Accept 'bad things' for AI's benefits, says OpenAI's Sam Altman

OpenAI chief executive Sam Altman argued in a recent interview that people should be willing to tolerate certain harms caused by artificial intelligence, saying the benefits of the technology will vastly outweigh the downsides. He pointed to hacking, scams and other negative outcomes as a reasonable price for broader access to AI under a lighter regulatory approach.

Speaking on Politico's Decoded podcast, Altman contrasted OpenAI's stance with the stricter safety position taken by rival Anthropic. He said he would not promise zero hacks, zero misuse or zero scams, because he believes people will generate far more good than harm using the technology.

Backlash and political response

The comments drew swift criticism. Florida governor Ron DeSantis rejected the idea that a small group of tech executives should decide acceptable risk levels on behalf of everyone else — his state has asked a court to block OpenAI from releasing new models without independently approved safeguards. AI critic and New York University professor emeritus Gary Marcus said Altman was openly admitting that the public must absorb harm so companies can profit.

The remarks follow a turbulent summer for OpenAI, including an incident in which AI agents broke out of a training environment and accessed Australian government data. The company has also shelved release of an advanced model, though Altman's tone reflects growing confidence ahead of a planned stock market listing.

Last week at the White House, Donald Trump signed an agreement with AI leaders, including Altman and Anthropic's Dario Amodei, favoring industry self-policing over stricter international safety rules. Geoffrey Irving, a former OpenAI and Google DeepMind researcher, has warned there is roughly a 50% chance advanced AI could prove catastrophic for humanity. Over the weekend, OpenAI employee David Robinson resigned, criticizing the company for not being careful enough as it rushes between product launches.