Researchers say OpenAI revoked their access to limited cyber program

OpenAI has been revoking access for some security researchers in its limited cyber program, and the company says the problem came from its side. The affected users were in Trusted Access for Cyber, or TAC, which gives vetted researchers access to OpenAI’s stronger models with fewer cybersecurity guardrails than regular ChatGPT users get.

The complaints started showing up on Wednesday, on OpenAI’s support forums and on X. Researchers said the Cyber page in ChatGPT told them their identity could not be verified, or that their account was “ineligible at this time.” TechCrunch spoke with five researchers who said they ran into the problem.

OpenAI told at least one researcher that access to Daybreak Blue, TAC’s newest vetted tier, had been revoked “due to a technical issue affecting a limited number of users.” Another researcher said support gave a similar explanation and asked people to reapply and go through verification again. OpenAI later pointed TechCrunch to a post saying a “limited set of users’ access to Daybreak Blue is no longer active and they will need to re-verify to maintain their access.”

Daybreak Blue launched on August 10 and is meant as the starting point for defenders doing things like vulnerability discovery, secure code review, malware analysis, incident response, and patch validation. OpenAI also offers a higher tier, Daybreak Red, for authorized vulnerability research, exploit validation, and security testing. The catch is obvious: these programs are trying to keep powerful models in the hands of trusted defenders, while keeping them away from people who want to use them for break-ins instead. The researchers who spoke to TechCrunch all said they live outside the U.S. and Europe, which may point to a regional glitch, but OpenAI hasn’t explained that part.