29 September 2026 | 07:38

READING: 1 minute

OpenAI has postponed the debut of its new artificial intelligence model, GPT-6.1 Astra, citing “safety issues.” The Wall Street Journal reported this in recent hours, and the BBC later confirmed it with Saachi Jain, OpenAI’s head of safety systems, who said the new system “did not fully meet the company’s standards.”

The announcement from the startup led by Sam Altman comes just 24 hours after OpenAI issued “apologies” to the Australian government. Yesterday, the US company said it should have shared preliminary findings of an investigation earlier with Australian authorities concerning some “incidents that occurred last June but were only made public last week,” after its AI models had accessed government websites. “Our goal was to provide the agencies involved with a detailed report once the investigation was complete,” OpenAI said in a blog post. “However, we should have shared preliminary conclusions earlier and kept Australian agencies updated as new information emerged.”

As the BBC reports, OpenAI’s decision to delay the model’s release is a rare example of a major AI developer taking action for safety reasons. The model was found lacking in its “ability to remain within its intended scope and authorized permissions, as well as in communicating to the user what work it actually performed,” Jain explained.

Read also