Beyond METR: 20 AI safety evaluators working with AI labs, from Apollo to Redwood
kevin_klyman · x · 2026-09-15
Responding to criticism of METR, the author argues that METR is just one of many AI evaluators, and objections to METR don't undermine embedded audit requirements. Lists 20 evaluators working with AI labs: Transluce, Grey Swan, Apollo, AVERI, SecureBio, Faculty, Vaultis, Dreadnode, Irregular, RAND, Far AI, ActiveFence (now Alice), Active Site, Deloitte (via Gryphon), Nemesys, Mercor (via Sepal), AE Studio, Scale, Frontier Design, Redwood.
More from Safety
- Ro Khanna backs bill banning artificial superintelligence, calls for federal AI regulator — ShakeelHashim · 2026-09-15
- Houthi militants reportedly used Claude Code to vibecode missile software — JohnnyNi13 · 2026-09-15
- Altman hints at AI-safety pact with rivals; Amodei pledges permanent insider access for independent evaluators — fortune · 2026-09-15
- Anthropic co-founder says mandatory AI 'kill switch' may be needed — HomeBrewUser · 2026-09-15
- New European Transformative AI Strategy warns inaction risks Europe's prosperity and sovereignty — arthurmensch · 2026-09-15
- Senator Rick Scott urges frontier AI slowdown while warning US can't cede lead to China — TheZvi · 2026-09-15