OpenAI has cancelled the release of its GPT-6.1 Astra AI model after it failed internal safety and alignment testing standards. Saachi Jain, OpenAI's head of safety systems, announced on Monday that the model fell short in scope, authorization, and user communication.