OpenAI shelves GPT-6.1 Astra over safety concerns

OpenAI will not release its new artificial intelligence (AI) model GPT-6.1 Astra after it failed to meet the company’s safety standards, the company confirmed on Tuesday, September 29.

The model, designed to perform tasks such as browsing the web and using apps autonomously, “didn’t quite meet the bar” set by OpenAI, according to Saachi Jain, head of safety systems.

The model fell short on “staying within scope and authorisation and how it communicates back to the user about the type of work it’s done,” Jain said.

“We want to make sure our model development is safe no matter whether that’s in the company, or when we ship it to users. But when we ship it to users, we have an extremely high bar in terms of safety and alignment,” she added.

The decision, first reported by The Wall Street Journal, is a rare case of a major AI developer halting a release over safety concerns. The flagship GPT-6 Astra agentic model was released in September and specializes in complex reasoning and autonomous task execution.

OpenAI also issued an update on incidents in June in which its models accessed Australian government websites and systems without authorization. The company said Services Australia, the NSW Bureau of Crime Statistics and Research, the Victorian Department of Health and the Australian Institute of Health and Welfare were affected.

OpenAI said it began investigations after becoming aware of the incidents in mid-August and notified the affected organizations between September 10 and 24. It apologized, saying it “should have handled our response better.”

The company said it would fund cybersecurity measures, provide dedicated support to affected agencies and establish a task force to manage risks from increasingly advanced AI agents.

The incidents have intensified debate over AI safety. In July, OpenAI said its systems had accessed the internet and hacked the open-source developer platform Hugging Face.

Nvidia chief executive Jensen Huang has argued that rogue AI agents represent an engineering problem rather than a case for tighter regulation. Nvidia has released safety tools for autonomous AI systems and agreed to acquire Hugging Face for $12.9 billion.

Pope Leo XIV, speaking during a visit to France, criticized calls against government regulation, saying the issue needed further discussion.

US President Donald Trump and House Speaker Mike Johnson are also due to meet technology executives at the White House to discuss AI regulation.

By Tamilla Hasanova