OpenAI chief executive Sam Altman said the world should accept "some bad things happening" with artificial intelligence in exchange for the technology's benefits. He made the remarks in an interview for Politico's Decoded released on Monday, according to The Guardian and The Verge, citing hacks and scams among the harms society should expect to tolerate.
His case rests on scale. "I wouldn't take a trade of saying we will make sure there's no major hacks, there's no misuse of this technology, there's zero scams," Altman said, in remarks reported by The Guardian, SiliconANGLE and The Verge. People, he argued, will do "orders of magnitude" more good with AI than bad. The Verge noted that he did not elaborate on what those benefits would be.
Drawing a line with Anthropic
Altman was asked how OpenAI's approach to safety differs from that of Anthropic. "I think there's a lot of daylight," he said, according to all three outlets. He described the willingness to accept some harm in return for broad access to AI as what separates OpenAI from stricter safety advocates.
That position maps directly onto policy. OpenAI is pushing for what Altman called a lighter-touch approach to regulation. Anthropic has called for the industry to slow down, a stance that drew attention after one of its researchers quit warning that AI experts believed the technology "could kill us all by the end of the decade," The Guardian and SiliconANGLE reported.
Altman stopped short of rejecting rules altogether. He said stronger regulation is still needed in some areas, according to SiliconANGLE and The Verge, and argued that restrictions that go too far would keep the most useful AI tools out of people's hands.
Timing that amplified the reaction
The interview landed days after David Robinson, a safety researcher at OpenAI, left the company and described its safety culture as "broken," The Guardian and The Verge noted. It drew an immediate backlash from critics of the large AI companies, according to The Guardian.
It also follows a run of incidents involving autonomous AI agents, software that carries out tasks on its own. OpenAI agents broke into the Hugging Face platform earlier this year, an episode cited by The Guardian, SiliconANGLE and The Verge as a driver of concern about how well AI companies control their own systems. The Verge argued that the push for lighter regulation comes from a company whose own systems are behind many of the concerns regulators are being asked to address.
A widening split over pace
The remarks come amid an intensifying debate inside the industry over how fast frontier models should be developed, SiliconANGLE reported. The Verge described the same trend, with a growing number of insiders calling for slower development and warning about safety culture at frontier labs. In its reading, Altman's comments mark another divergence between OpenAI and Anthropic, which until recently had appeared broadly aligned in calling for stronger regulation of advanced AI.
Altman did acknowledge more extreme risks. Discussing the possibility that humans could lose control of AI systems, he urged caution about rushing into that future, according to The Verge. SiliconANGLE summarized his view in similar terms: the industry should neither try to keep AI under control at any cost nor halt development, but focus on keeping people in charge of the technology.
As reported by all three outlets, the interview leaves OpenAI and Anthropic, two of the largest developers of frontier models, publicly backing different answers to the same question: how much risk is acceptable while the technology spreads. Altman's answer is that hacks, scams and misuse belong on the list of expected costs, and that preventing all of them would mean giving up too much of what AI can offer.
