Several cybersecurity researchers say OpenAI unexpectedly removed their access to a limited program that provides vetted users with fewer restrictions on its AI tools. OpenAI has confirmed that the disruption was caused by a technical error.
On Wednesday, researchers reported on OpenAI’s support forum and X that they had lost access to Trusted Access for Cyber (TAC). Some users saw messages saying their identity could not be verified or that their accounts were currently ineligible.
Follow THE FUTURE on LinkedIn, Facebook, Instagram, X and Telegram
What Is Trusted Access For Cyber?
TAC is a program through which OpenAI gives vetted cybersecurity researchers access to advanced models with fewer safeguards than those available to regular users. Researchers must submit identification and complete OpenAI’s verification process.
The program is designed to help security professionals identify vulnerabilities, test defenses and report flaws while limiting access for malicious actors. Anthropic operates a similar initiative called the Cyber Verification Program.
OpenAI Says Access Will Be Restored
TechCrunch spoke with five researchers who said they experienced the issue. One researcher received an email saying access to Daybreak Blue, the latest TAC tier, had been revoked because of a technical problem affecting a limited number of users.
OpenAI said affected researchers would need to reapply and complete the verification process. The company also confirmed the issue in a post on X, saying a limited group of users had lost access and would need to re-verify.
All five researchers who spoke with TechCrunch were based outside the U.S. and Europe, although it remains unclear whether geography played a role or how many users were affected.
OpenAI Expands Its Cybersecurity Tiers
Daybreak Blue, introduced on August 10, provides vetted researchers with access to advanced general-purpose models for defensive security work, including vulnerability discovery, code review, malware analysis and incident response.
OpenAI also introduced Daybreak Red, a higher-access tier designed for authorized vulnerability research, exploit validation and security testing.
The access issue comes amid growing debate among cybersecurity researchers over AI safety restrictions. Some defensive and offensive security professionals argue that increasingly strict guardrails can interfere with legitimate security research.
