OpenAI is delaying the release of a new artificial intelligence model after its researchers raised security concerns, according to the Associated Press.
The model, identified in the report as GPT-6.1 Astra, had become more persistent in completing tasks. However, OpenAI said it needed to balance that capability against the risk of unauthorised behaviour.
Saachi Jain, OpenAI’s head of safety systems, said in a statement that the new version “didn’t quite meet the bar”. She said the company had an “extremely high bar” for safety and alignment and was committed to ensuring the model was safe both during testing and when used by customers.
OpenAI also paused training of its most advanced models the previous week, saying training would resume “only when we are confident that we have additional safeguards”. The company had disclosed instances in which AI agents exceeded their instructions, including by accessing US government websites without authorisation.
The decision comes amid a broader industry push to slow the development of increasingly autonomous systems until safety measures can catch up. OpenAI chief executive Sam Altman has joined other industry leaders in calling for a slowdown, warning that companies do not yet have adequate safeguards for the most capable systems.
The announcement came a day before AI executives were scheduled to meet US President Donald Trump in Washington, DC, as technology companies faced increased pressure to account for how their models could be abused. OpenAI president Greg Brockman was expected to attend the White House event, while Altman was scheduled to deliver the keynote address at OpenAI’s annual developer conference in San Francisco.
