AI safety concerns intensified after an Anthropic employee warned on X that the technology could destroy humanity and accused major frontier labs of "gambling with our lives." The debate centers partly on p(doom), the publicly discussed probability of an AI-driven catastrophe: Elon Musk has put his estimate at 10% to 20%, while Anthropic CEO Dario Amodei has cited 10% to 25%. In an exclusive hour-long interview on Friday, OpenAI’s Sam Altman discussed AI’s risks to humanity, the pace of model development, whether more powerful systems can be controlled and what he would do if safe development proved impossible. Altman said the company’s IPO would be "ill-timed" amid safety concerns and would not take place until 2027. He said no gamble with humanity is acceptable and that he would be willing to challenge investors by pausing or stopping AI development if necessary. He also addressed national and international regulation and referred to a possible pact with industry peers to slow development while safety and AI alignment solutions catch up with model capabilities.