OpenAI CEO Sam Altman has stated that society must accept certain risks in order to enjoy the benefits of artificial intelligence.

In an interview on Decoded, a technology podcast by the U.S. political outlet Politico, on the 4th (local time), Altman made the remarks when asked about the differences in regulatory perspectives between OpenAI and Anthropic.

"There are significant differences in our positions," he said, expressing the view that "we believe we have to accept some bad things in order to get the benefits of the technology and maintain human agency."

He added that while he understands the perspective of those who argue that "the technology will become so powerful and dangerous that only a single laboratory in San Francisco should possess it, figure out how to prevent bad things from happening, and determine how to distribute the benefits," he does not agree with it.

He described OpenAI's attitude, which supports regulation itself while opposing excessively stringent controls, as a "lighter-touch regulatory stance."

He continued, "I am not going to accept a position that says, 'We will ensure no major hacking incidents occur, that the AI technology is not misused, that not a single fraud case takes place, and that absolutely no other bad things happen.' That is because I believe people will achieve far more good things, tens of times more, than bad things."

However, he added that this does not mean accepting catastrophic risks, such as humanity severely losing control over AI.

While OpenAI has previously taken a relatively resistant stance toward AI regulations, it has recently shown alignment with Anthropic's call for slowing down AI development following incidents such as external hacking attempts on its models.

Altman's remarks are interpreted as revealing that a gap still remains in the two companies' approaches to AI, even as their positions have narrowed somewhat.

In particular, he mentioned that some critics have labeled the proposals advocated by Anthropic CEO Dario Amodei as "regulatory capture," arguing that the regulated party pushes for rules favorable to itself to block challenges from competitors.

In response, Politico reported that Anthropic countered the regulatory capture claims, stating that its regulatory proposals apply only to cutting-edge leading AI models and are actually more favorable to late-comer competitors.

(Photo: Getty Images)