OpenAI Revokes Cyber Research Access for Vetted Security Researchers

OpenAI has unexpectedly suspended access to its Trusted Access for Cyber program for a number of vetted security researchers, prompting concerns over a technical malfunction within the company. Affected individuals reported receiving error messages indicating their accounts were ineligible or unable to verify identity when attempting to access ChatGPT’s cybersecurity interface. The incident primarily impacts researchers operating outside the United States and Europe, though the exact scope remains unconfirmed. OpenAI attributed the disruption to a recent technical issue, acknowledging the error in official correspondence and directing impacted users to reapply through the standard verification process. The Trusted Access for Cyber program was designed to provide vetted cybersecurity professionals with access to OpenAI’s most advanced frontier models, including the recently launched Daybreak Blue tier featuring GPT-5.6 Sol, and the specialized Daybreak Red tier. These tiers intentionally reduce standard safety guardrails to facilitate legitimate defensive security work, such as vulnerability discovery, secure code review, malware analysis, and exploit validation. The program operates similarly to Anthropic’s Cyber Verification Program, reflecting an industry-wide effort to equip trusted defenders with more capable AI tools while preventing malicious actors from leveraging unfiltered models for offensive operations. The sudden access revocations arrive amid growing frustration within the security community regarding the restrictive nature of AI safety protocols. Both defensive and offensive researchers have repeatedly criticized OpenAI and Anthropic for implementing guardrails that inadvertently hinder authorized vulnerability testing and bug reporting. By limiting model accessibility during routine maintenance or technical errors, companies risk disrupting critical security workflows and delaying the identification of high-severity flaws. OpenAI’s decision to request reverification suggests the suspension was applied broadly rather than targeted, likely triggered by a backend authentication or compliance check malfunction. Researchers remain dependent on these vetted tiers to conduct advanced security assessments that standard ChatGPT interfaces cannot support. Until the technical issue is fully resolved and verification systems are stabilized, the incident underscores the operational fragility of specialized AI research programs and the delicate balance between enforcing safety protocols and maintaining uninterrupted access for legitimate cybersecurity professionals.