2 articles
- 5d agoApollo ResearchEarliestApollo Research
- 5d agoApollo Researchhttps://apolloresearch.ai/blog/pilot-towards-embedded-evaluations
Why it’s here: This item covers an embedded AI alignment evaluation pilot with OpenAI, including access, disclosure, and meta-transparency rules that bear directly on AI oversight policy and how evaluators can be given government-relevant assurance of frontier model safety.
Trace across the board