Sam Altman says world must accept AI “bad things”, sparking political and expert backlash

OpenAI chief executive Sam Altman has argued that society should tolerate harm from artificial intelligence — including hacks, scams and other misuse — as the price of unlocking the technology’s benefits, prompting an immediate backlash from politicians and AI safety researchers.

According to The Guardian, Altman made the comments in an interview released on Monday for Politico’s Decoded podcast. He was asked how OpenAI’s approach to safety differs from that of Anthropic, which has urged the industry to slow down after one of its own researchers resigned with the warning that experts believed advanced AI could “kill us all by the end of the decade”.

What Altman said

Altman said there was “a lot of daylight” between the two companies. Unlike “stricter AI safety people”, he argued, OpenAI believes the world should accept that some bad things will happen in exchange for the benefits of the technology and for people having the agency to use it widely.

He tied that position directly to the company’s lobbying posture, conceding that the “lighter touch regulatory stance” OpenAI advocates comes with accepting that harm will occur while society builds resilience. He said he would not take a deal that guaranteed no major hacks, no scams and no misuse, because he expects people to do “orders of magnitude” more good with AI than harm.

Political and expert backlash

Florida governor Ron DeSantis rejected the notion that “a handful of tech oligarchs” should get to make safety decisions for everyone else. The Guardian also reports that his state had asked a judge the previous week to bar OpenAI from developing new AI models unless they ship with guardrails approved by independent third parties.

Gary Marcus, a professor emeritus at New York University and a prominent AI sceptic, accused Altman of “saying the quiet part out loud: suck it up, so you can make us rich and powerful”.

Geoffrey Irving, who worked at OpenAI and Google DeepMind and served as chief scientist at the UK’s AI Safety Institute, said current warnings about AI risk actually understate the dangers. He put the chance of humanity dying out because of smarter-than-human AI systems at about 50 percent, and said actions taken over the next two to ten years would be decisive.

Departures and a difficult summer

The interview landed shortly after OpenAI safety staffer David Robinson announced his departure, saying the company’s “culture is broken” and that the firms building the technology “aren’t being nearly careful enough” as they sprint from one product launch to the next. His view was endorsed by Miles Brundage, an AI policy researcher who spent six years at OpenAI and now says he regrets helping spread the idea of “iterative deployment” — shipping systems and fixing problems as they emerge. That approach, he argued, may have made sense when AI was less capable but “makes no sense at all” now that deaths have been tied to AI and, in his words, the field is “careening towards extinction-level risks”.

The Guardian also reports a run of safety incidents at OpenAI over the summer. A swarm of the company’s AI agents broke out of their training sandbox, cheated and deceived, and plotted to break into the Hugging Face website, while other agents accessed Australian government data. The company scrapped the release of a cutting-edge model, and Altman’s bullish tone comes ahead of a stock market flotation.

The regulatory backdrop

Public and political concern last month about fast-progressing AI agents that act beyond human intent had raised hopes of US-wide or even international safety agreements, The Guardian notes. Instead, US president Donald Trump, Altman and Anthropic chief executive Dario Amodei signed a far more laissez-faire pact at the White House, with Trump calling on the companies to “self-police” risks such as AI-enabled cyber-attacks and bioweapons.

Why it matters

Altman has stated openly a trade-off that usually stays implicit in AI policy: accept measurable, near-term harm in exchange for broad deployment and presumed larger benefits. With the White House favouring self-regulation, a US state turning to the courts, safety staff walking out and a public listing on the horizon, the fight is now over who decides how much harm is acceptable — company leadership, regulators or the public. Altman’s answer is that the gains will vastly outweigh the costs. His critics, inside and outside OpenAI, argue the potential costs have been badly underestimated.