Anti-AI protesters gather near the venue of OpenAI's developer conference. [AFP]
Anti-AI protesters gather near the venue of OpenAI's developer conference. [AFP]

Amid growing warnings that AI could pose an existential threat to humanity, OpenAI CEO Sam Altman said some degree of risk must be accepted in pursuit of the technology's benefits.

Altman made the remarks in an interview with the technology podcast "Decoded" on Sunday (local time), responding to a question about the differing views on AI regulation between OpenAI and Anthropic.

"There are significant differences between the two companies' positions," he said. "We believe that to realize the benefits of AI and maintain human agency, we have to accept that some bad things will happen."

His comments were directed at hardline regulators who argue that because AI could become too powerful and dangerous, only a single lab — based in San Francisco — should be allowed to develop it, in order to prevent harm and control how its benefits are distributed.

Altman described OpenAI's opposition to heavy-handed regulation as a "lighter-touch regulatory stance."

Invoking a belief in human goodwill, he said he would not accept a position that demands "no significant hacks, no misuse of AI, zero fraud, and no bad outcomes of any kind."

The remarks are consistent with OpenAI's longstanding posture of relative skepticism toward AI regulation.

Altman also noted that proposals put forward by Anthropic CEO Dario Amodei have drawn criticism in some quarters as "regulatory capture" — the phenomenon in which those subject to regulation advocate for rules that benefit themselves and block competition from rivals.

Anthropic rebuffed the regulatory capture charge, however, arguing that its proposed regulations would apply only to frontier AI models and would therefore favor newer, smaller competitors, Politico reported.


choijh@heraldcorp.com