Experts say in public letter Anthropic and OpenAI need truly independent safety evaluators

RishiBommasani · x · 2026-09-18

According to CNBC, experts behind a public letter argue that Anthropic and OpenAI need truly independent safety evaluators, pushing back against the labs' practice of self-evaluating their own frontier models' safety. The letter adds public pressure for an independent oversight mechanism in the AI industry.

Related event: Over 100 AI Experts Sign Open Letter Setting Minimum Standards for Embedded Third-Party AI Evaluation(6 posts)→

Original post →

More from Companies & People

• Kastle raises $24M Series A for AI bank employees after bold pivot before Demo Day — ycombinator · 2026-09-19
• AIxBio Hack brings together OpenAI, Nvidia and wilbe_science — mmbronstein · 2026-09-19
• Disney hires former Character.AI CEO Karandeep Anand as its first CTO — 12exyz · 2026-09-19
• Robot startup's career fair line tops 300 students, outdrawing Google and Amazon — MatthewChang · 2026-09-19
• Decagon Dialogues conference in SF Oct 1 with Cognition CEO and PayPal execs — kimberlywtan · 2026-09-19
• NVIDIA x Nebius Global AI Hackathon offers $50K+ prizes, $20K grand prize — demian_ai · 2026-09-19

Companies & People channel →