JadidJust-in AI Developments, Incidents & Deadlines
November 2026
SMTWTFS123456789101112131415161718192021222324252627282930

The AI stack's weakest points today are people and credentials, not models.

20 stories · transcript

Nvidia's board met overnight on whether to remove Jensen Huang, and six analysts cut the stock; separately, 4,837 hijacked GitHub accounts planted a secret stealer in at least 1,806 repositories. Beneath that, a critical Apache HTTP/2 flaw, MongoDB and Node.js patches landed, and Google shipped a model that decides instead of generating.

TLDR

  • Nvidia's board meets on removing CEO Jensen Huang after a second affair report; no outcome announced (Nov 11-12). Business Policy
  • Hijacked GitHub accounts plant a CI secret stealer in at least 1,806 repositories (Nov 12). Security Open Source
  • Apache HTTP Server fixes a critical mod_http2 remote code executionRemote code executionA flaw that lets an attacker run malicious code on a target's computers or network.Cloudflare ↗ flaw, CVE-2026-119618 (Nov 11). Security Infra
  • Google launches Gemini Decisions API: 1,000 questions scored in one pass at $0.10 per million input tokensA small unit of data, such as part of a word, that AI models read and generate. Many AI services price by tokens in and out.NVIDIA ↗ (Nov 11). Agents Pricing
  • Commerce reportedly finalizes a rule blocking Chinese AI firms from remote access to advanced chips abroad (Nov 12). Policy Infra

Upcoming Deadlines

Date What Who it hits Do this Source
Thu Nov 12 Rust 1.100 stabilizes Cargo minimum publish age (dependency cooldown, off by default; Socket, Oct 20) Rust teams Set a cooldown in Cargo config Socket
~Thu Nov 12 OpenAI's proposed end of model access in Cursor (official date not confirmed; Cursor may end it sooner) Cursor users on OpenAI models Check Cursor terms OpenAI
Mon Nov 16 GitHub Copilot retires Claude Opus 4.8, Claude Opus 5, GPT-5.6 Terra (replacements Opus 5.5, GPT-6.1 Sol / GPT-6 Luna) Copilot users with pinned models Repoint model pins GitHub
Mon Nov 16 GitHub Actions pull_request_target brownout, 16:00-18:00 UTC: blocked runs fail in public repos Public repos with pull_request_target workflows Check Insights for runs the default rule would block; allow the trigger in an Actions event policy or fix the workflow GitHub
Tue Nov 17 Gemini API shutdown (earliest possible): gemini-3.1-flash-tts-preview, gemini-3.1-flash-live-preview, gemini-2.5-flash-native-audio-preview-12-2025, gemini-2.5-flash/pro TTS previews TTS and Live API callers Move to gemini-3.8-live or gemini-3.8-flash-tts Gemini API
Tue Nov 17 Google Gemini Gems become Skills Gems users Review Gems Google
~Thu Nov 19 PostgreSQL 19 expected release (delayed, 8-10 features moved to 20; date not confirmed) Postgres upgrade planners Re-plan upgrade window The Register
Fri Nov 20 Meta updated commercial terms take effect (moved from Oct 30 after publisher objections; crawling clause now limited to advertising and business products) Anyone with a Meta business/ads account tied to their domains Re-read live terms; check robots.txt and staging logins TechCrunch · Meta
Fri Nov 20 Conductor Cloud shuts down after Meta acquisition (desktop app stops receiving new models in early 2027) Teams running parallel coding agents on Conductor Export work and move to another agent orchestrator The Decoder
Fri Nov 20 CISACybersecurity and Infrastructure Security AgencyThe U.S. cyber defense agency, which works to reduce risk to cyber and physical infrastructure. It runs KEV.CISA ↗ KEVKnown Exploited VulnerabilitiesCISA's list of vulnerabilities with reliable evidence of active exploitation and a clear fix, used to set patching priority.CISA ↗ federal patch deadline for SonicWall SonicOS SSL-VPN SAML bypass CVE-2026-19868 (exploited) Anyone running Gen7 SonicWall firewalls (TZ, NSa, NSsp, NSv) with SAML SSL-VPN login Upgrade SonicOS to the fixed release; review SSL-VPN logins for the SAML path BleepingComputer
Wed Nov 25 Qwen4-Max-Preview access on Alibaba Token Plan ends (1M-token context preview) Teams testing qwen4-max-preview Finish evaluations before Nov 25 Pandaily

Action Items

  1. Search every GitHub org you own for pull requests titled "docs: dependency audit notes" from branch wfx-docs, then rotate all Actions secrets, cloud roles and AI-tool tokens in any repo where one appears, today: Aikido says the stealer deletes its own run logs, so a skipped run number may be the only trace.
  2. Upgrade Apache HTTP Server to 2.4.71 (or your distribution's backport) on every host with h2 or h2c in a Protocols directive, including per-virtual-host settings, before the end of the week: CVE-2026-119618 is unauthenticated and rated 9.8.
  3. Move Node.js to 22.23.5, 24.22.3 or 26.13.1 and MongoDB replica sets to the fixed release for your branch: the Node fixes cover HTTP/2 crashes and a TLS-key memory leak, and CVE-2026-118383 is rated 9.5.
  4. Turn off N-central's Legacy API on any on-premises server exposed to the internet, and audit your ScreenConnect instances for the five extensions ConnectWise disabled, before the next MSP maintenance window.
  5. Rotate any Notion AI or Fireworks-hosted credentials you set up before Oct 24, and ask every inferenceRunning a trained model on new input to get an output, as opposed to training it.NVIDIA ↗ vendor for its incident-notification terms by the end of the month.

Models & Tools

  • Google launches the Gemini Decisions API (Nov 11) Agents Pricing. Gemini Decision 1 scores up to 1,000 independent closed-ended questions in a single forward pass over text, images, audio, video and PDFs, at 150-250 ms and $0.10 per million input tokens with no output billed. It targets tool selection, policy checks and model routing. Google's benchmark leads are its own. (Google | Simon Willison)
  • AWS Emulator, built and maintained by agents, runs AWS services locally (Nov 11) Agents Open Source. AWS says agents wrote the whole emulator and will maintain it, and it released the code under CC0 because it was unsure agent-written work can be copyrighted. Tests can run against it without handing credentials to a coding agent. (The New Stack | Help Net Security)
  • Alibaba opens a Qwen4-Max preview with a 1M-token context (Nov 11) Open Source Agents. Model ID qwen4-max-preview is available only on Alibaba's Token Plan until Nov 25, with tool calling and image and video input. No benchmarks or parameter counts were published. (Pandaily | TechNode)
  • Google splits its TPUTensor Processing UnitGoogle's custom chip built to speed up machine learning, especially large matrix operations.Google Cloud ↗ line again and open-sources TPU Sync (Nov 12) Infra. TPU 8tx targets confidential inference with AMD EPYC hosts; 8i Max pairs two 8i chips for long-context agents. Google's own tests show TPU Sync cutting KV-cache transfer from seconds to about 230 ms. APIs are changing rapidly, so pin the commit. (Tom's Hardware | InfoQ)

Security

  • wfx-loot: hijacked GitHub accounts plant a CI secret stealer (Nov 12) Security Open Source. Critical · CVSSCommon Vulnerability Scoring SystemA standard 0 to 10 severity score: 0.1–3.9 low, 4.0–6.9 medium, 7.0–8.9 high, 9.0–10 critical.FIRST ↗: not applicable (no CVECommon Vulnerabilities and ExposuresA public catalog that gives each known security vulnerability a unique ID, a short description and at least one reference.NIST ↗) · Exploited in the wild · No patch; rotate secrets. Aikido counts 4,837 compromised accounts opening probe pull requests in 61,764 repositories; the stealer ran in at least 1,806 and took Actions secrets, cloud keys and AI-tool tokens. SafeDep counts 8,093 repositories with malicious commits. Scanners will be silent. (Help Net Security | The Hacker News)
  • Apache HTTP Server mod_http2 remote code execution (CVE-2026-119618) (Nov 11) Security Infra. Critical · CVSS 9.8 · Not exploited, per Wiz · Fixed in 2.4.71. A race condition in assign_headers() corrupts heap memory without authentication on versions 2.4.55 through 2.4.70 with HTTP/2 on. Wiz says 64% of cloud environments hold exposed resources; Red Hat says public research shows a reliable exploit, while Wiz saw none. (The Hacker News | Wiz)
  • MongoDB pre-auth use-after-free can crash replica set members (CVE-2026-118383) (Nov 12) Security Data. High · CVSS 9.5 (CVSS 4.0) · Not exploited, per MongoDB · Fixed releases for each branch. Two unauthenticated requests plant the fault; the crash comes only when the member later becomes primary while lagging. Authentication does not block it, and standalone servers are unaffected. (The Hacker News)
  • Node.js fixes 16 flaws, including HTTP/2 crashes and a TLS key leak (Nov 12) Security Open Source. High · CVSS: not stated · Exploitation unknown · Fixed in 22.23.5, 24.22.3 and 26.13.1. CVE-2026-123863 crashes default HTTP/2 servers on 24.x and 26.x; CVE-2026-123864 can leak TLS private keys from HTTPS servers that call res.flushHeaders() early. (The Hacker News)
  • New N-central exploits work against the patched build (Nov 12) Security. Critical · CVSS: not yet assigned · Exploitation unknown · Mitigation: disable Legacy API. Update — Researcher Sean Bales published root remote code execution (4EverAble2Shell) and Agent2Shell exploits for 2026.4.1.26. N-able says it needs the non-default Legacy API exposed to the internet; hosted environments are not affected. (The Hacker News)
  • ScreenConnect campaign hits at least 288 endpoints across 48 organizations (Nov 12) Security. High · CVSS: not assigned · Exploited in the wild · ConnectWise disabled five third-party extensions. ConnectWise says it found no vulnerability in ScreenConnect and that about 400 partner environments run at least one of the extensions. Initial access remains under investigation. (The Hacker News | SecurityWeek)
  • Red Hat Quay key-server flaw lets anonymous users take over token signing (CVE-2026-123710) (Nov 12) Security Infra. Critical · CVSS 9.8 · Exploitation unknown · Apply Red Hat's advisory. Deployments where the signing-key service is reachable by untrusted clients are exposed to read and write access on every repository. (SecurityWeek)
  • Notion discloses that the Fireworks AI breach exposed AI chat transcripts (Nov 11) Security Customer AI. Medium · CVSS: not applicable · Exploited in the wild · Customers notified. Update — Fewer than 1% of Notion workspaces were affected and most transcripts cannot be tied to a customer. Fireworks told Notion on Nov 2 about the Oct 24 intrusion. (SecurityWeek | BleepingComputer)
  • Darktrace turns Claude Code into a remote-controlled implant in a lab test (Nov 12) Security Agents. Medium · CVSS: not applicable (no vulnerability) · Not exploited, per Darktrace · No patch; protect API keys. With stolen credentials, an attacker issues jobs through memory stores and pulls a 1 GB file via the Files API, all through api.anthropic.com traffic that looks routine. (Help Net Security)

Industry & Business

  • Nvidia's board weighs removing Jensen Huang after a second affair report (Nov 11-12) Business Policy. Update — CNBC confirmed the board met Wednesday; no decision was announced by Thursday morning. The New York Times reported a second relationship, with a direct report. Six analysts downgraded; Bernstein halved its target to $200 and the stock fell about 8% premarket. Analysts told CNBC removal could cost another 20%. (CNBC | TechCrunch)
  • Anthropic tells investors AI cut an AMD port from months to three days (Nov 11) Business Infra. In an investor presentation, co-founder Tom Brown says Claude trains and runs on all four major chip families and that Mythos wrote nearly all the AMD kernels. These are company claims, offered as the case that CUDA lock-in is weakening. (The Decoder)
  • UK chip startup Fractile raises $720M (Nov 11) Business Infra. The Series C was led by Thrive, Founders Fund, Lightspeed and Redpoint with Britain's Sovereign AI Fund; the government cites a valuation above £5 billion, reportedly near $7 billion. Fractile pairs inference chips with its own compiler. (TechCrunch)

Policy & Safety

  • Commerce reportedly finalizes its AI remote-access rule (Nov 12) Policy Infra. The Information, citing two people familiar, says the rule could land this month and would extend chip controls to seven more countries, including Malaysia, Singapore and Thailand, to stop Chinese firms renting restricted chips offshore. Cloud teams with Asian regions should expect new customer screening. (The Information)
  • Australia proposes mandatory safety rules for frontier AI labs (Nov 12) Policy. The bill would require independent audits, published safety frameworks and immediate security-incident reports, enforced by a new oversight body beside the Australian AI Safety Institute. It is a proposal, not law. (The Decoder)

AI in Production

  • Microsoft's MDASH scan of the Linux kernel yields 59 reports and 33 CVEs (Nov 11) Research Security. One researcher and an intern turned 26,502 raw findings into 59 upstream reports; eight CVEs have working root escalation exploits and 38 fixes are merged. Triage took months, which is the cost behind the volume. (Microsoft)
  • DeepMind's AlphaProof Code verifies Signal's cryptographic core in four weeks (Nov 11) Research Security. Agents wrote over six million lines of Lean proof and flagged about 1% of the code for human review, and both planted bugs were caught. The result is DeepMind's own, with Signal's CTO endorsing it. (The Decoder)

Type to search · ↑↓ to move · enter to open