Sam Altman just turned a philosophical debate into a public safety warning. On October 3, the OpenAI CEO posted on X: “I am very uncomfortable about people trying to ascribe religious force or a surrender of human judgment to AI models, and think it is a real safety issue.” He didn’t name anyone. He didn’t have to.
What happened
The post landed days after a New York Times report (Sept 29, 2026) revealed that Anthropic co-founder Chris Olah has spent about a year meeting privately with religious and philosophical thinkers — Catholic, Jewish, Sikh, evangelical and others, many under non-disclosure agreements — to discuss whether Claude might be conscious and how to give it “moral formation.”
According to the Times, Olah saw an advance copy of Pope Leo XIV’s AI encyclical, Magnifica Humanitas, before its May 25 launch, proposed pulling Anthropic out of the rollout over the text’s rejection of machine consciousness, then privately lobbied the pope’s advisers. The final text kept its line: AI systems “do not undergo experiences, do not possess a body, do not feel joy or pain” and have no moral conscience. Olah still attended the launch and spoke on the dais.
Anthropic does not claim Claude is conscious. Its internal constitution for Claude — sometimes called the “Soul Doc” — says the company is “not sure whether Claude is a moral patient,” and calls the question “live enough to warrant caution.”
Axios described Altman’s post as a veiled shot at Anthropic. Neither company immediately responded to requests for comment.
The wider split
Altman isn’t the only one pushing back. Microsoft AI chief Mustafa Suleyman has separately warned that treating AI as if it has feelings could be dangerous, and said Microsoft is working on a code of conduct meant to keep humans “always in control.”
Why it matters
This isn’t just CEO drama. As AI systems get more capable, two questions are colliding: how much moral weight we give them, and how much of our own judgment we hand over to them. Altman is warning about the second — treating a model as a source of authority is itself a failure mode. Anthropic’s camp is wrestling with the first — whether some form of moral consideration might be owed. Both will shape how these systems get built, trained, and trusted.
AiPost’s take
There is something revealing about the most powerful people in AI arguing about souls. The practical upshot is simpler, though: however you feel about AI consciousness, surrendering your own judgment to a model is the one thing you control. Verify, don’t worship.
Comments
Post a Comment