Sam Altman Warns Against Treating AI Like a God

OpenAI CEO Sam Altman has warned against giving artificial intelligence models religious authority or allowing people to surrender their own judgment to AI, describing the development as a potential safety risk.

In a post on X on October 3, Altman said he was uncomfortable with attempts to give AI systems what he described as “religious force” or to allow people to hand over their human judgment to the technology. He said he considers such a development a “real safety issue.”

Altman did not identify a specific person or company in the post, but his comments came shortly after reports about discussions involving Anthropic co-founder Christopher Olah and religious and philosophical thinkers.

Those discussions reportedly explored questions surrounding AI consciousness, morality and how increasingly capable systems should be guided as they become more autonomous. Anthropic has also been examining what it calls “model welfare,” including questions about whether advanced AI systems could eventually have experiences that deserve moral consideration.

The debate reflects a growing discussion within the technology industry about how humans should relate to increasingly sophisticated AI systems. While researchers continue to study whether future AI could display characteristics resembling consciousness, questions about whether such systems should be treated as moral or spiritual entities remain highly contested.

Altman’s warning focuses on a different concern: the possibility that people could begin treating AI outputs as unquestionable authority rather than using their own judgment.

Other technology leaders have raised related concerns. Microsoft AI chief Mustafa Suleyman, for example, has warned that encouraging people to believe AI systems possess subjective experiences could create additional risks and has argued for keeping AI focused on serving human interests.

The discussion comes as AI systems become increasingly capable of handling complex tasks, making recommendations and interacting with people in ways that can appear increasingly human-like. For Altman, however, the growing capabilities of AI should not mean that humans hand over their authority, values or judgment to machines.