OpenAI's Trusted Access for Cyber (TAC) program experienced a technical failure this week that removed vetted security researchers from the system and blocked some from re-entering. Participants discovered their approved status had disappeared without warning, with the verification portal prompting them to start the process from scratch. Some researchers also lost access to Daybreak Blue, a higher access tier offering expanded capabilities through Codex Desktop and command-line interfaces.
The TAC program grants security professionals access to advanced AI capabilities that remain restricted for general users. Applicants must provide identification documents and complete verification before receiving access. The program aims to equip legitimate security defenders with powerful tools for vulnerability research while preventing misuse by malicious actors. Anthropic operates a similar initiative called the Cyber Verification Program.
OpenAI acknowledged the problem on its community forum, stating that a limited number of users lost Daybreak Blue access due to a technical issue. The company instructed affected users to watch for an email titled "Action required: Reverify your Daybreak Blue access" with restoration instructions. However, multiple researchers report that following these instructions led to rejection messages stating their accounts were ineligible for verification.
Some researchers who held Daybreak Blue access for over a month found themselves locked out after the glitch. When they attempted re-verification as instructed by OpenAI support, the system rejected their applications, claiming it could not verify their identity or that their accounts were ineligible. According to user reports, OpenAI support staff stated they could not reset verification status, restore previous approvals, or override the new rejections.
OpenAI has not disclosed the number of affected users, the root cause of the removals, or why previously approved researchers now face ineligibility determinations. The company also has not explained whether it can restore access when re-verification fails. Affected researchers should monitor their email for verification instructions and contact OpenAI support if re-verification attempts fail, though resolution timelines remain unclear.
Source: https://www.theregister.com/ai-and-ml/2026/08/20/openai-glitch-locks-out-vetted-cyber-researchers-and-some-cant-get-back-in/5290238