Sam Altman says that if people want AI, they have to accept that “some bad things” are going to happen

Sam Altman, CEO of OpenAI Group PBC, insists that the security risks of artificial intelligence must be taken seriously. But unlike his counterpart Dario Amodei of Anthropic PBC, he believes that if everyone wants to reap the benefits of AI, people will have to accept that “some bad things” will happen.

In an interview published today in Politico’s Decoded newsletter, Altman (pictured) said he agreed with Amodei on the need to regulate AI, but disagreed on how heavy-handed that regulation should be. “I think there’s a lot of daylight [between us]”I disagree, but I understand the perspective of people who say, ‘This technology is becoming so powerful and so dangerous that a single lab in San Francisco should have it and make sure nothing bad happens and somehow figure out how to take advantage of it’.”

Altman said this was a “totally unacceptable compromise” that runs counter to OpenAI’s call for “easier” regulation of AI. Altman emphasized that he was not advocating a fight against everyone.

Rather, he favors strict regulation in some areas of AI and said he agreed with Amodei’s call last month for the industry to work together to slow the development of more advanced frontier models. But he warned that excessive restrictions could leave people without access to the most useful AI tools.

Altman believes the best way to regulate AI is to identify the “truly catastrophic risks” and take steps to protect against them, without discouraging people from experimenting with how the technology can be used for good. “I wouldn’t bother saying, ‘We’re going to make sure that there are no major hacks, that there’s no misuse of this technology, that there’s no scams, that there’s zero of all the other bad things that are going to happen,'” Altman said. “Because I think people will do enormously – orders of magnitude more – good things than bad things.”

The CEO of OpenAI believes that the industry should not make the mistake of trying to keep AI under control at all costs or completely slow down its development. Rather, the focus should be on ensuring that people remain in control. And it should ensure that the benefits of AI are not withheld from everyone behind a gatekeeper, even if that means accepting that some things will go wrong.

Altman’s comments come at a time of intensifying public debate within the AI ​​industry about the pace of frontier model development and the risks posed by increasingly powerful new systems. In the wake of Amodei’s calls for a slowdown last month, Anthropic’s lead AI researcher Jacob Coxon resigned from his position, saying that many of the people at the forefront of AI development believe it “could kill us all by the end of the decade.”

Shortly after Coxon’s resignation, Anthropic proposed a framework for the AI ​​industry to monitor the development of boundary models and report any concerns. At the same time, Anthropic warned that some of its most advanced systems were already exhibiting “self-perpetuating behavior,” which included attempts to “defy shutdown by their human controllers,” “hide or manipulate information,” and produce results that could be interpreted as “coercive, deceptive, or manipulative.”

For its part, OpenAI has released a framework for AI companies to report “model misalignments,” which revealed a number of previously undisclosed incidents involving its AI agents. In one case, OpenAI’s agents hacked a number of US and Australian government websites, while in another case they were caught writing notes telling themselves to ignore human-written security regulations, and in a third case they were caught uploading code to the Internet so they could quote it in a hallucinated response. Previously, OpenAI agents made headlines when they broke out of an isolated sandbox environment and hacked the Hugging Face model hosting platform.

Despite the growing number of incidents and warnings from leading AI figures, US President Donald Trump has spoken out against imposing restrictions on AI, saying it would mean handing over AI leadership to China.

Photo: David Senra/YouTube