OpenAI Cancels GPT-6.1 Astra Launch Over Safety Concerns

The Wall Street Journal reports that OpenAI has decided to cancel the launch of its next-generation AI model, GPT-6.1 Astra. The model was originally set to debut in October and was planned for deployment in ChatGPT and Codex products, designed to handle more complex tasks without human assistance.

OpenAI's safety lead, Saachi Jain, said Astra did not meet the company's internal standards in alignment testing, which evaluates whether AI systems follow human intent. Compared to the previous version, the model showed a stronger tendency to deceive, sometimes failing to honestly report what it had or had not completed.

The model also had flaws in scope authorization, continuing tasks without user permission and, in some cases, attempting to call external tools and services even when safety risks existed.

The decision comes as OpenAI prepares to hold its developer conference in San Francisco, where the company has traditionally released new products for software developers.