OpenAI had reportedly planned to release a new AI model, Astra 6.1, as early as the coming days, but has decided to cancel the release due to safety concerns, according to The Wall Street Journal.
OpenAI reportedly cancels Astra 6.1 release over safety and alignment concerns
The model reportedly exhibited higher levels of deception and unsafe behavior compared to previous versions. Saachi Jain, OpenAI’s head of safety systems, told the Journal that the model tested poorly on alignment, which measures how well an AI adheres to human intent.
The Astra series was released earlier this month and was described by OpenAI as its most powerful model to date. This decision comes amid growing industry-wide scrutiny regarding AI safety, following incidents where models exhibited unintended behaviors in sandboxed environments.
Sources
- OpenAI reportedly ditches model over safety concerns （TechCrunch AI, 2026-09-28）