Anthropic is working with religious thinkers to help guide the moral development of its AI model, Claude. The company's effort to bring in a range of theological perspectives is aimed at building an ethical framework that doesn't rely on a single tradition or worldview.
The move stands out in an industry where AI ethics has often been treated as a technical or legal puzzle. By inviting religious scholars into the conversation, Anthropic is testing whether pluralism — the idea that multiple moral systems can coexist — can be baked into a product used by millions.
Why bring religion into AI ethics
AI models like Claude are trained on vast amounts of text and then fine-tuned to follow certain values. Those values usually come from internal policy teams, user feedback, and legal requirements. They rarely come from the pew or the pulpit.
Anthropic's decision to engage religious thinkers suggests the company sees a gap. Technical guidelines can tell you what's allowed and what isn't, but they don't always settle deeper questions about dignity, mercy, or the common good. Different faith traditions have spent centuries arguing about those questions, and Anthropic appears to want that argument in the room.
The company hasn't said which specific religious groups or individuals it's talking to, or what the format looks like. But the effort is described as ongoing engagement, not a one-off consultation.
A test case for ethical pluralism
If Anthropic succeeds, it could set a precedent. Most AI ethics frameworks today are secular and procedural — they focus on harm reduction, transparency, and fairness. Adding religious voices doesn't replace those goals, but it could widen the circle of people who feel represented by the technology.
That matters for trust. When users from different backgrounds see their moral traditions reflected in how an AI answers questions about family, work, or end-of-life care, they're more likely to accept the system. When they don't, the opposite happens.
Pluralism isn't easy, though. Religious traditions disagree with each other, sometimes sharply. An AI model can't endorse every position at once. It has to choose, or at least find a way to acknowledge disagreement without pretending it doesn't exist. That's a design problem as much as a philosophical one.
What Anthropic gets out of it
Anthropic is one of several companies racing to build safe, reliable AI. Its competitors have their own ethics boards and advisory councils, often stocked with technologists, lawyers, and civil society groups. Religious thinkers are a less common ingredient.
By adding them, Anthropic may be trying to differentiate itself. It may also be responding to a real demand: people want AI that aligns with their values, and for many, those values are religious.
The company hasn't promised any specific changes to Claude based on these discussions. The engagement is described as shaping the model's moral compass — a phrase that suggests influence over time, not a quick patch.
Anthropic hasn't set a public timeline for when this work will show up in Claude's behavior, or how it will measure success. There's no published list of participants, and no details on whether the religious thinkers are paid advisers, informal consultants, or something else.
What is clear is that the company is treating moral pluralism as an engineering input, not just a PR talking point. If that approach spreads, other AI labs may face pressure to do the same — and to explain why they didn't.
For now, the next concrete step is more engagement. Anthropic says it's continuing the conversations, but it hasn't said when or how the results will be made public.
