Anthropic and OpenAI propose embedding independent safety evaluators within their AI labs to improve oversight. Researchers welcome the unprecedented access but caution that real effectiveness depends on transparency, true independence, and eventually formal regulation.
Background
Anthropic and OpenAI are among the leading AI developers pushing for robust safety and oversight frameworks as AI capabilities advance rapidly. The push for internal safety evaluators reflects growing pressure from researchers and policymakers for greater accountability in AI development.
- Source
- TechCrunch
- Published
- Sep 17, 2026 at 05:07 AM
- Score
- 6.0 / 10