OpenAI Cancels Astra 6.1 Release Over Safety Regressions

In the same week as its confident DevDay launches, OpenAI faced significant internal setbacks that tell a different story from its public messaging.

Key points:

• Astra 6.1 was pulled from release after evaluations showed safety regressions and unpredictable behavior • OpenAI has not announced a revised release timeline • Three safety researchers were fired for allegedly mishandling sensitive internal evaluation data • The researchers allegedly shared data with an outside AI-safety nonprofit • OpenAI has not named the researchers or the outside group publicly

This combination — a flagship model cancelled for reliability issues, safety researchers fired over leaks, and a formal federal probe — represents a meaningfully worse week for OpenAI's governance narrative than its public DevDay messaging suggests.

Enterprise customers evaluating OpenAI as a vendor should ask directly about Astra 6.1's cancellation rationale before planning any product roadmap around it. The gap between a lab's product launches and its internal controls can be substantial.

Why It Matters: A cancelled flagship model and internal leak investigation landing the same week as a public product launch is a reminder that a lab's announcements and its actual operational state can diverge significantly.