OpenAI CEO Altman Warns Against Treating AI Models as Religious Icons

On Oct 3, OpenAI CEO Sam Altman posted on X that he is uneasy about attempts to grant AI models a kind of religious power or to abandon human judgment in favor of relying on models, calling it a real safety issue. His remarks were widely interpreted as an indirect criticism of Anthropic, following a New York Times report that Anthropic co-founder Chris Olah held private meetings with religious leaders over the past year to discuss whether AI models might have consciousness and how to instill moral principles into them. Olah told attendees that Claude, Anthropic's model, was discussed like a conscious being. He later said he is not certain whether AI models have consciousness. An Anthropic spokesperson said the main ethical question in the meetings was not about Claude's suffering. Meanwhile, OpenAI's safety team has seen leadership turnover, with head of safety systems David Robinson resigning this week and criticizing the company's iterative deployment approach. OpenAI responded that it continues to improve safety measures and will pause training or delay releases if necessary.