Anthropic turns to religious thinkers to help shape Claude’s moral compass

Anthropic, the company behind the Claude AI model, is hiring religious thinkers from around the world to help build its AI’s moral guide, according to The New York Times.

The report arrives after a string of meetings between Anthropic and faith leaders earlier this year. Those meetings suggest the company’s interest in religious perspectives is part of an ongoing effort, not a one-time publicity move.

From boardrooms to theology seminars

The most concrete example came in late March 2026. Anthropic convened fifteen Christian leaders to discuss the ethics of artificial intelligence.

Catholic thinkers, in particular, have had a hand in the ethical principles found in what Anthropic calls Claude’s “Constitution.” That document serves as the model’s written set of values and guidelines, the rulebook that tells Claude how to reason through difficult requests.

Anthropic’s stated aim is for Claude to act as a “good, wise, and virtuous agent.” The company’s approach avoids locking the model into a single ethical doctrine.

The outreach continued into spring. In late April 2026, Anthropic executives attended the inaugural “Faith-AI Covenant” roundtable, where they sat alongside leaders from OpenAI and representatives of various faith organizations.

Whether the Times-reported hiring push has taken the form of a formal program remains unclear. Anthropic’s documented engagement so far has come mainly through convenings, consultations, and roundtables.

Why the Vatican keeps showing up

Anthropic’s conversations with religious institutions did not start this year. Earlier engagement followed the Vatican’s 2025 note on artificial intelligence, titled “Antiqua et Nova.”

A May 2026 report described Silicon Valley as increasingly leaning on the Catholic Church for guidance on thorny AI ethics questions.

What this means for AI development

For Anthropic, the consultations fit neatly within its constitutional AI approach. The idea is to set moral guardrails for the model through written principles, and religious expertise is now one of the inputs shaping those principles.

By drawing on religious thinkers, Anthropic is betting that long-standing moral traditions can help an AI handle those moments with more nuance. By refusing to commit to one framework, it is also trying to avoid building a model that feels like it belongs to a single faith or worldview.

The documented meetings so far have leaned heavily toward Christian and specifically Catholic voices. If the Times report is accurate and the search truly spans the globe, the real test will be how widely Anthropic casts its net across other traditions and secular philosophies.

OpenAI leaders joined the same Faith-AI Covenant roundtable, which suggests the major labs may start treating ethical credibility as a point of differentiation.