Altman: don't treat AI like a religious force

OpenAI CEO Sam Altman said he is very uncomfortable with people treating AI models as a religious force or handing over human judgment to them. He called that pattern a real safety issue. The comment landed on X on 3 October 2026 and quickly drew coverage from Business Insider, Times of India, and NDTV.

What happened

Altman posted that he is "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." The wording is careful: he is not only warning about hype, but about a mindset that treats models as something to defer to rather than tools to check.

The timing matters. His post followed New York Times reporting that Anthropic co-founder Chris Olah had met religious leaders to discuss whether Claude might be conscious and how to build morality into AI. That reporting put questions about machine consciousness and moral design into the mainstream news cycle. Altman's reply frames the other side of that debate: if people start treating AI as sacred or as a substitute for their own judgment, that itself becomes a safety problem.

Business Insider covered the post under a clear headline: do not treat AI like a religious force. Other outlets, including Times of India and NDTV, repeated the same quote and context.

Why it matters

Safety talk usually centers on model behavior: jailbreaks, misuse, or uncontrolled agents. Altman is pointing at a human failure mode. When users stop questioning outputs, stop verifying claims, or treat a chatbot as an authority with moral weight, errors and manipulation get easier, not harder. That risk grows as models sound more confident and more human.

The Anthropic thread adds fuel. Talks about consciousness and morality can be serious research. They can also invite a public that already anthropomorphizes chatbots to go further and treat them as beings to trust or obey. Labs, media, and users all share that framing risk.

Dany's take

I agree with the core warning. AI can be useful and still deserve zero worship. Keep human judgment in the loop: ask for sources, double-check high-stakes advice, and do not outsource your values to a model card. Curiosity about consciousness is fine. Surrendering judgment is not. I will keep watching how OpenAI, Anthropic, and the press talk about these ideas, because the language we use shapes how people behave around the tools.

Source: Business Insider: Sam Altman don't treat AI like a religious force.