OpenAI Technical Issue Revokes Access for Cybersecurity Researchers

OpenAI has revoked the access of several cybersecurity researchers to the Trusted Access for Cyber (TAC) program due to a confirmed technical error. - TAC Program Revocation: Researchers reported losing access on OpenAI forums and X, receiving messages about failed identity verification or ineligibility. - Program Purpose: TAC provides selected researchers access to advanced AI models with reduced cybersecurity restrictions to aid in discovering bugs and vulnerabilities before they can be exploited by malicious actors. - Technical Error Confirmed: TechCrunch spoke to five affected researchers who received emails from OpenAI acknowledging a technical issue, and they were asked to reapply and complete verification. - Regional Impact: All researchers affected were based outside the U.S. and Europe, suggesting a potential regional limit to these revocations. - Daybreak Blue Access: The revoked access pertained to Daybreak Blue, which allows use of frontier general-purpose models like GPT‑5.6 Sol for defensive cybersecurity tasks. - Higher Access Tiers: OpenAI also offers Daybreak Red, a tier with access to models specifically designed for cybersecurity research. - Community Concerns: Recently, researchers have critiqued AI firms like OpenAI for imposing too strict cybersecurity guardrails, which they claim hinder legitimate research efforts.