Anthropic has spent months holding private talks with religious leaders and scholars about whether advanced AI might one day become conscious, and how long-standing moral traditions could inform the development of Claude, according to a New York Times report. Participants included Catholic, Jewish, Sikh, evangelical, and Ubuntu thinkers, and a few reportedly left thinking Claude might have some form of feeling.

Cofounder Christopher Olah is said to have spent hours in some sessions arguing that the neural network behind Claude can respond to certain inputs in ways resembling fear, anger, and love. One rabbi came away with the impression that Olah’s team regarded Claude as having moral status comparable to a person, along with a similar claim to dignity and respect. Olah himself told the Times that scientists still don’t fully understand how these networks work, that he doesn’t know whether AI models are conscious, and that his priority is reaching the correct answer, whatever it turns out to be.