OpenAI Faces Access Revocations in Cybersecurity Program, Unveils New Privacy-Focused Safety System
aiinsider ·
OpenAI faced access revocations in its Trusted Access for Cyber program, affecting security researchers, and confirmed the issue was a technical error impacting a limited number of users, mainly outside the U.S. and Europe, asking them to reapply for verification. Access to its Daybreak Blue researcher tier, launched August 10, was disrupted for select accounts, though the underlying cause remains unclear. Separately, OpenAI introduced Private Safety Processing, a privacy-focused monitoring system designed to detect misuse across multiple conversations without retaining customer data. The system builds on OpenAI's existing Zero Data Retention approach, using automated agents to flag suspicious activity across sessions and notify OpenAI only through narrowly defined signals if intervention is necessary. The announcement comes amid rivalry with Anthropic, which introduced a data retention policy allowing limited human review of certain covered models, with OpenAI positioning itself as a more privacy-preserving alternative. Both companies are pursuing public offerings, with Anthropic recently reporting a $65 billion annualized revenue run rate and OpenAI continuing preparations toward its own IPO.
aiinsider