METR publishes independent review of Anthropic's August 2026 Risk Report misalignment section
connoraxiotes · x · 2026-10-21
METR (Nikola Jurkovic, Tim Hua, Beth Barnes) released an external review of the "Misalignment in high-stakes settings" section of Anthropic's August 2026 Risk Report, which had concluded the risk of catastrophic harm from misalignment was low as of July 15, 2026.
- Anthropic held veto/redaction rights but made no redactions and did not change METR's conclusions; edits were limited to wording, clarifications, and one experiment's methodology.
- The review was lightweight and predates Anthropic's "independent evaluators" announcement, with more limited access.
- METR recommends Anthropic consider risks from AI systems undermining safety-relevant decisions within the company in future reports.
Related event: METR Reviews Anthropic Risk Report: Agrees on Low Risk, Faults Methods(4 posts)→
More from Safety
- Package manager outage forces skipping release-age checks, ideal for supply chain attacks — sloppenheimer · 2026-10-21
- Anthropic ran Claude autonomously over 25 OSS codebases and found 493 attack-surface issues — chrisrohlf · 2026-10-21
- Same Text, Opposite Verdicts: DIY Test Exposes Unreliable AI Detectors — CommonExplanation711 · 2026-10-21
- Claudini: AI autoresearch discovers SOTA jailbreak algorithms with 80% ASR — maksym_andr · 2026-10-21
- How are you building your enterprise AI agent security stack for 2026? — Medical-Analyst2671 · 2026-10-21
- ChatGPT banned a user writing an abuse memoir under 'Child Sexualization' policy — throwRAthe1 · 2026-10-21