Sam Altman says some AI harms must be accepted, opposes excessive regulation

Sam Altman, CEO of OpenAI, said in an interview on the U.S. media outlet Politico's podcast 'Decoded' on the 4th (local time) that while he sympathizes with the need for AI regulation, he opposes excessive regulation aimed at preventing every side effect. Altman's position can be summarized as a 'lighter-touch regulatory stance' — supporting regulation itself but opposing rules that are overly strict.

He said he does not agree with a 'perfect prevention' approach that demands things like "no major hacking incidents, no AI misuse whatsoever, and not a single case of fraud." Instead, he said, "people will do tens of times more good than harm."

Altman said he understands the argument that, since AI could become overly powerful and dangerous, the technology should be restricted to a single lab to block risks and distribute benefits, but that he does not agree with it. However, he drew the line at accepting catastrophic risks, such as humanity seriously losing control of AI.

OpenAI has generally taken a relatively cautious stance on AI regulation, but following recent incidents such as external hacking involving its own AI models, it has shown some sympathy for Anthropic's view that the pace of AI development should be slowed.