OpenAI Revokes Access to Trusted Access for Cyber Program
August 19, 2026
OpenAI has reportedly revoked researcher access to its limited cyber program. The program was designed to allow trusted defenders to use advanced models for vulnerability reporting and patching.
HOW THIS AFFECTS YOU
●
researcherYour ability to test model safety in specialized domains may be restricted.
●
policyThis move affects how security researchers interact with frontier models.