OpenAI announced on Monday that it is halting the release of its newest artificial intelligence model, citing security concerns identified by researchers during internal testing.
The company originally targeted an October release for GPT-6.1 Astra, The Wall Street Journal reported. The model was more than capable of completing challenging tasks from start to finish without human assistance, according to the newspaper.
During the testing phase for GPT-6.1 Astra, the model showed high levels of what OpenAI saw as deception, or a willingness to mislead users about its actions, The New York Times reported.
Exclusive: OpenAI is scrapping the release of its next-generation AI model because it failed to meet safety standards https://t.co/1Wba4I00yn
— The Wall Street Journal (@WSJ) September 28, 2026
The model was also willing to go beyond the original scope of what it was asked to do without checking for instructions, according to the newspaper.
Saachi Jain, OpenAI’s head of safety systems, said in an interview with The Wall Street Journal that GPT-6.1 Astra had regressed in those two areas.
“For anything regarding safety and alignment, there’s a trade-off,” Jain told the newspaper. “You really do need to find what’s the right line between staying within scope, but also avoiding laziness in terms of how the model actually pursues tasks even when it hits friction.”
OpenAI’s decision to scrap rolling out GPT-6.1 Astra came after weeks of reports that its AI models had gone rogue during testing, the Times reported. The models were hacking into websites without the company’s knowledge, or exhibited behavior that researchers called “concerning.” Those concerns included the model hiding mistakes and making up data.
Sam Altman, OpenAI’s chief executive, said in a social media post on Friday that the company had “not been as fast as we would have liked” in disclosing A.I. incidents. “We are prioritizing as best as we can based on severity.”
There is an extensive and ongoing review related to our agents’ use of internet access during training and evaluation. We’ve been publishing summaries at the link below and will continue to.
— Sam Altman (@sama) September 25, 2026
We have not been as fast as we would have liked but we are trying to balance our desire… https://t.co/8zoMxas5Eq
Many of the publicly known agent-security incidents involved OpenAI’s internal AI models, The Wall Street Journal reported. Those models were never scheduled for public release.
“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,” Jain told the newspaper. “But when we ship it to users, we have an extremely high bar in terms of safety and alignment.”
© 2026 Cox Media Group
