AI Radar

Security

Vulnerabilities, advisories and incidents in AI tools, models and agents from the last 14 days, newest first. Check the linked advisory before acting; evidence shows how well each report is confirmed.

  1. Nov 10

    GPUStack auth bypass grants unauthenticated admin (GHSA-gj9m-f8ph-pmhr)

    high evidenceheat 441 source

    CriticalCVSS 9.8CVE-2026-102895GHSA-gj9m-f8ph-pmhrgpustack (pip)Affected < 2.2.0 · Fixed in 2.2.0Advisory

    A GitHub advisory says GPUStack before v2.2.0 treats requests from a loopback source as the built-in administrator when force_auth_localhost is False, the default.

  2. Nov 10

    Claude Code Edit tool race allows writes outside workspace (GHSA-5xjx-9qx6-jcrv)

    high evidenceheat 401 source

    HighCVE-2026-88968GHSA-5xjx-9qx6-jcrv@anthropic-ai/claude-code (npm)Affected < 2.1.251 · Fixed in 2.1.251Advisory

    GitHub published an advisory for a time-of-check to time-of-use race in the Claude Code Edit tool.

  3. Nov 10

    Fireworks AI breach exposes Notion AI chat sessions

    high evidenceheat 553 sources

    Notion says a security incident at its vendor Fireworks AI may have exposed some Notion AI chat sessions.

  4. Nov 10

    GPUStack authentication flaw CVE-2026-102895 exploited in the wild

    high evidenceheat 451 source

    CVE-2026-102895GPUStackAdvisory

    CVE-2026-102895 is an improper authentication flaw in GPUStack, listed in CISA's known exploited vulnerabilities catalog.

  5. Nov 10

    Microsoft FORGE Lab finds Rust memory-safety bugs with agents

    high evidenceheat 401 source

    Microsoft's FORGE Lab published a post saying it uses agentic tools to find memory safety issues in Rust code.