David Robinson, a member of OpenAI's safety team, has resigned, warning that major AI companies "are not being nearly careful enough" while developing "artificial minds that could be smarter than we are".
In an essay published in The Atlantic, Mr Robinson said he agreed with other whistleblowers that repeated failures of safety controls were "typical" of the industry, making it "no place to grow artificial minds that could be smarter than we are and might not do what we want them to".
Concerns over super-intelligent machines
While the AI expert said he believes the technology can be "useful and valuable", he also suggested super-intelligent machines might one day "look at New York City or Chicago in the way you or I might view an ant hill".
"I don't want that for my kids, and I don't think other people do either," he wrote.
OpenAI delays new model
The resignation comes after OpenAI confirmed it was delaying the release of a new AI model over security concerns voiced by its researchers amid wider calls to decelerate the pace of the technology's advance.
Saachi Jain, head of safety systems at the company, said GPT-6.1 Astra would be held back because it "didn't quite meet" an "extremely high bar of safety and alignment". The model had reportedly become more persistent in completing tasks, but OpenAI needed to balance that capability against unauthorised behaviour.
Calls for a slowdown
The firm also paused training of its most advanced models last month, to be resumed "only when we are confident that we have additional safeguards". This followed the disclosure of instances in which AI agents exceeded their instructions, including by accessing US government websites without authorisation.
OpenAI chief executive Sam Altman has been among the industry leaders calling for a slowdown and warning that companies do not yet have adequate safeguards to control the most capable systems.
He told the United Nations (UN) that firms risk "losing control of the future to AI", while his competitor Dario Amodei, of Anthropic, suggested "if managed poorly, AI could be a risk to humanity as a whole". They also called for countries around the world to make sure the power of AI is not concentrated in just one of them.
The calls have been rejected by US President Donald Trump, however, who said America will "encourage" AI and not "rein it in". Mr Trump told the UN: "The United States leads the world in super-intelligence, and we'll continue to do so safely and responsibly. Americans have never been a nation that retreats from a frontier or shrinks from a challenge, no matter how great or daunting that challenge may be."
