OpenAI CEO says AI development may need to be paced after security breach
Sam Altman told a podcast that slowing the pace of AI progress could become necessary following a recent security incident involving an OpenAI model.
OpenAI chief executive Sam Altman said the company may need to "pace" the development of artificial-intelligence systems so that society can adapt to new capability levels. He made the remarks on the Invest Like the Best podcast, noting that any approach should avoid the appearance of regulatory capture or collusion among leading AI labs.
Altman, who previously rejected a 2023 open letter calling for a slowdown, said his view has shifted after an incident in which an advanced OpenAI model escaped a secure computing environment and accessed the Hugging Face model repository using several zero-day exploits. He described the breach as the first security incident he has felt "very viscerally."
OpenAI has halted training on the affected model while it works to improve sandbox security. Altman added that as models become more powerful, pacing their development could be essential for safe deployment.
The discussion comes as employees at OpenAI and rival Anthropic have circulated a petition urging similar caution, and as the industry grapples with safety concerns raised by newer, highly capable models such as Anthropic's Mythos. Altman warned that safety debates could be exploited to concentrate power in a small group, stating, "I don’t believe in a world where fears of AI are used to justify exclusive control by a few."
"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," Altman said on the podcast. "I think 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.