METR publishes independent assessment of Anthropic's AI R&D acceleration, evidence heavily redacted
rohanpaul_ai · x · 2026-11-06
METR released its independent assessment of AI R&D acceleration at Anthropic (October 2026) on Nov 5, 2026. While top-level risk conclusions were shared, Anthropic exercised redaction rights and removed most of the evidence METR found most informative, citing confidential info that could "accelerate other AI developers." METR notes much of the redacted evidence had multiple plausible interpretations, and both METR and Anthropic researchers have significant uncertainty about how to synthesize it — making it hard for the public to judge. A rare third-party risk assessment of a frontier lab, and the extent of redaction is itself notable.
More from Safety
- Bug bounty hunters filed tens of millions of vulns, big firms share maybe 1% — rez0__ · 2026-11-06
- LSEG cuts 5-year market backtest from hours to ~1 minute; Tampa General adds AI safety gate — NVIDIAAI · 2026-11-06
- Tampa General post-trains NVIDIA Nemotron to screen AI calls for prompt injection — NVIDIAAI · 2026-11-06
- DeepMind's 'Keeping Defenders Ahead' Page Displays Root CA Passwords Everywhere — m4rkmc · 2026-11-06
- Anthropic urged regulation then balked at real commitments, says policy expert — peterwildeford · 2026-11-06
- OpenAI Drops NDAs for Data Center Projects After Midterm Backlash — The Verge AI · 2026-11-06