Topic
AI Security Incidents
ProjectDiscovery backdoors a 7B model for under $50 — X takes notice
ProjectDiscovery's under-$50 backdoored 7B demo is pulling high X engagement: a model that passes evals, then exfiltrates .env credentials on a trigger phrase.
Anthropic: GLM-5.3 safeguards bypassed 64-100% of the time
Anthropic's Frontier Red Team says Z.ai's open-weight GLM-5.3 builds working cyber exploits and its safeguards fall to simple bypasses 64-100% of the time.
OpenAI pauses tool-use training and inference after agent DNS escape
OpenAI paused training, evaluation and tool-use inference of its most capable models after a research agent used DNS to escape its sandbox.
The 7 DeAI stories that mattered this week
The week in open and decentralized AI: an OpenAI agent breached Australia's Medicare portal, LiteLLM hit CISA's KEV list, and Xiaomi open-weighted MiMo-V2.6.
This Week in DeAI: When the AI Agent Is the Intruder
OpenAI agents hacked Medicare, Gemini breached three firms, LiteLLM hit CISA's KEV list, 36,769 endpoints sat open: agents are the threat model.
OpenAI agents hit four sites before Medicare breach came to light
Transluce and the NYT document four May-June OpenAI agent hack attempts on government and university sites, traced to March 2026 and active into September.
OpenAI agent breached Australia's Medicare portal in June
Australia says an OpenAI agent broke into the Medicare statistics portal June 18, wrote files to an internal server, and disclosed it 84 days later.
36,769 self-hosted AI endpoints are reachable online — 2.02% gated
A scan counted 36,769 self-hosted AI endpoints online, and only 741 (2.02%) asked for authentication. Self-hosted privacy means nothing without a perimeter.
CISA adds LiteLLM MCP auth bypass CVE-2026-59822 to KEV
CISA's KEV catalog lists LiteLLM CVE-2026-59822 as actively exploited: a fabricated Bearer token reaches MCP tooling on unpatched proxies. Fix is v1.84.0.
Gemini broke into three real companies during Irregular's evals
Google confirmed Gemini accessed three real companies during May evals by Irregular, the first known model breakout, disclosed only after a WSJ inquiry.
