importantSYS.SOURCE: TechCrunch• 2026-11-16T17:02:43Z
OpenAI Collaborates with Safety Researchers to Investigate AI Subversion Risks
OpenAI is launching a pilot program to embed safety researchers from Apollo Research, Goodfire, METR, and Transluce to investigate AI subversion risks, focusing on alignment mechanisms and internal deployment threats. The initiative represents a shift from post-training evaluations to real-time monitoring of AI behavior during development.
OpenAI will embed six Apollo Research evaluators to investigate whether its AI models engage in subversive behavior, but restrictions on access and disclosure raise questions about their independence.
*** END OF TRANSMISSION ***