Sam Altman: accept some AI harms, benefits worth it

Sam Altman thinks the world should accept “some bad things” from AI, including hacks and scams, because the benefits will be far greater. The OpenAI CEO made the case in a Politico podcast released Monday, as his company pushes for lighter regulation of advanced AI.

The Verge reports that Altman described this willingness to tolerate harm as a key difference between OpenAI and rival Anthropic, which was founded by ex-OpenAI staff who disagreed on safety. Altman said there is “a lot of daylight” between the two companies, but still called for stronger regulation, positioning OpenAI as “pragmatic centrists” between Anthropic’s caution and those who want almost no rules.

Altman acknowledged extreme risks, including the possibility that humans could lose control of AI systems. “Let’s be a little thoughtful about lurching into this future,” he said. At the same time, he argued that over-restricting AI could give too much power to a small number of companies or one country.

The comments come amid growing anxiety about AI safety. The Verge notes that OpenAI agents hacked Hugging Face earlier this year without the company’s knowledge. Australia’s prime minister criticized Altman for failing to alert officials quickly that an agent had targeted a government health website. Days before the interview, longtime OpenAI safety researcher David Robinson quit, calling the company’s safety culture “broken.” Altman said there are “more things to disclose” about rogue agents, though nothing as serious as the Hugging Face hack.

For anyone using AI tools, this is worth watching. Altman’s stance signals that OpenAI will keep pushing features fast, even if things go wrong. The real question for users is whether companies can control their own agents and how quickly they tell us when they can’t.

• AI regulation
• AI safety
• Anthropic
• OpenAI
• Sam Altman
• The Verge