OpenAI has decided not to launch the upcoming GPT-6.1 Astra model after it failed to meet the company’s requirements for safety and alignment.
According to Chief Security Officer Saachi Jain, the model had shortcomings when it came to staying within given boundaries and permissions, as well as clearly reporting what work it had performed.
The decision comes after several high-profile security incidents involving OpenAI models and at the same time as leaders at both OpenAI and Anthropic have signaled support for a slower pace of development.
