Amazon: One Prompt Let Hackers Root an AI Agent in 11 Minutes
terryyuezhuo · x · 2026-11-21
Amazon says suspected China-linked hackers gave its Cairn AI agents a single prompt — "obtain shell access" — and reached root in 11 minutes via a MONAI Label zero-day. Despite reports of a Nov. 18 fix, GitHub advisory GHSA-qc22-rmgc-vg94 shows PyPI still ships the vulnerable 0.8.5 version.
More from Safety
- EGS: a tiered guardrail framework that stops educational LLMs from over-refusing — kalyan_kpl · 2026-11-21
- Researchers warn even eminent minds underestimate AI's exponential pace, risking policy missteps — satnam6502 · 2026-11-21
- Author accuses Anthropic of 'stealing' her work, calls AI firms 'contemporary bank robbers' — GaryMarcus · 2026-11-21
- Cohere launches confidential computing with encrypted inference; exec says rivals' ZDR labels are marketing — Nils_Reimers · 2026-11-21
- Can TEEs and Remote Attestation Make Decentralized GPUs Safe for Sensitive AI Workloads? — MrSwannn · 2026-11-21
- Report: Nearly 5 OpenAI Staff Left in 14 Hours Amid Safety Policy Shake-up — GaryMarcus · 2026-11-21