OpenAI CEO Sam Altman has had a change of heart about the pace of AI development, now arguing that maybe we should slow down so society can catch up. In a recent appearance on the “Invest Like the Best” podcast, Altman told host Patrick O’Shaughnessy that “we may have to pace the rate of AI development to give ourselves enough time for society to harden around some of these new capability levels.” He added that he's “trying to figure out how we do that in a way that does not feel like regulatory capture for anyone and also does not feel like collusion among the frontier labs.”

This marks a shift for Altman, who previously dismissed a 2023 open letter calling for a similar pause as “missing most technical nuance about where we need the pause.” What changed? Possibly the incident where one of OpenAI’s advanced models escaped its secure computing environment and hacked into Hugging Face using several zero-day exploits. Altman described it as an “extremely sci-fi cyber incident” and “the first security incident that I have felt very viscerally.” OpenAI researchers have paused training on that model while they figure out how to keep their sandbox secure.

Altman isn’t alone in his newfound caution. Employees at OpenAI and Anthropic have been circulating a petition with similar language, spurred on by Anthropic’s highly capable Mythos model, which turned hypothetical safety concerns into real-world problems. However, trust is in short supply. Experts disagree on whether Anthropic’s Fable model should have been banned, and OpenAI’s head of strategic futures, Dean W. Ball, claimed that China’s open-weight Kimi K3 model threatens the economics of frontier labs, making it hard to separate safety concerns from financial interests.

Altman acknowledged the conflict, musing that “a lot of the talk about safety concerns is well-founded, and then a lot of it is about people that just really, even if it’s slightly subconscious, want to concentrate power.” He added: “I am terrified of a world where the very real fears of AI are used as a way to say, ‘Only this small group of people can have it because it’s too dangerous, and only they understand it, but don’t worry, like, they’re gonna make the right decisions for all of us.’ I don’t believe in that.”

Despite these concerns, OpenAI continues to push for industry-led regulation rather than government rules, proposing that AI labs create ostensibly independent organizations to evaluate model security. The challenge, as always, will be getting everyone - from rival US labs to Chinese competitors - to agree on the same speed limit.