OpenAI accidentally locked cybersecurity researchers out of its restricted AI program

Key points

• OpenAI confirmed a technical error revoked access for a limited number of users from its Trusted Access for Cyber (TAC) program, as first reported by TechCrunch.
• The affected tier is Daybreak Blue, launched 10 August 2026, giving vetted researchers access to advanced models including GPT-5.6 Sol.
• All five researchers who spoke to TechCrunch live outside the US and Europe, suggesting the glitch may be region-specific.
• OpenAI has asked affected researchers to reapply and complete identity verification again.
• A higher tier called Daybreak Red also exists, giving access to models built for offensive security testing.

Imagine spending weeks proving you're a trustworthy security researcher, passing ID checks, getting cleared by OpenAI, and then one morning your access simply vanishes. That is what happened to a number of cybersecurity professionals this week.

Multiple researchers reported on OpenAI's official support forums that when they opened ChatGPT's Cyber page, a message told them their identity couldn't be verified or their account was "ineligible at this time." OpenAI has confirmed it was their fault.

What is this program and why does it matter?

TAC, which stands for Trusted Access for Cyber, is OpenAI's scheme for giving pre-approved security researchers access to its most powerful AI models with fewer safety guardrails than the public gets. Anthropic, OpenAI's main rival, runs a comparable scheme called the Cyber Verification Program.

The logic is straightforward. Cybercriminals are already trying to use AI to find software vulnerabilities and build exploits. Defenders need equally capable tools to find those same flaws first. TAC is meant to be that equaliser.

Daybreak Blue is the entry tier for individual researchers, supporting vulnerability discovery, secure code review, and patch validation, among other defensive tasks. Daybreak Red sits above it, offering models purpose-built for offensive security testing. We covered the launch of both tiers on 13 August in "OpenAI splits its Daybreak cybersecurity programme into two tiers and launches a new security-focused AI model".

What went wrong?

OpenAI's own error triggered the lock-outs. One researcher shared an email from OpenAI stating that access to Daybreak Blue had been revoked "due to a technical issue affecting a limited number of users." A separate support thread on OpenAI's forums produced an almost identical explanation, citing a "recent technical issue."

The company pointed TechCrunch to a public post confirming that "a limited set of users' access to Daybreak Blue is no longer active and they will need to re-verify to maintain their access."

All five researchers TechCrunch spoke to are based outside the US and Europe, pointing to a possible geographic pattern in whatever broke.

What should affected researchers do?

Re-verify. Locked-out users need to go through the TAC application process again: submit ID, complete vetting, start over.

It's inconvenient, but OpenAI is framing it as a one-time fix rather than a policy change. Nobody's being permanently removed.

One broader tension sits behind this story. Security researchers working both defensively and offensively have publicly complained that even TAC's loosened guardrails still block legitimate work. That debate will outlast this week's glitch by a long way, and it's worth watching whether incidents like this erode researchers' confidence in the program before it's even found its feet.

Common questions

Does this mean OpenAI is restricting researchers more than before?

No. OpenAI says this was a technical error, not a policy decision. Researchers who re-verify should regain the same access they had before.

Can ordinary ChatGPT users access these reduced-restriction models?

No. Daybreak Blue and Daybreak Red are only available to researchers who pass OpenAI's identity and vetting checks. Standard ChatGPT accounts are unaffected.