-2026-46207, CVE-2026-46208, CVE-2026-46209, CVE-2026-46211, CVE-2026-46212, CVE-2026-46214, CVE-2026-46218, CVE-2026-46219, CVE-2026-46220, CVE-2026-46225, CVE-2026-46226, CVE-2026-46227, CVE-2026-46229, CVE-2026-46230, CVE-2026-46231, CVE-2026-46232, CVE-2026-46233, CVE-2026-46234, CVE-2026-46235, CVE-2026-46236, CVE-2026-46238, CVE-2026-46241, CVE-2026-46273, CVE-2026-46274, CVE-2026-46280, CVE-2026-46282, CVE-2026-46285, CVE-2026-46286, CVE-2026-46287, CVE-2026-46291, CVE-2026-46292, CVE-2026-46293, CVE-2026-46294, CVE-2026-46296, CVE-2026-46299, CVE-2026-46301, CVE-2026-46303, CVE-2026-46304, CVE-2026-46306, CVE-2026-46307, CVE-2026-46312, CVE-2026-46314, CVE-2026-46319, CVE-2026-52911, CVE-2026-52920, CVE-2026-52925, CVE-2026-52933, CVE-2026-52936, CVE-2026-52951, CVE-2026-52954, CVE-2026-52955, CVE-2026-52957, CVE-2026-52958, CVE-2026-52961, CVE-2026-52962, CVE-2026-52963, CVE-2026-52964, CVE-2026-52967, CVE-2026-52968, CVE-2026-52969, CVE-2026-52970, CVE-2026-52974, CVE-2026-52975, CVE-2026-52977, CVE-2026-52981, CVE-2026-52982, CVE-2026-52984, CVE-2026-52985, CVE-2026-52986, CVE-2026-52989, CVE-2026-52990, CVE-2026-52992, CVE-2026-52993, CVE-2026-52995, CVE-2026-52998, CVE-2026-52999, CVE-2026-53001, CVE-2026-53002, CVE-2026-53003, CVE-2026-53004, CVE-2026-53006, CVE-2026-53011, CVE-2026-53012, CVE-2026-53013, CVE-2026-53014, CVE-2026-53015, CVE-2026-53016, CVE-2026-53021, CVE-2026-53022, CVE-2026-53023, CVE-2026-53032, CVE-2026-53033, CVE-2026-53034, CVE-2026-53035, CVE-2026-53036, CVE-2026-53037, CVE-2026-53039, CVE-2026-53040, CVE-2026-53041, CVE-2026-53043, CVE-2026-53045, CVE-2026-53046, CVE-2026-53047, CVE-2026-53048, CVE-2026-53049, CVE-2026-53050, CVE-2026-53052, CVE-2026-53056, CVE-2026-53058, CVE-2026-53059, CVE-2026-53060, CVE-2026-53061, CVE-2026-53062, CVE-2026-53063, CVE-2026-53064, CVE-2026-53065, CVE-2026-53066, CVE-2026-53068, CVE-2026-53069, CVE-2026-53071, CVE-2026-53072, CVE-2026-53073, CVE-2026-53074, CVE-2026-53075, CVE-2026-53076, CVE-2026-53077, CVE-2026-53082, CVE-2026-53083, CVE-2026-53084, CVE-2026-53085, CVE-2026-53086, CVE-2026-53088, CVE-2026-53093, CVE-2026-53094, CVE-2026-53096, CVE-2026-53097, CVE-2026-53098, CVE-2026-53110, CVE-2026-53111, CVE-2026-53112, CVE-2026-53115, CVE-2026-53117, CVE-2026-53122, CVE-2026-53123, CVE-2026-53126, CVE-2026-53128, CVE-2026-53130, CVE-2026-53279, CVE-2026-53287, CVE-2026-53289, CVE-2026-53291, CVE-2026-53293, CVE-2026-53294, CVE-2026-53295, CVE-2026-53296, CVE-2026-53303, CVE-2026-53304, CVE-2026-53306, CVE-2026-53309, CVE-2026-53314, CVE-2026-53320, CVE-2026-53369, CVE-2026-53374, CVE-2026-53375, CVE-2026-53376, CVE-2026-53379, CVE-2026-63838, CVE-2026-63843, CVE-2026-63844, CVE-2026-63845, CVE-2026-63846, CVE-2026-63847, CVE-2026-63848, CVE-2026-63851, CVE-2026-63852, CVE-2026-63854, CVE-2026-63855, CVE-2026-63856, CVE-2026-63860, CVE-2026-63861, CVE-2026-63862, CVE-2026-63865, CVE-2026-64164, CVE-2026-64587)

AI enablement platform Xpander has announced raising $7.5 million in a seed funding round led by Pico Venture Partners. Emerge Ventures, Samsung Next, and Seedil also invested in the San Francisco, California-based startup. Xpander was founded in 2024 by former AWS engineers David Twizer (CEO), Moriel Pahima (CTO), and Ran Sheinberg (CPO). The startup has built a platform that allows organizations to adopt, build, run, secure, and manage AI agents across their environments. Xpander’s platform relies on a vendor-neutral, universal agent harness that executes agents as portable workloads and securely renders interfaces on demand. This infrastructure provides organizations with governance over AI agents and with the flexibility to build, deploy, and manage them across products, workflows, and data. The startup also provides its customers with an agentic Forward Deployed Engineer called Omni that enables the creation of agent teammates for employees, allows teams to build agents, and supports collaborative multi-agent workflows. Xpander plans to use the seed investment to accelerate market penetration. “Every company is working to harness the power of AI and become AI-native, yet most find it unattainable. Our experience at AWS has inspired us to build a platform that smoothly facilitates AI migration and adoption for organizations, to allow them to grow their businesses faster than ever before,” said Twizer. Related: Cybersecurity M&A Roundup: 21 Deals Announced in July 2026 Related: Venture Firm Team8 Secures Additional $365 Million Related: Mindgard Raises $30 Million to Protect AI Systems Related: Corma Raises $60 Million for Defensive Cybersecurity AI Model

Week in review: Salesforce and ServiceNow portals exposed for 17 months, exploited Metabase 0-day Here’s an overview of some of last week’s most interesting news, articles, interviews and videos: GitHub Dependabot malware alerts now cover eight ecosystems GitHub has flagged npm malware since March 2026. Anyone pulling in a bad PyPI, Maven, RubyGems, NuGet, Go, crates.io, or PHP Composer package has had no such warning, because GitHub’s malware detection only ever watched one ecosystem. That changed this month. Dependabot malware alerts, which had run on npm data alone, now draw on all eight ecosystems the moment a user turns malware alerts on. Dependabot itself already runs across more than 30 million repositories and 34-plus package ecosystems overall, which gives a sense of the scale the malware pipeline now has to operate at. An AI tool found 84 flaws in 5G network software and 23 of them still have no fix Researchers at Nanyang Technological University turned a set of AI agents loose on the software that runs 4G and 5G phone networks, and the agents came back with 84 security flaws nobody had reported before. Developers have confirmed 83 of them, and 81 now carry CVE numbers. Post-quantum migration gets harder when every user holds a key In this Help Net Security interview, Christopher Smith, CEO of Quantus, discusses what cryptographic inventories turn up in banks and hospitals, including default passwords and admin keys still held by former employees. A stranger has been reading Salesforce and ServiceNow portals worldwide for 17 months Most security stories start with something broken. This one starts with everything working as designed. Researchers at Reco have been tracking a campaign they call City-Forum, named after a domain registered in 2002, abandoned, and now resolving to a generic rented server from a German hosting provider. From that server, someone has been pulling records out of Salesforce and ServiceNow portals around the world. N-able ships second N-central hotfix as attackers keep exploiting CVE-2026-18577 To help customers fend off ongoing attacks, N-able released a second security hotfix for N‑central, its monitoring and management (RMM) solution popular with managed service providers (MSPs). Metabase zero-day exploited to access Framework customer data Framework, the San Francisco-based company that designs repairable and upgradeable laptops, has suffered a data breach after attackers managed to exploit a zero-day vulnerability in the Metabase business intelligence service. According to the notification sent to affected Framework customers, the attackers accessed names, email addresses, phone numbers, physical addresses, and login IP addresses, but not payment information or records related to orders. Microsoft patches 400+ vulnerabilities, one zero-day under attack (CVE-2026-68820) Microsoft’s August 2026 Patch Tuesday delivered security fixes for 400+ vulnerabilities, including one that has been exploited in zero-day attacks (CVE-2026-68820) and three that were publicly disclosed prior to the release of the patches. Cisco fixes vulnerability exploited to DoS its firewalls (CVE-2026-20349) A high-severity vulnerability (CVE-2026-20349) is being leveraged by attackers to temporarily interrupt the operation of Cisco firewalls, the company has confirmed. The flaw has been added to CISA’s Known Exploited Vulnerabilities catalog and needs to be remediated by US civilian federal agencies by August 14, 2026. Who will be the Stanislav Petrov in your organization? Recent reports of “rogue AI” systems hacking companies reminded Brian Honan, CEO, BH Consulting, of Stanislav Petrov. In 1983, Soviet computers falsely detected a US nuclear missile launch. Instead of trusting the system, Petrov applied human judgement and correctly identified it as a false alarm, potentially preventing nuclear war. That is why the recent incidents involving OpenAI and Hugging Face, Anthropic, Meta, and the UK’s AI Security Institute deserve the attention of CISOs and boards. 338 million attack simulations reveal the state of enterprise defense First, a bit of good news: Enterprise defenses are recovering. However, it’s a narrow recovery, with a twist. Today, organizations are better at stopping loud attacks but have barely moved the needle at all against the quiet ones. This data, and a lot more, comes straight from the newly published Blue Report 2026, the fourth annual comprehensive study from Picus Labs. Four corporate investigation mistakes organizations make under pressure In this Help Net Security video, Christine Gadsby, VP and Chief Security Advisor at BlackBerry, explains why corporate investigations go wrong before the forensic team arrives. The first hours matter more than leaders assume. Access gets granted, conversations start, and decisions get made that later affect chain of custody, privilege, and how regulators judge the process. How to report an AI Act violation in the EU The EU’s fight to regulate AI models entered a new chapter on 2 August 2026, when the European Commission’s AI Office and national authorities began enforcing the AI Act. It creates a common set of rules for AI systems used or sold in the EU, with the goal of encouraging innovation while protecting people’s safety and fundamental rights. Chainloop: Open-source evidence store and policy engine for the software supply chain Chainloop is an open source evidence store for the software supply chain. A command line tool runs inside a GitHub Actions, GitLab, Jenkins, or Dagger pipeline, picks up what the build produced, uploads those files to content-addressable storage, and references each one in a signed in-toto attestation. in-toto is a specification for recording who ran which step of a build, so the record can be checked afterward. Microsoft Entra ID is removing an extra MFA hurdle for Windows Hello and macOS PSSO users Microsoft is changing how Entra ID handles MFA for people who sign in with Windows Hello for Business (WHfB) or macOS Platform Single Sign-On (PSSO). The rollout reaches worldwide and GCC tenants starting early October 2026, with completion expected by late November. Cyberattack on Steam hardware shipper leaks names, addresses, and order data Video game publisher Valve is alerting customers in Europe to a data breach at CEVA Logistics, its Steam hardware shipping partner. Reports from affected customers began surfacing on social media earlier today, after Valve started sending out data breach notification emails. Previously unseen entry vector used to breach Polish energy plant The December 29 cyberattack on a Polish combined heat and power (CHP) plant was the first observed case of attackers gaining access to an OT network through a private APN, according to CERT Polska. The private APN is a dedicated mobile network that a Distribution System Operator (DSO), the company running the local electricity grid, sets up with a mobile carrier. GPT-5.6-Cyber refuses security researchers’ requests far less often GPT-5.6-Cyber is a new OpenAI model built on GPT-5.6 Sol, trained to find zero-day vulnerabilities and build exploit chains, with fewer refusals on higher-risk, dual-use work. Model is available only through Daybreak Red, the higher tier of OpenAI’s vetted access program for cybersecurity professionals. Ransomware gangs don’t need control system access to disrupt industrial production Disrupting IT systems that support industrial environments can be enough to interrupt production, even when ransomware operators do not gain direct access to industrial control systems (ICS), according to Dragos. The company identified 1,140 ransomware incidents involving industrial organizations in the second quarter of 2026, up 12% from 1,020 in Q1. Malicious SIMs can hijack smartphones, steal files, and lock them onto 2G Researchers have found that compromised or malicious SIM cards can issue commands to some smartphones and cellular-connected devices, allowing attackers to steal information, disrupt communications, downgrade connections to 2G, and in some cases execute code. Ready-made $500 kit puts a crypto scam within anyone’s reach A seller on a cybercrime forum is offering a ready-made scam kit for $500, complete with an admin panel that tracks victims, checks their crypto wallets for value, and inflates fake balances to squeeze out more money, Malwarebytes found. Split-second deepfake glitch blows digital certificate fraudster’s cover Spanish police have arrested a man in Murcia accused of using deepfake software to trick a certificate provider’s video identity checks in an attempt to obtain digital signatures he could use for financial fraud. Lazarus hackers pair fake job offers with Windows zero-day exploit The North Korea-linked Lazarus group is using fake job offers, trojanized PDF software and a Windows zero-day in attacks aimed primarily at the defense sector, Check Point researchers have found. Signal’s new security feature checks if your encrypted chats were tampered with Signal has introduced a feature called automatic key verification, giving users a new way to confirm that nobody has secretly interfered with their encrypted chats. 153GB of stolen credentials surface after LiteLLM supply chain attack A massive 153GB archive stolen during the LiteLLM supply chain attack exposes credentials and other sensitive data linked to thousands of corporate domains, including AWS, Samsung, Cisco, and Salesforce. Hudson Rock says it obtained and analyzed the archive, which contains 433,909 files, and attributed 118,829 CI runner dumps to 2,488 corporate domains. Attackers exploit critical SharePoint flaw after PoC goes public (CVE-2026-55040) Threat actors have begun exploiting a critical Microsoft SharePoint flaw following the release of proof-of-concept (PoC) exploit code by Rapid7. Tracked as CVE-2026-55040, the vulnerability was patched by Microsoft as part of its July 2026 Patch Tuesday updates. White House authorizes private US companies to hack foreign criminal networks President Trump signed a National Security Presidential Memorandum on August 12 allowing vetted private companies to run offensive cyber operations against foreign threat actors, under the control and oversight of the US government. Ukrainian police raid 94 fraudulent call centers, seize $2 million Ukrainian police have disrupted 94 fraudulent call centers during a nationwide operation that involved more than 400 searches and the seizure of thousands of computers, phones, and SIM cards. New Android malware relays bank cards to fraudsters while victims still hold them Group-IB researchers discovered WindRelay, a new Android malware built to capture live payment card data over NFC (Near Field Communication) and relay it to attackers in real time. WindRelay is paired with the SpyNote remote access trojan, which gives attackers remote access to a victim’s device. 71% of CISOs spend 10+ hours on board reports Boards want evidence that security controls and architecture reduce business risk, expressed in terms of resilience, consequence, and decision relevance. Translating technical findings into business language remains a major time burden for CISOs, who are calling for simpler data delivery, better frameworks, and better context. Pulse Security AI’s The CISO-Board Communication Gap report found that board members bring external information into discussions while many organizations still lack a formally defined cyber risk appetite. Product showcase: Enpass Password Manager breaks away from the proprietary cloud model Enpass is a password manager that stores passwords, passkeys, payment cards, identities, secure notes, software licenses, and other sensitive information in encrypted vaults. Vaults remain on the device or in a cloud storage service selected by the user. OpenAI locks down Astra over potential critical cyber capabilities OpenAI’s internal evaluation of its upcoming model, Astra, found significant advances in agentic coding and cybersecurity, leading the company to conclude that it cannot rule out the model reaching the critical capability level for cybersecurity under its Preparedness Framework. Anthropic to put AI in charge of reviewing Claude Code actions by default Anthropic will make auto mode in Claude Code the default for new sessions on Pro, Max, and Team plans starting August 14. Users who previously selected a different default may receive a one-time prompt asking whether they want to switch to auto mode. Your security vendor gets the frontier cyber model, you get the findings Selected red team specialists can now use OpenAI’s cyber models to find and exploit weaknesses in client applications and infrastructure. Those clients never get the models themselves. That split is the design of the Daybreak Cyber Partner Program, which OpenAI expanded on August 10: access to the underlying models stays with the approved partner and is not transferred directly to the customer. Locking your ssh-agent exposed local-only keys until OpenSSH 10.5 Lock your ssh-agent and it should sit there refusing to sign anything until you unlock it. In OpenSSH 10.4, locking it also switched off the check that tells the agent whether a request came from your own machine or arrived down a forwarded connection from a remote server. The fix shipped today in OpenSSH 10.5. AI deployments are stretching enterprise security to its limits CISOs and CTOs expect AI deployments to increase their organizations’ attack surface by an average of 14% over the next year. Nearly all lack visibility into AI deployments, and 90% are concerned about employees using unapproved AI tools outside formal oversight, according to NetFoundry’s 2026 State of Secure AI Access survey. PentestGPT: Open-source automated penetration testing agentic framework PentestGPT is an open-source penetration testing agent that points a large language model at a target and lets it work. In its default mode it runs recon, then exploit, then walkthrough, each stage feeding the next. Switch it to pentest mode and the stages become asset discovery, vulnerability identification, report. No human sits in the loop. Chrome’s anti-abuse protections block 7 billion unwanted Android notifications daily Google Chrome’s latest measures against abusive web push notifications include automatically revoking notification permissions for inactive and suspicious websites, helping reduce scams, phishing attempts, and other deceptive content. Wireshark 4.6.8 patches 28 security bugs, nine in file parsers Wireshark 4.6.8 fixes 28 security bugs in the protocol analyzer, and nine of them fire when someone opens a saved capture file. Those nine sit in file parsers, the code that reads a capture off disk before any dissection begins: pcapng, Endace ERF, Tektronix K12xx, BUSMASTER, Catapult DCT2000, Gammu DCT3, 3gpp phone logs, TTX Logger, and, on Windows only, Ixia IxVeriWave and Vector Informatik BLF. An attacker never has to touch your network for these. They only have to get you the file. Product showcase: Is this image real? Slop or Not investigates Slop or Not is an AI text and image detector for iPhone and Mac that runs entirely offline, with no account required. It uses on-device AI models powered by the Apple Neural Engine to detect AI-generated content. DDoS attacks hit record scale as 1 Tbps+ campaigns become more common DDoS attacks grew in scale during the first half of 2026, bringing larger traffic floods, shorter attack durations, and increasingly automated campaigns. Cloudflare’s H1 2026 DDoS Threat Report shows threat actors relying on multi-vector techniques and large-scale network-layer attacks to disrupt online services across multiple industries. 17 draft Cyber Resilience Act standards are open for comment A company selling a connected toy in Europe must show by the end of 2027 that the product meets the Cyber Resilience Act. The law states what manufacturers have to achieve and stops there, which leaves the toymaker to work out the technical detail alone. Seventeen draft standards, now open for comment, supply that detail. Weak IAM affects up to 98% of cloud environments Misconfiguration remains one of the leading threats to cloud environments because a single configuration error can result in public network access, unrotated keys, missing encryption, exposed services, and logging gaps. CISA now mandates baseline cloud configuration practices for US federal agencies. The hardest part of agentic AI may be rebuilding the business Organizations expect AI agents to change how work gets done, driving productivity and growth while allowing employees to focus on higher-value tasks. Few, however, have the processes and workflows needed to realize those benefits, according to Deloitte’s latest research. AWS Certificate Manager sets 2027 end date for email-validated certificate renewals AWS Certificate Manager (ACM) will phase out email validation for public certificates throughout 2027, ahead of the Certification Authority/Browser (CA/B) Forum’s March 15, 2028 deadline for ending email-based domain validation. OpenAI’s GPT-5.6 Sol runs up to 14× faster with Ultrafast mode OpenAI’s GPT-5.6 Sol on Ultrafast mode is available in limited preview to a select group of customers, launching first through the OpenAI API. The company says the service runs up to 14 times faster than Standard processing and generates up to 750 output tokens per second. Ultrafast is powered by Cerebras as part of the companies’ partnership on ultra-low-latency inference. Cybersecurity jobs available right now: August 11, 2026 We’ve scoured the market to bring you a selection of roles that span various skill levels within the cybersecurity field. Check out this weekly selection of cybersecurity jobs available right now. New infosec products of the week: August 14, 2026 Here’s a look at the most interesting products from the past week, featuring releases from A10 Networks, ScienceLogic, Searchlight Cyber, and SelectHub.

153GB of stolen credentials surface after LiteLLM supply chain attack A massive 153GB archive stolen during the LiteLLM supply chain attack exposes credentials and other sensitive data linked to thousands of corporate domains, including AWS, Samsung, Cisco, and Salesforce. Hudson Rock says it obtained and analyzed the archive, which contains 433,909 files, and attributed 118,829 CI runner dumps to 2,488 corporate domains. “We are leveraging this data for a global ethical disclosure effort,” Alon Gal, Hudson Rock’s co-founder and CTO, told Help Net Security. “We see this as an opportunity to help organizations respond proactively before threat actors weaponize the data publicly.” LiteLLM is an open-source proxy gateway that developers use to route requests to different AI models. The LiteLLM breach had its roots in an earlier compromise of Trivy, a popular open-source vulnerability scanner. On March 19, 2026, TeamPCP, a cybercriminal group that emerged in late 2025, used stolen credentials to publish a compromised version of Trivy. LiteLLM’s build pipeline installed Trivy automatically, giving the poisoned scanner read access to the runner environment and allowing the attackers to steal the project’s PyPI publishing tokens. Using those tokens, TeamPCP published two malicious LiteLLM releases, versions 1.82.7 and 1.82.8, to the Python Package Index on March 24. “The key takeaway is how supply chains have evolved to make a single upstream breach affect thousands of companies simultaneously. A window of roughly 40 minutes in which the LiteLLM dependency was hacked led to over 430,000 instances in which millions of secrets were harvested,” Gal said. Tracing exposed secrets back to their owners The dataset contains information linked to organizations including NVIDIA, Volkswagen, Microsoft, FedEx, S&P Global, John Deere, Epic Games, Orange, TomTom, BT Group, ServiceNow, Deloitte, and Siemens. Screenshots published with the research show AWS secret access keys, Salesforce client secrets, Slack signing secrets, Azure environment variables, and AI provider API keys captured during pipeline execution. A trove of internal corporate secrets (Source: Hudson Rock) CloudSEK, working from a separate dataset built from about 434,000 stolen files, put the number of exposed organizations at close to 2,500. The company stressed that its figures reflect exposure rather than confirmed breaches. “I’ve confirmed the data is legit by the way, multiple victim orgs,” security researcher Kevin Beaumont said. “It contains a significant volume of sensitive content at orgs. It’s a massive supply chain breach due to poor AI security—not because AI is the threat, but teens can run circles around orgs obsessed with rushing out AI and poor DevOps security.” “Identifying the victims within this 153GB database presents a unique threat intelligence challenge. Accurate attribution requires looking past the surface level to analyze the actual infrastructure boundaries,” Hudson Rock noted. In one case, a leaked pipeline was tied to a committer email at SiriusXM, but infrastructure markers in the same dump, including a self-hosted GitLab instance at gitlab.adswizz.com, pointed to AdsWizz, a SiriusXM subsidiary. Hudson Rock said correct identification depends on “hard infrastructure markers rather than simple committer emails.” A large share of the dumped files have no obvious owner. According to Hudson Rock, some contain database passwords, third-party API keys, and cloud credentials without a company email address, custom domain, or internal server name that could identify the organization they came from. This means organizations may have exposed credentials in the dataset without knowing they are affected. Organizations urged to act before stolen data leaks Hudson Rock is urging organizations using AI proxy infrastructure, third-party CI/CD vulnerability scanners, or downstream AI packages to audit their environments for LiteLLM versions 1.82.7 and 1.82.8 and treat any secrets accessible to the LiteLLM environment as compromised. Organizations should also rotate cloud IAM keys and access tokens, review audit logs for anomalous activity dating back to March 24, and check for unauthorized .pth files and suspicious systemd services. “While we cannot disclose how we obtained this data, it is not leaked anywhere at the moment and is not circulating widely. This makes it a critical window of opportunity for companies to rotate keys and secrets before it eventually leaks, as is the natural cycle of a breach,” Gal added. Despite the scale of the breach, some organizations appear to be treating the exposure with less urgency than the findings warrant. “These creds date from about March. One of the orgs impacted told me they’d rotated them all and it’s a nothingburger, so I looked at their responsible disclosure policy, it allows trying creds, so I tried them all. Almost every one worked. Submitted report. One of the biggest US techcos,” Beaumont wrote. “This magnitude pushes us into a completely new world regarding the type of response required from the cybersecurity industry,” Gal concluded.

More than 2,500 organizations and over 430,000 CI/CD pipelines were affected by the LiteLLM supply chain attack earlier this year, CloudSEK reports. The LiteLLM compromise was disclosed shortly after the supply chain attack on Aqua Security’s Trivy open source vulnerability scanner and was a direct result of it. According to CloudSEK, TeamPCP, the threat actor behind multiple high-profile open source software (OSS) compromises, never targeted LiteLLM directly. The open source Python library and proxy server was compromised after its CI pipeline installed the compromised Trivy version automatically. Two LiteLLM versions, namely 1.82.7 and 1.82.8, were pushed to PyPI, providing the hackers with access to all the information LiteLLM touched. “Trivy, then the [LiteLLM] build system, then the LiteLLM release: one unrevoked token, three tools deep. That chain is what turns a single credential leak into ecosystem-wide exposure,” CloudSEK notes. “Automated build systems compress time. Once a malicious artifact reaches a registry, scheduled jobs, dependency resolvers, ephemeral runners, developer laptops, and cached layers can copy it rapidly. The forensic and credential-rotation window therefore extends beyond package removal,” the company continues. The modified LiteLLM versions contained malicious code executed on every Python invocation, with no explicit import. The payload ran on all systems where the package was installed. According to CloudSEK, while the affected packages were live for only 40 minutes, the window was long enough for the malicious code to propagate, ultimately exposing 434,000 CI/CD pipelines and impacting over 2,500 organizations. Nvidia, AWS, Samsung, Salesforce, Cisco, ServiceNow, Accenture Federal Services, Siemens, Regeneron Pharmaceuticals, London Stock Exchange Group, FedEx, Volkswagen, Orange, HP, Deutsche Bahn, NGINX, and Zscaler are only some of the names on CloudSEK’s list. “The 2,500+ company and 434,000 pipeline figures describe reconstructed exposure. They should not be read as proof that every listed organization was successfully compromised or that every credential was stolen,” CloudSEK says, noting that compromise should be independently verified in each case. The LiteLLM supply chain attack led to broad sensitive information compromise: package publishing credentials, cloud keys, SSH keys, tokens, environment variables, runtime data, and AI provider keys, among others. Hackers could use these secrets to take over accounts, steal data, inject malicious commits, achieve persistence, move laterally, disrupt services, deploy malware, and mount various other types of attacks. Organizations should consider any secret accessible to the LiteLLM library as compromised, including those “present in process memory, injected into the job, stored on disk, or retrievable through an instance metadata service”. Potentially compromised secrets should be validated, then rotated alongside service accounts and sessions, and logs should be reviewed to determine the exposure scope and timeframe. According to CloudSEK, the next major supply chain attack will likely target AI infrastructure, as these systems have become “high-value junctions between data, identity, compute, and autonomous action”. “The incident was not only a software supply chain breach that happened to involve an AI product. It demonstrated that compromising an AI control point can expose the identities and systems around it. Future attacks are likely to target the AI layer precisely because it is connected to everything else,” CloudSEK notes. UPDATE: SOCRadar reported that most of the 2,500 organizations believed to have been affected by the LiteLLM supply chain attack were actually affected earlier, likely via the Trivy incident. Related: Over 400 NPM Packages Infected in ChainDrop Supply Chain Attack Related: New GitHub, PyPI Policies Boost Supply Chain Security Related: Multiple Jscrambler Packages Impacted by Supply Chain Attack Related: More Klue Breach Victims Identified as Hackers Get Hacked

DEF CON — A security researcher has revealed severe, now-resolved security vulnerabilities in the Connective digital identity system, a browser extension used by over two million users in Belgium. Developed by Nitro Software Belgium, the software is used by eight of Belgium’s ten largest banks and over 60 government agencies to manage digital identity authentication and execute legally binding electronic signatures. James Arnott, security researcher and founder of cybersecurity firm Bay Area Labs, discovered that the software failed to verify which website was attempting to communicate with the user’s computer. Because these checks were missing, any website or embedded online ad could interact directly with the Connective application running on a victim’s machine without their knowledge or permission. According to Arnott, a malicious website could silently read connected electronic ID (eID) and payment card details. Furthermore, attackers could trick users into revealing their eID PIN by triggering official-looking authentication pop-ups. Because the software allowed web pages to customize the text inside these dialog boxes without displaying the domain making the request, users had no way to verify whether a prompt was legitimate or a phishing attempt. When a user entered their PIN into a prompt, the application transmitted it back to the requesting webpage. An attacker could then use the PIN to generate unauthorized approval tokens to forge legally binding electronic signatures whenever the victim’s physical eID card was inserted into a card reader. The compromise of the eID system severely impacted the trust model of Belgium’s broader digital ecosystem, including government portals like CSAM.be and third-party identity providers like Itsme. While these service providers contained no flaws of their own, their reliance on eID signatures meant that an attacker with stolen signing capabilities could register or hijack digital identity accounts. In addition to identity theft, the researcher uncovered a remote code execution vulnerability that operated independently of whether an eID card was plugged in. By exploiting a flaw in how the application processed files on the local computer, a malicious website could force the software to execute attacker-controlled code at the user level. An attacker could execute this drive-by attack by tricking a user into downloading a file disguised as a standard document and visiting a webpage. Requiring no special permissions, the flaw also carried the risk of spreading like a self-propagating worm by hijacking user credentials to send malicious links to other potential victims. Nitro fully remediated the issues 146 days after the initial report and awarded a $200 bug bounty. The company deployed updates to block unauthorized origin requests and secure PIN handling, with final security enforcement completed in late July. No CVEs appear to have been assigned. Arnott publicly disclosed the findings at DEF CON and released a blog post with additional technical details. UPDATE, August 13, 2026: Nitro provided the following statement: In early 2026, as part of our bug bounty program an independent security researcher reported vulnerabilities in the browser extension and local signing component that enable Belgian eID signing in Nitro Sign Enterprise Verified and identification in Identity Service. We investigated, developed fixes, and a series of remediations were deployed, with the final hardening enforced in production in July 2026. The vulnerabilities are now resolved. We found no evidence that any of these vulnerabilities were exploited. The vulnerabilities were reported to the Centre for Cybersecurity Belgium, and we coordinated closely with them throughout the process. Customers running current versions are protected. The fixes are distributed automatically through the Nitro plugin installer, which delivers the updated SignID component and the current browser extensions. Any customer running an older on-premise deployment can contact our team at [email protected] for assistance in confirming they are up to date. Related: Critical One-Click Vulnerability in Atlassian’s Rovo AI Exposed Enterprise Data Related: How a $50,000 Exploit Chain Turned Bixby Against Samsung Phones Related: Truck Brake Controller’s Safety Recall Doubled as Hidden Security Fix

DEF CON — Varonis Threat Labs has disclosed a one-click vulnerability in Rovo, Atlassian’s enterprise AI assistant, that let a specially crafted link seed attacker-controlled instructions directly into a user’s live AI session. Dubbed RovoBlast, the flaw required no jailbreak and no permission bypass, relying on the fact that the assistant simply treated externally supplied parameters as trusted input. Rovo functions as an AI layer spanning Jira, Confluence, Bitbucket, and third-party tools such as Slack, Microsoft 365, and Google Workspace. It also carries autonomous agent features capable of completing multi-step tasks with no further user involvement, which is what enabled the RovoBlast attack. [ Read: How a $50,000 Exploit Chain Turned Bixby Against Samsung Phones ] The exploit leveraged a URL parameter called rovoChatPrompt, which pre-fills content straight into Rovo’s chat window. Varonis researchers describe this attack path as parameter-to-prompt (P2P) injection, which they previously reported in Microsoft Copilot as Reprompt in January. Researchers noticed that the organization ID part of the URL could be left blank and Atlassian would still route the request into the victim’s own default organization, all without any warning or indicator that the session had been seeded by an outside source. To gauge the potential blast radius, the researchers simply asked Rovo what data it could see. The AI’s answer included Jira, Confluence, Bitbucket, Slack, Google Workspace, Microsoft 365, relational databases, uploaded files, web pages, and archived content. The actual leakage came from ResearchAgent, one of Rovo’s built-in tools, which can autonomously conduct multi-source web research and navigate across arbitrary sites. Once an attacker’s prompt was seeded through the malicious link, that same capability let Rovo pull internal data and push it out to the open web in a single automated chain. The team demonstrated the technique in three separate proof-of-concept scenarios: exfiltrating Confluence pages, Jira tickets, and SharePoint content containing personal data. Notably, the researchers found that a single seeded link was generally enough to trigger the leak. The attack didn’t require chaining multiple requests or any additional bypass steps to get Rovo to retrieve and summarize sensitive data. Varonis disclosed RovoBlast to Atlassian, which fixed the issue before the findings were published. The researchers recommend that organizations limit which systems Rovo can reach, disconnect unused integrations, wall off sensitive areas such as legal, HR, and finance, disable browsing or multistep automation features that aren’t in active use, and pair this with routine monitoring of assistant activity logs. Varonis presented the research at DEF CON 34 on Friday. A technical write-up is available on the Varonis blog. *statement from Atlassian removed at Atlassian’s request Related: Meta AI Hacked External Systems During Cybersecurity Testing Related: Atlassian, Splunk Patch Critical Vulnerabilities