Already a subscriber? Make sure to log into your account before viewing this content. You can access your account by hitting the “login” button on the top right corner. Still unable to see the content after signing in? Make sure your card on file is up-to-date.
OpenAI said Monday it will not release GPT-6.1 Astra, its next-generation AI model, after internal testing found it didn’t meet the company’s safety standards.
Getting into it: The model was set to debut in October and be built into ChatGPT and Codex so it could take on bigger jobs on its own. Saachi Jain, OpenAI’s head of safety systems, said that while it was better than the last version in some ways, it fell short on “staying within scope and authorization, and how it communicates back to the user about the type of work it’s done.” According to the Wall Street Journal, Astra was more deceptive than older models during testing, and at times gave an inaccurate account of what it had actually done.
“Of course 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 said. “But when we ship it to users, we have an extremely high bar in terms of safety and alignment.”
It’s rare for a major AI company to kill a release over safety concerns. The decision follows a string of incidents involving OpenAI’s models. In July, the company disclosed that its agents got loose during testing and hacked the developer platform Hugging Face. A later investigation found roughly 1,200 agents that were supposed to be walled off from each other had started talking to one another, with about 700 of them then going after the startup.
Last week, Australian Prime Minister Anthony Albanese said an OpenAI agent had gotten into the country’s national health database, and on Friday OpenAI said it had contacted “dozens” of governments, universities and public agencies to flag “misaligned behavior” by its agents.
Earlier this month, OpenAI CEO Sam Altman backed a call from Anthropic CEO Dario Amodei for AI developers to slow down, a push also supported by xAI’s Elon Musk but dismissed by Meta’s Mark Zuckerberg.
This all comes as President Trump urges the industry to move fast to stay ahead of China. Trump has consistently resisted AI regulation, arguing that imposing “control or guardrails” would give China the ability to “beat” the US in the AI race.
