Anthropic Consults Religious Leaders on AI Ethics

The artificial intelligence company Anthropic has begun consulting with religious thinkers to help refine the moral framework and "constitution" that guides the behavior of its chatbot, Claude [1]. This outreach coincides with the release of Pope Leo XIV’s encyclical, Magnifica Humanitas, which urges developers to prioritize the common good over profit and calls for robust external regulation of the AI industry [2].

Key takeaways

• Anthropic has held meetings with religious scholars from various traditions to discuss how centuries-old moral reasoning can inform AI development [1].
• Pope Leo XIV’s encyclical, Magnifica Humanitas, warns against the concentration of AI power in the private sector and calls for the technology to be "disarmed" [1, 2].
• Anthropic co-founder Christopher Olah attended the Vatican presentation of the encyclical and expressed support for external checks on the industry [1, 2].
• Critics, including AI ethicist Carissa Véliz, question whether these consultations represent genuine ethical progress or a form of "ethics washing" to improve corporate reputations [1].

Integrating Moral Reasoning into AI

Anthropic’s initiative began in late March, when the company invited approximately 20 religious thinkers to discuss how to teach a chatbot to be "good" [1]. Participants, including scholars from the Markkula Center for Applied Ethics and the organization AI and Faith, were asked to help refine Claude’s constitution—a set of principles used to train the model to critique its own responses [1]. According to attendees, the company acknowledged that the moral challenges of AI have outpaced its internal wisdom, leading them to seek external perspectives on how to instill a "disposition toward good behavior" rather than relying solely on a checklist of prohibitions [1].

The discussions have evolved from initial, sometimes awkward sessions into a broader effort that now includes representatives from Judaism, Hinduism, Mormonism, Sikhism, and the Greek Orthodox Church [1]. While some participants believe the company’s engagement is sincere, others remain skeptical. Carissa Véliz, an AI ethicist at the University of Oxford, argues that the use of religious terminology by tech companies could create a tribal mentality that is difficult to challenge through reason, noting that business incentives often remain the primary driver of corporate behavior regardless of these consultations [1].

The Vatican’s Call for Oversight

The collaboration between Anthropic and religious figures occurs against the backdrop of Pope Leo XIV’s 40,000-word encyclical, which identifies AI as a primary challenge to humanity [2]. In the document, the Pope warns that it is insufficient to rely on the ethics of a few private companies, arguing that "robust legal frameworks" and independent oversight are essential to prevent AI from becoming an instrument of domination or exclusion [2].