FRONTIER · GLOBAL
Over 100 AI experts urge independent safety evaluators for Anthropic, OpenAI models
A coalition of 100+ AI researchers and safety experts issued a public letter demanding that Anthropic, OpenAI, and other foundation model labs use truly independent evaluators—not in-house teams—to assess safety risks before release.
WHY IT MATTERS
Enterprise procurement teams will face pressure to demand third-party AI audits and safety certifications; compliance frameworks may require independent eval reports as condition of use.
Source: CNBC · 2026-09-18