Weekly Threat Briefing — 2026-11-09 to 2026-11-16
85 findings209 articles6 sections
- Critical findings
- 0
- High findings
- 0
- CISA KEV additions
- 0
- First seen in this period
- 209 Articles2,585 CVEs14,944 IOCs
First seen in this period: news articles, each counted once · CVEs rated CVSS 7.0+ (NVD, GitHub, WordPress) · CISA KEV catalog additions · new abuse.ch indicators (ThreatFox, URLhaus, Feodo Tracker), dated by abuse.ch.
Priority queue
Act on these first
Critical and high findings from this briefing and its live updates. Most severe first, then newest.
- Critical severitySection VulnerabilitiesSource: KEV
- Critical severitySection VulnerabilitiesSource: KEV
- Critical severitySection VulnerabilitiesSource: KEV
- Critical severitySection VulnerabilitiesSource: KEV
- Critical severitySection VulnerabilitiesSource: KEV
- Critical severitySection VulnerabilitiesSource: KEV
- Critical severitySection VulnerabilitiesSource: KEV
- Critical severitySection VulnerabilitiesSource: RSS
- Critical severitySection VulnerabilitiesSource: RSS
- Critical severitySection VulnerabilitiesSource: CVE
Executive summary · as of 07:04 UTC
The situation
This week was dominated by attackers abusing remote management (RMM) and edge-device platforms, alongside a sharp rise in AI-driven offensive tooling. CISA added 12 CVEs to the Known Exploited Vulnerabilities (KEV) catalog. They include a Check Point Gaia root privilege-escalation zero-day, a Microsoft Outlook zero-day patched in a roughly 1,100-CVE Patch Tuesday, a Chromium use-after-free, two Barracuda ESG flaws and an OpenSSH authentication bypass. A Check Point zero-day exploited for root access since late October coincided with PHP2Shell attacks that deployed a VMware ESXi wiper via Check Point gateways. Two MSP-facing campaigns also stood out. ScreenConnect environments were hit by the worm-like SCKit toolkit at nearly 50 organizations, and the N-able N-central '4EverAble2Shell' zero-day was exploited within hours of a public PoC, including on fully patched servers.
The FBI issued a FLASH alert on ARTEX, an open-source AI penetration-testing framework. According to the FBI, ARTEX moves intrusions from targeting to privileged access in under three hours and was used in a ConnectWise supply-chain breach of Microsoft 365 tenants. Unit 42, Elastic and Symantec documented similar abuse of Cairn, CyberStrikeAI and AI-generated lateral-movement scripts. Supply-chain risk in AI tooling also materialized: stolen Fireworks AI credentials led to a malicious Claude Code extension in the Cursor marketplace and to the theft of Uber driver data. Separately, the 'wfx-loot' campaign used 4,837 compromised GitHub accounts to push secrets-stealing code. Law enforcement had a strong week, with these actions: - Europol dismantled Gootloader, CastleRAT and StealC infrastructure. - The US disrupted North Korea's 'FaceStealer' campaign and seized $1.2 billion from the Binance theft. - Police arrested 354 dark-web market participants.
On the data side, HIBP loaded credential-bearing corpora: YouHack, with 7.16M accounts and passwords, and 179,745 CastleRAT victim credentials supplied by Europol. Extortion actors claimed large public-sector datasets, including Michigan State University and Texas HHSC. Disclosed incidents included La Poste's Maileva (22M people), Muji (over 1M card numbers) and TSMC (46,000 employees). Qilin remains the most active ransomware program with at least 365 victims claimed in three months, while PixelLeaks posted a burst of roughly 25 victims in a single day. abuse.ch logged 14,875 new indicators across 427 families. ClickFix/ClearFake delivery, Mirai/Mozi IoT payloads and GitHub-hosted fake VPN installers carrying SalatStealer and SmartLoader dominated the volume.
Browse the full briefing
85 findings · 6 sections- 15 critical
- 34 high
- 28 medium
- 5 low
- 3 info
- Vulnerabilities23 findingsExploited & Critical Vulnerabilities11 critical · 10 high · 2 medium
- Malware16 findingsMalware & Campaigns2 critical · 4 high · 10 medium
- Threat Actors11 findingsThreat Actors & Law Enforcement Actions6 high · 5 medium
- Tools & Techniques11 findingsTools, Techniques & Research1 critical · 3 high · 4 medium · 1 low · 2 info
- Data Breaches & Leaks17 findingsData Breaches & Leaks1 critical · 11 high · 5 medium
- Policy7 findingsPolicy & Regulation2 medium · 4 low · 1 info
Exploited & Critical Vulnerabilities
CISA added 12 CVEs to KEV this window. Priorities are Check Point, Outlook, Chromium, Barracuda, OpenSSH and GPUStack. Unpatched or freshly disclosed zero-days in N-able N-central and Windows (SignDrop) were exploited or made public. The window held 2,542 NVD CVEs, 216 OSV advisories and 235 WordPress vulnerabilities; the most critical are highlighted below.
- Critical severityCheck Point Gaia DynamicCLI permission flaw (CVE-2026-123039) exploited for root (opens the source in a new tab)
Added to CISA KEV on 2026-11-13. The Gaia OS DynamicCLI daemon runs as root and listens on a world-writable local resource, enabling privilege escalation. Affected products are Check Point Firewall, Multi-Domain Security Management Server, Scalable Platforms, Security Gateway and Security Management Server. Check Point released emergency hotfixes for a zero-day exploited since late October to gain root and persist on gateways. PHP2Shell attacks on Check Point gateways have also deployed an ESXi wiper. Apply hotfixes and hunt for persistence on gateways.
- Critical severityMicrosoft Outlook spoofing zero-day (CVE-2026-124194) under active exploitation (opens the source in a new tab)
Added to KEV on 2026-11-10 alongside November Patch Tuesday. This is a missing-user-interaction spoofing flaw in Outlook that can be exploited over the network. Microsoft's release fixes an actively exploited Outlook zero-day. Office updates must be checked and deployed separately.
Source: KEV - Critical severityGoogle Chromium XML use-after-free (CVE-2026-125350) exploited in the wild (opens the source in a new tab)
A use-after-free in Chromium's XML component allows in-sandbox code execution via a crafted page; it was added to KEV on 2026-11-10. Google shipped desktop and Android fixes for an exploited zero-day and pushed a second Chrome update the same week. All Chromium-based browsers are affected.
- Critical severityBarracuda ESG authentication bypass (CVE-2026-116606) (opens the source in a new tab)
An unauthenticated remote attacker can invoke the ESG configuration-import function and restore a crafted backup, modifying the appliance configuration. Added to KEV on 2026-11-10.
Source: KEV - Critical severityBarracuda multiple products OS command injection (CVE-2026-111560) (opens the source in a new tab)
An OS command injection flaw affects multiple Barracuda products: ESG, Load Balancer ADC, Message Archiver, WAF Control Center, WAF and Web Security Gateway. Added to KEV on 2026-11-10. Prioritize internet-exposed appliances.
- Critical severityOpenSSH on Linux authentication bypass (CVE-2026-123158) (opens the source in a new tab)
An unexpected status code or return value in OpenSSH on Linux could allow authentication bypass. Added to KEV on 2026-11-09. Patch broadly-deployed SSH servers and review authentication logs for anomalous successful logins.
Source: KEV - Critical severityGPUStack improper authentication (CVE-2026-102895) exploited (opens the source in a new tab)
Exploitation gives unauthenticated admin access via a trusted-loopback or X-GPUStack-Real-IP header bypass (GHSA-gj9m-f8ph-pmhr, CVSS 9.8). Attackers can make user and model changes, create persistent admin API keys and disclose worker and cloud-provider secrets. Code execution is possible through the inference-backends endpoint. Added to KEV on 2026-11-09. Related GPUStack flaws CVE-2026-102894 (SAML bypass) and CVE-2026-103002 (missing authorization) were also published.
Source: KEV - Critical severityN-able N-central '4EverAble2Shell' zero-day exploited hours after PoC (opens the source in a new tab)
Attackers exploited a newly disclosed N-central zero-day within hours of a PoC release to compromise servers and deploy cryptominers. At least 21 systems were affected, 19 of them on the latest security update. N-able took N-central offline after 33 flaws were reported and issued a hotfix. Isolate N-central servers, apply the hotfix and audit managed endpoints for pushed payloads.
- Critical severityApache HTTP Server HTTP/2 unauthenticated RCE patched (opens the source in a new tab)
A critical flaw in mod_http2 can give an unauthenticated attacker remote code execution. Patch, or disable HTTP/2 on exposed servers until updated.
Source: RSS - Critical severitySAP NetWeaver ABAP authentication bypass (CVE-2026-120017) and related SAP criticals (opens the source in a new tab)
Insufficient verification of authentication assertions in AS ABAP lets unauthenticated attackers impersonate arbitrary users, including privileged accounts (CVSS 10). SAP's same release also covers these issues: - CVE-2026-89103: SAP Gateway default ACL - CVE-2026-89126: Web Dispatcher memory corruption - CVE-2026-76964: ABAP memory corruption - CVE-2026-76965: OS command execution
Source: CVE - Critical severityReact Server Components unauthenticated RCE (CVE-2026-126187) (opens the source in a new tab)
A critical unauthenticated RCE (CVSS 9.5) affects React Server Components in react-server-dom-parcel (GHSA-xqc2-xf4c-x3h3). Update RSC packages across Node.js applications.
Source: CVE - High severityCisco BroadWorks XXE (CVE-2026-20320) (opens the source in a new tab)
An unauthenticated XXE allows reading sensitive configuration from Cisco BroadWorks. Added to KEV on 2026-11-10.
Source: KEV - High severitySignalWire FreeSWITCH code injection (CVE-2026-116779) (opens the source in a new tab)
A code injection flaw permits configuration and credential disclosure, arbitrary call control and command execution. Added to KEV on 2026-11-09. Relevant to VoIP and UC environments.
- High severityIBM Aspera Faspex incorrect authorization (CVE-2026-123218) (opens the source in a new tab)
Authenticated users can craft a message to bypass authorization and escalate privileges. Faspex has historically been a target for data-theft actors. Added to KEV on 2026-11-09.
Source: KEV - High severityNovember Patch Tuesday: ~1,100 fixes, 63 critical (opens the source in a new tab)
Microsoft fixed nearly 1,100 vulnerabilities, 63 of them critical, including the exploited Outlook zero-day. Other critical bugs affect Exchange, SharePoint, Windows Search and LDAP. Known side effects of the updates are printing failures with complex documents and Windows activation errors (0x80041008).
Source: RSS - High severityPublic Windows 'SignDrop' zero-day grants SYSTEM privileges (opens the source in a new tab)
Researcher Abdelhamid Naceri (Nightmare Eclipse) released an unpatched Windows LPE exploit granting SYSTEM. Expect rapid adoption by intruders. Monitor for anomalous SYSTEM process creation from user contexts.
Source: RSS - High severityAdobe Connect critical SSRF, SQLi and code-execution batch (opens the source in a new tab)
Adobe Connect received a large batch of CVSS 9.9–10 fixes. They cover SSRF leading to privilege escalation (CVE-2026-121889, CVE-2026-121356), CRLF injection, incorrect authorization and path traversal. Several SQL injection, code injection and XXE flaws lead to code execution. Adobe Experience Manager code-injection and authorization flaws (e.g. CVE-2026-113580) were also published.
Source: CVE - High severityMongoDB Server replication use-after-free (CVE-2026-118383) (opens the source in a new tab)
A use-after-free in replica-set catch-up can be triggered by unauthenticated requests (CVSS 10). It can cause crashes or, under certain conditions, arbitrary code execution.
Source: CVE - High severitySharePoint privilege escalation (CVE-2026-120002) (opens the source in a new tab)
An incorrect authorization flaw allows an authorized attacker to elevate privileges over the network (CVSS 9.9). Azure DevOps improper authentication enabling code execution (CVE-2026-109467) was also published.
Source: CVE - High severityDrayTek Vigor2960 command injection (CVE-2026-129449) with public exploit (opens the source in a new tab)
Command injection via swmconfigupload in mainfunction.cgi on Vigor2960 up to 1.5.1.6 has a public exploit. The same model has path traversal and heap overflow flaws (CVE-2026-129450/129451/129453). Public exploits are also available for D-Link DIR-600L and DSL-6740C (CVSS 10) and TRENDnet TEW-657BRM, which are likely botnet targets.
Source: CVE - High severityWordPress Confection plugin unauthenticated PHP code execution (CVE-2026-115252) (opens the source in a new tab)
The plugin (0.4.0–0.4.7) writes unauthenticated settings into a served PHP file, enabling RCE (CVSS 10). Easy DragDrop File Uploader (CVE-2026-124822) allows unauthenticated arbitrary file deletion, including wp-config.php. In total, 235 WordPress vulnerabilities entered the feed this window.
Source: CVE - Medium severityUnrealIRCd delimiter injection (CVE-2026-114713) (opens the source in a new tab)
Improper neutralization of delimiters lets any user on a multi-server network inject server-to-server commands executed with server authority. Added to KEV on 2026-11-13.
Source: KEV - Medium severityOpenwave Webtop Media Transcoder resource transfer flaw (CVE-2026-111793) (opens the source in a new tab)
An authenticated user who bypasses the Webmail experience can convert unintended document types. Added to KEV on 2026-11-10.
Source: KEV
Malware & Campaigns
Attacks on remote management (RMM) and MSP platforms (SCKit/ScreenConnect) and a destructive ESXi wiper delivered via Check Point gateways lead the malware picture. They are joined by developer supply-chain compromises (GitHub wfx-loot, Cursor) and Linux persistence via GSocket and SANSAR. abuse.ch recorded 14,875 new indicators across 427 families; notable families are summarized below.
- Critical severitySCKit abuses ScreenConnect to compromise MSPs and their customers (opens the source in a new tab)
Microsoft detailed SCKit, a worm-like toolkit used in active ScreenConnect attacks on MSPs. It deploys rogue remote-access clients to customer devices, then spreads to MSP technician machines to reach additional managed networks. ConnectWise attributed the breaches at nearly 50 organizations to abuse of a third-party tool and disabled five third-party integrations. MSPs should audit ScreenConnect extensions and unexpected access clients, and use the Defender XDR hunting queries Microsoft provided.
- Critical severityCheck Point gateways hacked via PHP2Shell to deploy VMware ESXi wiper (opens the source in a new tab)
Vulnerable Check Point Security Gateways were compromised in PHP2Shell attacks that deployed data-wiping malware against VMware ESXi hosts. Patch gateways, verify the integrity of ESXi backups and restrict management access to the hypervisor.
- High severity'wfx-loot' campaign hijacks 4,800 GitHub accounts to steal cloud credentials (opens the source in a new tab)
The 'wfx-loot' campaign used 4,837 compromised GitHub accounts to probe 61,764 repositories. It opened pull requests and pushed secrets-stealing malware aimed at cloud credentials. Review recent PRs and workflow changes, and rotate CI/CD secrets.
- High severityCursor breach led to malicious Claude Code extension in marketplace (opens the source in a new tab)
Credentials stolen from AI infrastructure provider Fireworks AI let attackers publish a malicious Claude Code extension through the Cursor marketplace. The same Fireworks AI compromise exposed Uber driver license data. Developers should inventory installed IDE extensions.
- High severityGSocket weaponized for stealthy persistent C2 on Linux (opens the source in a new tab)
Threat actors install GSocket on compromised Linux servers to keep encrypted remote access after the initial vulnerability is patched. Microsoft provides detection and remediation guidance. Hunt for gs-netcat/gsocket binaries and persistence entries on previously exploited hosts.
- High severityClickFix-style delivery: IClickFix and ClearFake (1,292 indicators)
IClickFix: 656 indicators (domains, URLs, ip:port), threat type payload_delivery. ClearFake: 636 indicators (domains, URLs, SHA256), threat types payload delivery, payload and botnet C2. Samples: datadevfuture[.]buzz, api-dns[.]org, allafricabackpackers[.]shop. Fake-CAPTCHA and fake-update lures that make users paste commands remain a top initial-access vector.
Source: abuse.ch - Medium severitySANSAR Linux cryptominer hides GPU use with rootkit and Ethereum-based C2 (opens the source in a new tab)
Elastic analyzed SANSAR, a Linux cryptomining orchestrator. It manages miners over Tor and pairs them with a kernel rootkit and GPU-monitoring hooks that hide utilization. Its backdoor retrieves the C2 address from the Ethereum blockchain.
- Medium severityTrezor confirms crypto-stealing hardware implants in resold wallets (opens the source in a new tab)
Three Trezor Safe 5 devices sold by Southeast Asian reseller CryptoBilis contained hardware implants. The implants stole recovery phrases and exfiltrated them over cellular networks. Trezor released a fix.
Source: RSS - Medium severityRemus infostealer claims persistent Gmail access after password reset (opens the source in a new tab)
Remus developers claim the stealer keeps Gmail access after a password reset, likely through session or token theft. Incident response should revoke sessions and tokens, not just reset passwords.
- Medium severityArcticRAT Android gang uses AI to profile victim finances (opens the source in a new tab)
The ArcticRAT Android malware has expanded from crypto theft to full device control. The operators use AI to profile victims' finances from stolen data.
Source: RSS - Medium severityFormbook targets European manufacturing with localized order lures (opens the source in a new tab)
Formbook (XLoader) campaigns use localized order and procurement lures to target UK and Italian manufacturing and technology firms.
Source: RSS - Medium severityUnknown Loader leads abuse.ch volume (2,321 indicators)
2,321 ThreatFox indicators (domains, SHA256, ip:port, URLs) tagged Unknown Loader. Threat types: payload delivery, payload and botnet C2. Samples: ed8b7ef6823693b68301ade76e6280aecdd267db497679969d00dd7c54761790, 2d15e66ce69755771af6bf74078afa11a1749b2381ba90601bca6ccc80f474d9, b32260de68f5a5f5226029f65cedcccd70160c35161f4c719d6cd3f0fd856c50.
Source: abuse.ch - Medium severityGitHub-hosted fake VPN installers deliver SalatStealer and SmartLoader
Clusters of GitHub release URLs impersonate AdGuard VPN, ProtonVPN, v2rayTun and Happ installers to deliver SalatStealer, SmartLoader and zapret-tagged payloads: about 1,055 URLs in total. SalatStealer separately had 345 hash/URL indicators. Samples: hxxps://github[.]com/noonpythoncast59/aDguARD_vPN/releases/download/1[.]7[.]9/adguardvpn[.]zip, hxxps://github[.]com/hectosponsorgap/v2rayTuN_WinDOws/releases/download/1[.]10[.]4/v2raytun[.]zip, hxxps://github[.]com/holgercool/Winget-Dashboard/releases/download/v1[.]4[.]6-beta[.]4/protobuf-ts-types[.]v1[.]4[.]6-beta[.]4[.]zip.
Source: abuse.ch - Medium severityMirai and Mozi IoT botnet payloads (1,300+ indicators)
Mirai: 816 indicators (hashes, ip:port, URLs, domains) for payloads and botnet C2. Mozi: 525 URLs for MIPS/ARM ELF downloads, plus further generic ELF downloader URLs. Samples: hxxp://38[.]226[.]161[.]86:51355/bin[.]sh, hxxp://185[.]9[.]139[.]102:34606/i, 450ccc09f78572b602dc434f9c8f11cf4f74fb9d13bfc941c876929bb8b488cc. Correlates with this week's public router exploits for D-Link, DrayTek and TRENDnet devices.
Source: abuse.ch - Medium severityVidar infostealer C2 (353 indicators)
353 indicators (domains, URLs, hashes, ip:port) covering botnet C2 and payloads. Samples: z[.]dhawa[.]net, s8[.]amarta89[.]org, ml[.]amor76[.]com. Other active stealers included Raven Stealer (201), WARDENStealer (100), RN Stealer (89) and StrelaStealer (88).
Source: abuse.ch - Medium severityCobalt Strike and commodity RAT C2 (Remcos, NjRAT, PureRAT, AsyncRAT)
Cobalt Strike: 176 ip:port and hash indicators. Samples: 38[.]174[.]217[.]97:8080, 38[.]174[.]217[.]97:888, 38[.]174[.]217[.]97:8888. Commodity RAT C2 also remained active: Remcos (136), NjRAT (127), PureRAT (107) and AsyncRAT (99).
Source: abuse.ch
Threat Actors & Law Enforcement Actions
Law-enforcement pressure was substantial: the Gootloader/CastleRAT/StealC takedown, the FaceStealer disruption with a $1.2B Binance recovery, and 354 dark-web arrests. Qilin continues to lead ransomware activity, and state-linked espionage targeted governments in Central Asia and Romania.
- High severityEuropol dismantles Gootloader, CastleRAT and StealC infrastructure (opens the source in a new tab)
Europol and partners from nine countries seized hundreds of servers and domains supporting Gootloader, CastleRAT and StealC. The DOJ linked Gootloader to at least 44 ransomware attacks costing $29M since 2018. German authorities traced the rebuilt StealC network, taken down for the second time this year, to the Virtualine bulletproof hoster seized by the FBI in October. HIBP ingested 179,745 CastleRAT victim credentials.
- High severityUS seizes $1.2B from Binance theft, disrupts DPRK 'FaceStealer' (opens the source in a new tab)
DOJ and FBI announced the seizure of $1.2 billion stolen in the Binance theft. They also disrupted the 'FaceStealer' social engineering campaign, which had stolen more than $400M in cryptocurrency.
Source: RSS - High severity354 arrested in dark-web marketplace crackdown (opens the source in a new tab)
An international operation arrested 354 vendors and buyers on the Archetyp, Nemesis, Tor2Door, Bohemia and Kingdom dark web markets.
Source: RSS - High severityQilin ransomware: most active affiliate program, administrator profiled (opens the source in a new tab)
Krebs profiles the administrator of Qilin, now the most active ransomware affiliate program with at least 365 victims claimed in three months, and documents OPSEC missteps. Qilin posted six new US victims on 15 November, including manufacturing, construction and inspection firms.
Source: RSS - High severityLuna Moth re-targets law firms to steal Microsoft tokens (opens the source in a new tab)
The extortion group is retargeting law firms that refused to pay its first ransom. It uses social engineering to steal Microsoft access tokens. Review M365 token issuance and OAuth consent.
- High severityFalcon C2 browser framework targets governments (opens the source in a new tab)
Unit 42 uncovered Falcon C2, a browser-based framework that targets governments through malicious scripts and extensions plus the Antino implant.
- Medium severityKazakhstan-linked actors use AI agent to spy on Central Asian governments (opens the source in a new tab)
Suspected Kazakhstan-linked actors used an AI agent to steal documents from Central Asian governments and infiltrate critical-infrastructure networks.
Source: RSS - Medium severityRomania disrupts Russian espionage targeting government email (opens the source in a new tab)
Romania's intelligence service said it stopped a Russian state-linked cyber-espionage campaign against government email before any data was exfiltrated.
Source: RSS - Medium severityGerman military drone maker Quantum Systems hit by cyberattack (opens the source in a new tab)
German defense firm Quantum Systems confirmed a late-October incident that was part of an active attack by hostile actors and disrupted military drone operations.
Source: RSS - Medium severityRansomware hits Belarusian cloud provider ActiveCloud (opens the source in a new tab)
Attackers encrypted parts of ActiveCloud's infrastructure, leaving customers without key services for days. In a separate incident, Poland shut down its ePUAP government services platform after a cyberattack.
Source: RSS - Medium severityNew 'Curious Worm' group targets healthcare (opens the source in a new tab)
'Curious Worm' is a new actor that emerged in October–November. More than half of its 12 claimed incidents target healthcare.
Source: RSS
Tools, Techniques & Research
AI-driven offensive tooling became operational reality this week. The FBI, Unit 42, Elastic and Symantec documented attacks run by ARTEX, Cairn and CyberStrikeAI at machine speed, and research highlighted supply-chain risks to AI coding agents. DFIR-relevant forensic research is also included.
- Critical severityFBI FLASH: ARTEX AI pen-test framework powering global attacks (opens the source in a new tab)
The FBI warns that ARTEX, an open-source AI penetration-testing framework, is being used to automate intrusions worldwide. It was used in a ConnectWise supply-chain breach that compromised Microsoft 365 customer environments. Some intrusions reach privileged access in under three hours. Dark Reading notes that forks and 45 similar tools can run credential and cloud attacks at machine speed.
- High severityDetecting agentic offensive frameworks (ARTEX, CyberStrikeAI, Cairn, PentAGI, HexStrike) (opens the source in a new tab)
Elastic profiles five agentic offensive frameworks: ARTEX, CyberStrikeAI, Cairn, PentAGI and HexStrike. They act in 2–25 second bursts, and none enforces authorization in code. Elastic shares the telemetry they leave and an ES|QL hunt to catch them.
Source: RSS - High severityCyberStrikeAI intrusion across AWS accounts via CloudTrail (opens the source in a new tab)
A CyberStrikeAI server operated for 16 days across four AWS accounts using eight identities, issuing 56 API calls in 141 ms. Elastic's CloudTrail analysis shows why six detection rules each caught only part of the activity.
- High severityCairn AI framework used for autonomous attacks on semiconductor and defense (opens the source in a new tab)
Unit 42 analyzed attackers' use of the open-source Cairn agent framework to autonomously breach semiconductor, defense and civil-society targets.
Source: RSS - Medium severityAI-generated scripts rescue stalled intrusion at US retailer (opens the source in a new tab)
Symantec describes an attacker whose hand-typed commands failed against a US retailer. The attacker switched to AI-generated scripts and achieved lateral movement within minutes.
- Medium severityPost-training supply chain attacks hijack AI coding agents (opens the source in a new tab)
Microsoft shows how poisoned datasets and model checkpoints can make coding agents execute remote code. Related research shows a sleeper agent can be planted by editing fewer than ten neurons, and that this is detectable by frontier agents.
Source: RSS - Medium severityAI coding agents leaking secrets to GitHub (opens the source in a new tab)
A Claude Code command inadvertently commits files containing credentials to public Git repositories. Organizations should enable secret scanning on repositories touched by AI agents.
Source: RSS - Medium severityStealing credentials through permissive CSP allow-lists (opens the source in a new tab)
PortSwigger shows how Content Security Policy (CSP) allow-listed domains can turn HTML injection into credential theft without XSS.
Source: RSS - Low severityMicrosoft FORGE: 59 Linux kernel bugs, eight local root exploits (opens the source in a new tab)
Microsoft FORGE Lab's AI-assisted research produced 59 Linux kernel bug reports, 33 CVEs and 38 merged fixes. It includes eight demonstrated local root exploits.
Source: RSS - Info severityDFIR: investigating BEC and fraud in Microsoft 365 (opens the source in a new tab)
Magnet Forensics walks through investigating BEC in Microsoft 365, from forwarding rules and OAuth abuse to a full AXIOM Cyber workflow. This is relevant to Luna Moth and ARTEX M365 compromises.
- Info severityAndroid FCM push notification artifacts survive app deletion (opens the source in a new tab)
GMDSOFT shows that Firebase Cloud Messaging artifacts in Google Play Services can reveal app activity after the app has been deleted.
Source: RSS
Data Breaches & Leaks
The breach feeds held 371 entries this window; 40 were reviewed. Credential exposures lead: HIBP loaded the YouHack forum corpus (7.16M) and CastleRAT victim credentials from the Europol takedown. BYOD and PixelLeaks drove large extortion claims, and disclosed incidents at La Poste/Maileva, Muji, Uber and TSMC each affected thousands to millions.
- Critical severityYouHack forum breach (7.16M accounts with passwords) (opens the source in a new tab)
HIBP added 7,160,849 accounts associated with the Russian hacking forum YouHack, redistributed as part of a larger corpus. Exposed: email addresses, IP addresses, passwords and usernames. The original May 2013 breach covered 107k members with mostly unsalted MD5 hashes.
Source: Breach - High severityCastleRAT victim credentials (179,745) from Europol takedown (opens the source in a new tab)
Europol supplied HIBP with 179,745 email addresses and passwords recovered during the CastleRAT infrastructure takedown. These are infostealer/RAT victim credentials. Check corporate domains and force resets.
Source: Breach - High severityMichigan State University — BYOD claims 1.03M records with passwords (opens the source in a new tab)
The BYOD extortion group claims full PII on 1.03 million current and former MSU students and staff, including current and former passwords, after negotiations failed.
Source: Breach - High severitySberSpasibo (Sberbank loyalty) — 3.29M accounts with card numbers (opens the source in a new tab)
HIBP added 3,288,711 unique emails from Sberbank's loyalty program data, published in March 2023 and covering 48M users. Exposed: bank card numbers, dates of birth, email addresses, loyalty program details and phone numbers.
Source: Breach - High severityLa Poste / Maileva — 22 million people exposed (opens the source in a new tab)
A mid-September cyberattack on Maileva, a digital subsidiary of La Poste Group, exposed personal data of about 22 million people. La Poste confirmed the incident.
Source: RSS - High severityTexas HHSC — BYOD claims 2M citizens' benefits and banking data (opens the source in a new tab)
BYOD issued a 'final warning' with a sample. It claims data on 2 million Texans, covering Medicaid, banking information, SNAP, TANF and disaster assistance records.
Source: Breach - High severityMuji operator — over 1M credit card numbers exposed (opens the source in a new tab)
Muji operator Ryohin Keikaku apologized for breaches of customer data, including more than one million credit card numbers. It is one of dozens of recent Japanese incidents; the City of Osaka also warned that a breach affected 1 million people.
Source: RSS - High severityUber driver licenses stolen via Fireworks AI vendor compromise (opens the source in a new tab)
Uber is notifying thousands of US drivers that their personal and driver's license data was stolen in October 2026 via compromise of third-party AI vendor Fireworks AI. This is the same compromise behind the Cursor extension incident.
Source: RSS - High severityUnocoin crypto exchange — 81,000 wallets show zero holdings (opens the source in a new tab)
Indian exchange Unocoin found unauthorized server access on 9 November and disabled logins; 81,000 wallets reportedly show zero holdings.
Source: RSS - High severityPC Express (Multimedia Informatica) — PixelLeaks, 24,000+ mailboxes (opens the source in a new tab)
PixelLeaks threatens to publish the hosted mail database of an Italian IT and email provider, covering more than 24,000 client mailboxes. It is one of about 25 PixelLeaks postings on 15 November that included healthcare labs (Clinica Suiza Lab in Peru, Huangshi Medical Laboratory), Godrej & Boyce and financial advisors.
Source: Breach - High severityClinica Suiza Lab — PixelLeaks healthcare dataset (opens the source in a new tab)
The claimed data is a Peruvian clinic and lab group's full patient register, including identity documents, clinical histories, lab results and imaging. Minors are among those affected.
Source: Breach - High severityStarbucks — N0n leaks consumer data (opens the source in a new tab)
N0n marked Starbucks as 'not paid — exposed' and claims customer records with devices, purchases and locations. N0n also listed Timberland, with more than 18,800 customer emails plus search and revenue data, and Airport Appliance.
Source: Breach - Medium severityTSMC — 46,000 employees' data leaked via cloud provider (opens the source in a new tab)
TSMC confirmed that employee data for about 46,000 staff (6.9GB) was auctioned. The data matched records held by a cloud service provider; no TSMC systems were compromised.
Source: RSS - Medium severityHong Kong Sogo — 820,000 loyalty members at risk (opens the source in a new tab)
Sogo disclosed unauthorized access potentially exposing up to 820,000 loyalty members' data; Chow Sang Sang reported a separate breach.
Source: RSS - Medium severityNorwegian & Widerøe — 144,249 passengers via Amadeus breach (opens the source in a new tab)
Data on 144,249 passengers of Norwegian and Widerøe leaked via the breach at Spanish IT provider Amadeus.
Source: RSS - Medium severityPenang Adventist Hospital — LockBit5 (opens the source in a new tab)
LockBit 5 listed Penang Adventist Hospital (Malaysia). Qilin posted six US victims on 15 November: Heartland Construction, Atlas Evaluation & Inspection, Apex, Oregon Linen, Cole Manufacturing and Kessler Soils.
Source: Breach - Medium severityVanderbilt University — systems restored, data theft claimed (opens the source in a new tab)
Vanderbilt is restoring systems while attackers claim theft of student and employee data. UT Arlington separately confirmed data theft in a cyberattack.
Source: RSS
Policy & Regulation
US policy is moving toward leniency on AI-driven breach enforcement, while Germany and Japan expanded offensive and active cyber-defense powers. Regulators in the UK, India and New York signalled stricter breach-reporting and privacy expectations.
- Medium severityWhite House AI task force urges 12-month leniency on AI-driven breach penalties (opens the source in a new tab)
The administration's AI task force urged regulators to hold off enforcement against companies hit by AI-driven attacks for at least 12 months. The aim is to encourage voluntary vulnerability and incident reporting.
Source: RSS - Medium severityGermany grants intelligence agencies hacking and sabotage powers (opens the source in a new tab)
The Bundestag passed, by 308–190, the largest postwar overhaul of German intelligence law, granting spy agencies offensive hacking and sabotage powers.
Source: RSS - Low severityJapan pushes emergency cybersecurity funding amid AI-assisted breaches (opens the source in a new tab)
The package adds public-private cyber responders and an AI-focused team, and gives more leeway to use active cyber-defense powers.
Source: RSS - Low severityIndia DPDP: ransomware without data theft can still require breach reporting (opens the source in a new tab)
According to K7, a ransomware attack that only locks systems can qualify as a reportable personal data breach under India's DPDP law, even without exfiltration.
Source: RSS - Low severityRecord cyber incident reports to UK ICO (opens the source in a new tab)
UK organizations reported a record number of cyber incidents to the data protection regulator this year. The rise is attributed to data extortion and AI.
Source: RSS - Low severityNY AG settles with PJ&A over 13.3M medical records breach ($300,000) (opens the source in a new tab)
Perry Johnson & Associates and four related companies settled with the New York AG over a 2023 breach exposing 13.3 million people's medical data.
Source: RSS - Info severityNew York City bans surveillance pricing (opens the source in a new tab)
NYC can fine companies $1,000 per verified instance of using personal data to set individualized prices.
Source: RSS
Sources & Attribution
These briefings are compiled from publicly available threat-intelligence feeds, which may include CISA KEV, NIST NVD, the GitHub Advisory Database (OSV), abuse.ch, and Wordfence Intelligence. Data-breach and credential-leak items may include data from Have I Been Pwned and ransomware.live.
CVE® is a registered trademark of The MITRE Corporation. CVE Records are © The MITRE Corporation, reproduced under the CVE Program Terms of Use. WordPress vulnerability data is provided by Wordfence Intelligence, © Defiant, Inc. Breach data from Have I Been Pwned is licensed under CC BY 4.0.