OpenAI Revokes Security Researchers' Access Nine Days After New Tier Launch

OpenAI revoked Trusted Access for Cyber (TAC) privileges for at least five security researchers on August 19, 2026, nine days after launching its Daybreak Blue tier. The tier gives vetted researchers access to frontier models like GPT-5.6 Sol with reduced guardrails for defensive security work. OpenAI cited a "technical issue" and told affected users to reapply and repeat verification.