OpenAI confirmed it accidentally revoked access for a number of cybersecurity researchers enrolled in its Trusted Access for Cyber (TAC) program, a limited-access initiative that allows vetted defenders to use advanced AI models with fewer restrictions for security research.
On Wednesday, August 19, 2026, multiple researchers reported on OpenAI’s official support forums and on X that their TAC access had been cut off. When attempting to open ChatGPT’s Cyber page, they were met with messages stating their identity could not be verified or that their account was “ineligible at this time.” TechCrunch spoke with five researchers who experienced the problem, all of whom said they live outside the U.S. and Europe, suggesting the disruption may be limited to certain regions.
OpenAI attributed the outage to a technical error. In an email shared with TechCrunch, the company told one affected researcher that access to Daybreak Blue — the latest individual-researcher tier of TAC — was revoked “due to a technical issue affecting a limited number of users,” adding, “This was an issue on our end, and not the user experience we want to deliver.” OpenAI asked affected researchers to reapply and complete the verification process again.
Daybreak Blue, launched on August 10, 2026, grants access to frontier models including GPT-5.6 Sol for defensive security work such as vulnerability discovery, malware analysis, and patch validation. OpenAI also introduced a higher tier, Daybreak Red, which provides access to models built specifically for offensive security tasks including exploit validation and authorized vulnerability research.
TAC is designed to give vetted security professionals access to more capable AI tools so they can identify and report software vulnerabilities faster, while keeping those same tools out of the hands of malicious actors. Anthropic runs a comparable program called the Cyber Verification Program. To qualify for TAC, researchers must submit identification and pass an OpenAI vetting process.
The disruption comes as both defensive and offensive security researchers have raised concerns that guardrails imposed by OpenAI and Anthropic are limiting their ability to conduct legitimate security work.
Source: TechCrunch