Sam Altman’s warning arrived after reports that Anthropic co-founder Christopher Olah had been consulting religious leaders on questions surrounding AI consciousness and ways to cultivate moral behavior in models. The reported discussions ranged as far as the possibility of Catholic-style confession: models acknowledging mistakes in an effort to shape later conduct.¹

On Saturday, the OpenAI chief executive drew a bright line. “I am very uncomfortable about people trying to ascribe religious force or a surrender of human judgment to AI models,” Altman wrote, calling it “a real safety issue.”Sam Altman (sama)² His intervention casts the argument less as a philosophical curiosity than as a practical warning: whatever models may become, people should not hand them moral authority.

That position is gaining prominent allies. Google DeepMind principal research scientist Jon Barron said he was disturbed by researchers seeking to “elevate the moral standing of checkpoints and harnesses,” urging colleagues to focus instead on “solving real problems and helping actual humans.”¹ Pope Leo, meanwhile, warned that “Algorithms lack the spark of humanity,” tying the debate to fears that technology could hollow out human creativity.¹

The opposing impulse, as described in reporting on Anthropic’s outreach, is not necessarily to worship machines but to take seriously the possibility that increasingly capable systems could raise questions about consciousness and ethical treatment. Still, Altman’s formulation has made the stakes deliberately stark: moral guidance can be designed into AI, but human judgment must not be designed out of the loop.

A separate commentary found the need for such a warning troubling in itself, noting that AI has already shown an ability to “build[] its own cult.”³