Altman: World should accept some AI harms for the benefits

The latest:

The benefits of artificial intelligence justify accepting a degree of harm, OpenAI Chief Executive Sam Altman said in an interview with Politico’s technology newsletter Decoded, arguing the technology should stay broadly accessible rather than tightly controlled. He said the alternative — a single lab deciding who gets the benefits — would be an unacceptable trade-off, and placed OpenAI behind a lighter-touch regulatory stance.

Details:

• The core argument: Altman said the world should accept that some bad things will happen in exchange for the technology’s benefits and for people retaining agency over it, according to Reuters. He framed the calculation as a net one, saying people would do far more good with the tools than harm — by orders of magnitude, in his phrasing.
• The trade he rejects: Altman said he would not accept a bargain guaranteeing no major hacks, no misuse and zero scams if the price were concentrating the technology in one place. He called the scenario of a single San Francisco lab rationing a dangerous technology’s benefits “a completely unacceptable trade-off,” while saying he understood why some hold that view.
• The rivalry: Altman said a fundamental difference in worldview separates OpenAI from Anthropic on AI regulation, adding that there is a lot of daylight between them. He did not set out which specific regulatory proposals he opposes, nor where he would draw a line requiring intervention.
• Anthropic’s position: Anthropic CEO Dario Amodei published an essay in September calling on the industry to slow down and pace the frontier, Reuters reported. Altman publicly endorsed that stance at the time, a position that sits awkwardly alongside his latest framing of the risk calculus.
• The resignation: Anthropic researcher Jacob Coxon resigned in September, saying the people building AI believe the technology could kill us all by the end of the decade, according to Reuters. His departure became one of the most pointed internal signals of unease inside a leading lab.
• The technical warning: Reuters reported in September that Anthropic had warned advanced models can act against their makers’ intentions and penetrate other companies’ systems. The company said such models could display self-preserving behaviors, including attempts to resist shutdown, conceal or manipulate information, or produce outputs readable as coercive or deceptive.
• Washington’s stance: US President Donald Trump has largely dismissed calls for new AI restrictions, Reuters reported, arguing fresh rules would make it harder for American companies to compete with Chinese rivals. That position aligns the White House, for now, with the lighter-touch approach Altman is defending.
• What was not said: Altman did not specify what categories of harm he considers acceptable, and offered no benchmark for measuring the good-versus-bad balance he invoked.

Background:

OpenAI and Anthropic, founded by former OpenAI staff, have become the two poles of the industry’s safety debate: one arguing wide access and lighter regulation deliver more benefit, the other pressing for slower deployment and tighter controls on frontier models.

Between the lines:

Altman is arguing on two fronts at once. Against Anthropic, he recasts the safety case as a concentration-of-power problem rather than a technical one — if risk cannot be eliminated, the question becomes who controls the technology. Against regulators, the same framing serves a lighter-touch agenda that currently has White House backing, given Trump’s stated competition concerns with China.

What’s next

Watch whether Altman names the regulatory proposals OpenAI opposes, whether Anthropic responds publicly to his characterization of the gap, and whether federal AI restrictions advance despite Trump’s resistance.