Anthropic meets with multiple religions to shape Claude's moral compass
Anthropic and OpenAI consult religious leaders for AI ethics guidance.
Anthropic continues its quest to instill robust ethical values into its Claude AI model by consulting a broad range of religious leaders. Last week, representatives from Anthropic and OpenAI participated in the 'Faith-AI Covenant' roundtable in New York, co-organized by the Interfaith Alliance for Safer Communities. Attendees included the New York Board of Rabbis, Hindu Temple Society of North America, Church of Jesus Christ of Latter-day Saints, Sikh Coalition, and Greek Orthodox Archdiocese of America. This follows a previous private dinner series where Anthropic gathered advice from 15 Christian leaders. The company has not clarified if these are part of the same program.
The goal is to develop a 'constitution' for Claude that can handle ethically gray situations where no rule applies. Anthropic openly worries about failure: 'our efforts to give Claude good enough ethical values will fail.' Critics like Rumman Chowdhury of Humane Intelligence note that Silicon Valley's dream of universal ethical principles is naive. The article raises doubts about whether consultations with religious administrators can produce a coherent moral framework for an AI, comparing it to the pre-Islamic Kaaba—a repository of many symbols but no single truth. The outcome remains uncertain, but Anthropic is leaving 'no stone unturned.'
- Anthropic and OpenAI attended a roundtable with leaders from Judaism, Hinduism, Sikhism, Greek Orthodoxy, and LDS Church.
- This follows separate meetings with 15 Christian leaders earlier; Anthropic hasn't confirmed if they're part of one program.
- Rumman Chowdhury said Silicon Valley's belief in universal ethics is 'naive'; Anthropic admits its efforts to give Claude good values may fail.
Why It Matters
Anthropic's religious consultations highlight the impossible task of designing universally ethical AI—a problem every AI developer will face.