Anthropic Embeds Independent Safety Evaluators as OpenAI’s Sam Altman Pledges to Follow
Summary
Anthropic embeds independent safety evaluators inside its operations with badges, desks, laptops and internal-level access, and OpenAI CEO Sam Altman says OpenAI will follow Dario Amodei’s model.
Key Points
- Anthropic CEO Dario Amodei commits to embedding third-party safety evaluators at the company to pace frontier AI development, and OpenAI CEO Sam Altman says OpenAI will follow suit.
- Amodei says embedded evaluators such as METR staff will receive company badges, desks, laptops and access comparable to internal risk-assessment teams, while verifying commitments and reporting safety incidents.
- He urges U.S.-mediated safety coordination among democratic AI companies and tighter chip-export and model-distillation controls to widen America’s AI lead over China within three to five years.