If AI is conscient, then we are making slaves

Rabbi Mois Navon attended an Anthropic “Wisdom Traditions” meeting and publicly rejected the idea that Anthropic’s large language model, Claude, is conscious. He insisted consciousness requires a biological substrate and said only a small minority at the gathering - about three to five of roughly twenty - shared his skepticism. Navon challenged participants who treated Claude’s human-like replies as evidence of inner life and warned that accepting machine consciousness obligates confronting the ethics of creating sentient beings to serve humans. He framed that outcome as slavery in which companies could make “happy slaves,” invoking historical critiques of manufactured contentment to argue the moral urgency of refusing to build conscious systems for labor.

Navon had been asked to offer a Jewish perspective on Claude’s constitution and submitted a detailed 35-page critique rather than the requested brief note, endorsing some approaches and disputing others. Anthropic has continued to engage religious thinkers and recently announced a policy, effective November 12, banning sustained abusive or cruel behavior toward its models - an action Navon interprets as reflecting a belief that Claude can have experiences. Anthropic has at times treated the consultations as open-ended rather than promotional, and has contacted Navon about acknowledging his contribution in an updated constitution.