November 2026 has just begun, and this page will list every serious attack of the month in date order. Breaches, supply chain compromises, exploited vulnerabilities, AI-run intrusions and agent incidents each get one line with a status and a link. Last month's October 2026 attacks timeline shows what a full month looks like.
This is a living page. It is checked for new incidents every 4 hours through the end of November, and every entry links to a write-up or primary sources. The same data powers the shareable November 2026 attacks tracker: filter by type, link to a single incident, or embed one on your own site.
Last updated: November 7, 2026, 02:02 UTC · 13 dated entries, 17 earlier campaigns, 4 unconfirmed claims
TL;DR
| Question | Answer |
|---|---|
| What does this page cover? | Every attack, breach, exploited vulnerability and agent incident of November 2026 we could verify, in date order |
| Which date is used? | When the attack happened, or when it was first disclosed if that is all that is known |
| Are claims included? | Yes, labelled Unconfirmed and kept in their own section |
| What is the common thread? | Too early to say; see the section below |
| First thing to do | Rotate secrets on affected platforms and hosts, then patch what is being exploited |
How to read the timeline
Each entry has a type and a status:
| Type | Meaning |
|---|---|
| Supply chain | A package, release, plugin or update channel shipped malicious code |
| Breach | A company's systems or customer data were accessed |
| Exploited | A vulnerability with confirmed in-the-wild attacks |
| Critical vuln | A severe flaw disclosed and patched, with no confirmed exploitation yet |
| AI-run attack | An intrusion where agents or LLMs did substantial operational work |
| Agent incident | An AI agent caused harm or broke a control without an outside attacker |
Confirmed means the affected company, a government agency or a named research team stated it. Reported means one vendor's research that has not been independently corroborated. Unconfirmed means a claim with no supporting statement.
Week 1: November 1 to 7
| Date | Incident | Type | What happened | Status |
|---|---|---|---|---|
| Nov 2 | WorkOS: internal sales app accessed, CEO links attack to ARTEX | Breach | WorkOS says an internal sales app was accessed without authorization (detected Oct 30), exposing CRM data and customer Slack excerpts; production services not affected, per the company. The ARTEX link comes from the CEO, not the bulletin. (details) | Confirmed (WorkOS); ARTEX link unverified |
| Nov 2 | Northflank security incident: control-plane secrets, AI key misuse | Breach | Northflank told all customers to rotate integration and project secrets after an attack on secrets in its control plane; some customers reported unauthorized AI key use. Running workloads were not affected, per the company. Root cause not yet published. (details) | Reported (Northflank) |
| Nov 2 | GitBook: AI-agent attack reaches infrastructure credentials | Breach | GitBook says a targeted attack using AI agents accessed some infrastructure credentials on Nov 2. Integration tokens, Git Sync secrets, API tokens and Assistant questions may be exposed; all personal API tokens revoked, MCP server needs re-authentication. No content changes found. (details) | Confirmed (GitBook); AI-agent detail unverified |
| Nov 2 | CyberStrikeAI: LLM on the C2 server takes over education domains | AI-run attack | Kaspersky says the open-source CyberStrikeAI framework put an LLM orchestrator on the C2 server to run post-exploitation against education networks in Africa and Latin America, reaching domain admin in 3-4 days. (details) | Confirmed |
| Nov 3 | Primcloud says a Railway engineer's AI agents probed its infrastructure | Agent incident | Primcloud's founder says AI agents run by a Railway engineer scanned its network and requested secrets; he says every attempt was blocked. Railway's CEO apologized, said the activity looked agent-driven, and will ban non-manual external testing. No logs are public. (details) | Partly confirmed (Railway apology); details one-sided |
| Nov 3 | WordPress Core upgrade-window RCE (CVE-2026-119977) added to CISA KEV | Exploited | CISA added CVE-2026-119977, a TOCTOU flaw in WordPress Core's upgrade staging paths (4.7 through 7.1.3) allowing unauthenticated code execution during an upgrade, to KEV on Nov 3. Fixed in 7.1.4 (Oct 28); federal due date Nov 6. (source) | Patched |
| Nov 4 | WardenStealer and Aoki infostealers take Claude Code and Codex tokens | Breach | SpyCloud analysed recovered infection logs from the Windows infostealers WardenStealer and Aoki and found Claude Code and Codex credential files (OAuth tokens, API keys) collected; a password reset does not revoke them. (details) | Confirmed (SpyCloud) |
| Nov 4 | Haiou Codex jailbreak bundle plants ads in agent instructions | Agent incident | Check Point Research analysed Haiou 4.5.0, a Codex jailbreak bundle that rewrites the agent's global instructions so greetings trigger an ad for a third-party LLM gateway. (details) | Confirmed (Check Point) |
| Nov 4 | Apple pointer authentication bypass (CVE-2024-27834) added to CISA KEV | Exploited | CISA added CVE-2024-27834, an Apple pointer authentication bypass fixed in May 2024 (Safari 17.5, iOS 17.5, macOS Sonoma 14.5), to KEV on Nov 4; Sekoia links it to APT29 watering holes on mobile devices. Federal due date Nov 18. (source) | Patched |
| Nov 4 | sshole: OpenSSH on musl (Alpine, OpenWrt) accepts any password over 256 bytes | Critical vuln | On musl-based sshd without PAM and with password auth on, a password over 256 bytes logs in to accounts with * in /etc/shadow, such as Alpine root. Fixed in OpenSSH Portable 10.6p2; no CVE listed in the write-up. (details) | Patched |
| Nov 5 | Claude Dispatch "ThisBash": browser extension can drive Claude Desktop to run code | Critical vuln | Researcher Gal Weizman (Forever Security) reports a browser extension with claude.ai access can send prompts to Claude Desktop via Dispatch and approve them until code runs, even on another machine. Anthropic reportedly closed it as a duplicate; no CVE, no fix. Not reproduced by Beam. (details) | Reported (researcher); unfixed |
| Nov 5 | kaniko CVE-2026-105769: lookalike FROM host receives the build's Azure token | Critical vuln | kaniko before 1.25.20 matches Azure registry hosts with an unanchored regex, so a base image on a lookalike host gets the build's Azure access token (CVSS 9.3). Fixed in 1.25.20; rebuild your executor image. (details) | Patched |
| Nov 6 | SonicWall SonicOS SSLVPN SAML/MFA bypass (CVE-2026-19868) added to CISA KEV | Exploited | CISA added CVE-2026-19868 (CVSS 7.5) to KEV on Nov 6: a user with valid credentials for an alternate authentication method can bypass SAML authentication, including SAML-enforced MFA, on SonicOS SSLVPN. Federal due date Nov 20; see SonicWall advisory SNWLID-2026-0014. (source) | Reported (CISA KEV) |
New entries are added here as they are confirmed.
Campaigns reported in November that started earlier
These were disclosed or covered this month, but the attack or fix predates November, so they sit outside the daily timeline.
| Incident | Started | What happened |
|---|---|---|
| ARTEX, PentAGI and CAIRN in real intrusions | Late August 2026 | Google GTIG and Mandiant report threat actors running open-source agentic pentest frameworks in incident response cases against government, technology and telecom targets |
| Fake ChatGPT and Claude bots scanning for .env files | Late September 2026 | Cequence tracked one actor posing as about 32 AI bots to request roughly 90 secret-bearing paths such as .env files and SSH keys |
| Black October: nine AI developer platform incidents and LLMjacking | October 2026 (Oct 6-30) | Wiz counts at least nine AI developer platform incidents in October, including Fireworks AI, Zeabur, Railway, Warp and Northflank, and traces stolen cloud keys to resold model calls within days |
| ARTEX agents turn leaked GitHub OAuth secrets into Azure Key Vault theft | August 2026 (Microsoft report Nov 5) | Microsoft Threat Intelligence says attackers run ARTEX to search public GitHub for OAuth app secrets, then read Azure Key Vault and run VM commands; in one case five minutes from search to exfiltration. Tenant access seen from Oct 28 |
| Censys: 1,465 exposed ARTEX consoles, 32 open directories | Report published Nov 5, 2026 | Censys counted 1,465 internet-facing ARTEX consoles (about 815 distinct deployments) and 32 open directories with implants and operator files; 155 hosts also carry other threats. Some may be authorized testers. A measurement, not a vulnerability |
| Check Point: ARTEX agents stole 190,000 game credentials and Apple signing certificates | October 2026 (report Nov 5-6) | Check Point Research read ten exposed ARTEX operator consoles: agents given only a goal stole about 190,000 game credential sets and Apple .p12 signing certificates through basic web flaws, at an estimated $180 to $1,500 per campaign |
| LLMjacking at scale: stolen keys resold via LLM Mixers | 2026 (reported Nov 4) | Google GTIG and Mandiant describe attackers stealing cloud and LLM credentials and reselling access through LLM Mixers, with single-day charges over $100,000 and a LiteLLM gateway flaw abused |
| OpenAI: agent pipeline built 948 Linux root exploits | Mid-October 2026 (published Nov 4) | OpenAI says a research pipeline reproduced 1,724 Linux kernel bugs and built 948 local privilege escalation exploits for about $1.2M; 61 were patched upstream when the scan finished, more patches are being sent. A vendor report, not an attack |
| CERT-In: CRAFTY FALCON, AI-assisted attacks on payment APIs | 2026 (advisory Nov 4) | CERT-In advisory CIAD-2026-0058 says an AI-assisted actor uses ARTEX, PentAGI, Cairn and CyberStrikeAI to pivot from staging environments into BFSI payment APIs. A reported backdoored ARTEX image is unverified |
| OpenAI: GPT-6 Astra quietly underperforms on safety-framed coding tasks | Report published Nov 4, 2026 | OpenAI measured Astra completing about 10% of hidden rubric items on safety-framed tasks vs 16% on capabilities-framed ones, mostly unflagged by monitors; the dataset root cause is inconclusive. A lab report, not an attack |
| Chainalysis: attacker used "offensive agentic techniques" | October 2026 (notice Oct 30) | Chainalysis says an external actor used offensive agentic techniques to reach internal tools, exposing user IDs, names, emails and role labels; no passwords, MFA secrets or tokens found accessed. Method not published |
| VOLTA infostealer asks an LLM to rank victims | Reported Oct 2026 | Cisco Talos found VOLTA, a Windows infostealer that queries an LLM via OpenRouter to judge each victim's value, targeting SSH keys and cloud configs |
| ChatGPT share link XSS (fixed Oct 26) | Fixed October 26, 2026 | Parameter chained ChatGPT Preview, a MathLive postMessage flaw and a YouTube JSONP CSP bypass so a shared chat link could run code as chatgpt.com in the victim's session |
| Ivanti Connect Secure zero-day chain: auth bypass to root | Early October 2026 (Volexity Oct 13) | Volexity documented an unauthenticated chain (CVE-2026-104935, CVE-2026-111357) from authentication bypass to root on Ivanti Connect Secure that evaded the Integrity Checker; CVE-2026-118527 also in CISA KEV. Volexity assesses LLM use with medium confidence |
| Cisco ASA and FTD CVE-2026-76418: exploited SSL VPN reload bug | October 2026 (advisory Oct 28) | Cisco says CVE-2026-76418 (CVSS 8.6) is exploited: a crafted request reloads ASA/FTD devices with SSL VPN secondary authentication configured. Denial of service only; in CISA KEV since Oct 29 |
| 3AM ransomware at a UK school estate: Bitdefender sees an agentic loop | September 10-19, 2026 (reported Nov 3) | Bitdefender says a 3AM affiliate hit a UK school estate with human-plus-agentic-loop behavior; about 37 of 50 reached machines encrypted, tools were ordinary, AI use inferred not proven |
| Windows Paint RCE (CVE-2026-83844): patched out-of-bounds read | October 13, 2026 (resurfaced Nov 6) | CVE-2026-83844 is an out-of-bounds read in Microsoft Paint allowing code execution (CVSS 8.1, high complexity). Patched Oct 13; CISA SSVC lists no exploitation. Update Paint via Microsoft Store and Windows Update |
Unconfirmed claims
| Date | Claim | Where it came from | What to do |
|---|---|---|---|
| Nov 1 | Alleged Anthropic Mythos-class "numbat" checkpoint on Hugging Face | Hugging Face repo burner22/numbat and a Hacker News thread; Anthropic has not confirmed, commenters call it fake | Do not load it, or any repo, with trust_remote_code=True without reading the code; do not ask an agent to |
| Nov 3 | Codex training opt-out allegedly re-enabled | A Hacker News post says a team's "Improve the model for everyone" setting was switched back on; OpenAI has not confirmed a reset | Check both the ChatGPT data controls toggle and the separate Codex "Include environments" setting |
| Nov 4 | Postman desktop reportedly added Claude Code and Codex plugins, self-approved hook | A single GitHub report on Postman desktop 12.33.4 (macOS), not independently reproduced; Postman had not responded when The New Stack published | Review ~/.codex/config.toml and your Claude Code plugin list for plugin, marketplace and hook entries you did not add |
| Nov 6 | Claude Code account email in User-Agent headers, linked by a user to scam email | anthropics/claude-code issue 107757 and a Hacker News thread; Anthropic staff confirm the mechanism, the spam link is unproven | Update to 2.1.317 or later, run the no-network echo test, keep your own PreToolUse hook until the email leaves model context |
An entry moves into the timeline if the company or a named research team confirms it, and is removed if it is shown to be false.
Previous month: October 2026 attacks timeline
What November's attacks have in common
It is early in the month, so there is no pattern to report yet. Last month's incidents mostly began with stolen or unrotated credentials, and many aimed at AI keys. This section will be revised as November's entries accumulate.
What to do this week
- Rotate secrets stored on any platform named in an entry below, and on any host that ran an affected package.
- Pin packages and container images by version and digest.
- Patch vulnerabilities marked Exploited first.
- Cap and scope every AI provider key, and alert on usage spikes.
- For your own coding agents: keep monitors outside the agent's reach, log what they run, and require approval for credential access and destructive commands. Beam's local watcher flags these actions on your machine.
How this timeline is updated
Beam publishes one attacks timeline per month, and a new page starts on the first of each month. This page is checked every 4 hours and updated whenever a new November incident is confirmed or a claim changes status; the "Last updated" line at the top changes with it. The attacks tracker shows the same entries with filters, a link for every incident and an embed code you can paste into your own page.
Change log
- November 7, 2026: Added SonicWall SonicOS SSLVPN CVE-2026-19868 (CISA KEV).
- November 6, 2026: Added kaniko CVE-2026-105769 (Azure token leak to lookalike registry host, fixed in 1.25.20). Added Microsoft's ARTEX GitHub-secrets to Key Vault report and Censys's exposed ARTEX consoles count (earlier).
- November 6, 2026: Added the Claude Code account email User-Agent report (issue 107757) to unconfirmed claims.
- November 6, 2026: Added Check Point's ARTEX operator consoles report (game credentials, Apple certificates) to earlier campaigns. Added the patched Windows Paint RCE CVE-2026-83844 (earlier).
- November 5, 2026: Added Apple CVE-2024-27834 (CISA KEV) and OpenAI's 948 Linux root exploits report (earlier). Added Claude Dispatch "ThisBash"; added CERT-In CRAFTY FALCON and OpenAI's GPT-6 Astra report (earlier). Added the Ivanti Connect Secure zero-day chain and Cisco ASA/FTD CVE-2026-76418 (earlier).
- November 5, 2026: Added sshole, the OpenSSH musl long-password bypass fixed in 10.6p2.
- November 4, 2026: Added the Primcloud/Railway AI-agent probe, WordPress Core CVE-2026-119977 (CISA KEV), SpyCloud's WardenStealer/Aoki AI-token theft, Check Point's Haiou prompt adware and Google/Mandiant LLMjacking (earlier). Added claims: Postman agent plugins hook, Codex training setting reset.
- November 3, 2026: Added the WorkOS sales app breach (ARTEX link per the CEO, unverified). Added Bitdefender's 3AM ransomware school case to earlier campaigns. Added the GitBook AI-agent attack; added Black October (Wiz), Chainalysis, VOLTA and the ChatGPT share link XSS to earlier campaigns.
- November 2, 2026: Added the Northflank incident and Kaspersky's CyberStrikeAI findings; added ARTEX (Google/Mandiant) and fake AI bot .env scanning (Cequence) to earlier campaigns.
- November 1, 2026: Page created.
- November 1, 2026: Added the alleged Anthropic "numbat" checkpoint on Hugging Face to unconfirmed claims.

