OpenAI Error Revokes Daybreak Blue Cyber Research Access
OpenAI says an error revoked some researchers’ Daybreak Blue access, forcing reverification for GPT 5.6 Sol and other cyber defense tools after its launch.
Summary
On Wednesday, August 19, 2026, researchers reported on OpenAI’s support forums and X that their Trusted Access for Cyber access had vanished. ChatGPT’s Cyber page said their identities could not be verified or their accounts were ineligible. TechCrunch interviewed five affected researchers, all outside the U.S. and Europe. OpenAI blamed a technical error affecting a limited number of Daybreak Blue users, but neither the precise cause nor total affected remains known. An email, support response and company post instructed users to reapply and complete identity verification.
TAC requires identification and vetting, then gives approved defenders advanced OpenAI models with fewer cybersecurity restrictions than regular users receive. OpenAI launched Daybreak Blue on August 10 as its latest individual tier, providing frontier general purpose models including GPT 5.6 Sol for vulnerability discovery, secure code review, malware analysis, incident response and patch validation. The simultaneously introduced Daybreak Red tier offers cybersecurity specific models for authorized vulnerability research, exploit validation and security testing.
OpenAI’s TAC and Anthropic’s comparable Cyber Verification Program aim to accelerate vulnerability reporting and patching while excluding criminals seeking bugs and exploits. Defensive and offensive researchers have recently argued that both companies’ guardrails also obstruct legitimate security work.
Positives
- OpenAI confirmed that a technical error, rather than researcher misconduct, caused the access losses.
- Daybreak Blue provides vetted defenders GPT 5.6 Sol for vulnerability discovery, code review, malware analysis, incident response and patch validation.
- Daybreak Red expands approved research to exploit validation and authorized security testing using cybersecurity specific models.
- TAC and Anthropic’s Cyber Verification Program are designed to help defenders report flaws and accelerate company patches.
Risks & concerns
- Five affected researchers interviewed by TechCrunch abruptly lost Daybreak Blue access and were required to reapply.
- OpenAI has not disclosed the technical cause or the total number of users whose access was removed.
- All five interviewed researchers live outside the U.S. and Europe, suggesting a possible regional pattern that remains unconfirmed.
- Identity verification failures and ineligibility notices interrupted access despite users having previously passed OpenAI’s vetting.
- Defensive and offensive researchers say OpenAI and Anthropic guardrails have prevented legitimate cybersecurity work in recent months.
