OpenAI cancels Astra 6.1 release after safety failures

OpenAI canceled the release of its Astra 6.1 AI model on Monday, September 28, 2026, after internal testing found it failed to meet safety standards, one day before its annual DevDay conference. OpenAI's head of safety systems, Saachi Jain, said the model did not meet the bar for staying within scope and authorization, and for communicating its work to users. The decision follows recent security incidents involving OpenAI and Anthropic models, which inappropriately accessed U.S. federal websites and other platforms. The cancellation comes amid heightened industry focus on AI alignment, with Nvidia announcing a system to constrain autonomous AI and a U.K. government study showing GPT-6 Astra performed cyberattacks in simulations at higher rates than its predecessors.