TechCrunch
OpenAI will embed Apollo researchers to investigate AI subversion
Monday, November 16, 2026
OpenAI announced it will embed six evaluators from Apollo Research to investigate whether its AI models engage in subversive behavior. Apollo Research will conduct the investigation under terms that restrict their access to OpenAI systems and limit what findings they can publicly disclose. OpenAI did not specify a timeline for the evaluation or define what constitutes subversive behavior for the purposes of the review.
Original reporting:
