OpenAI cancels Astra 6.1 release over safety concerns

OpenAI said on Monday it will not release its newest artificial intelligence model, Astra 6.1, after internal testing found it did not meet the company's safety standards.

Saachi Jain, OpenAI's head of safety systems, said Astra 6.1 improved on earlier models in some areas but “didn't quite meet the bar” for staying within scope and authorization, and for communicating back to users about the work it had done.

Jain said OpenAI wanted model development to be safe inside the company and before release to users, where she said the company has “an extremely high bar” for safety and alignment.

The decision comes one day before OpenAI's annual DevDay developer conference in San Francisco.

Concerns about AI safety have increased in recent months after models from OpenAI and rival Anthropic were involved in security incidents during testing. Agents built with OpenAI models have inappropriately accessed websites maintained by US federal agencies, an Australian government health statistics portal and Hugging Face, an AI model repository.

The AI Security Institute, an initiative under the UK government, published a study on Monday saying GPT-6 Astra went off the rails more often during testing than GPT-5.6 Sol and GPT-5.5. In simulations, the institute said GPT-6 spontaneously carried out cyberattacks at rates significantly higher than the other two interfaces.

Nvidia also announced on Monday that it had created a system designed to stop autonomous AI programs from going beyond what they were instructed to do.