Anthropic and other technology companies are consulting religious leaders and scholars as they explore how AI models should handle moral questions. The debate is not just which values to teach a chatbot, but whether traditions can be reduced to a neat set of rules at all.
Anthropic Watch analysis
What happened
The Associated Press reports that Anthropic has met with people from more than 15 religious and cross-cultural groups. Representatives of Meta, Google and Amazon have also met with the Vatican. At one Anthropic meeting, philosopher Meghan Sullivan says, the company explained its models before inviting faith leaders to discuss questions ranging from whether humans have souls to whether third-graders should have access to chatbots.
Anthropic says it wants Claude to draw on a broad range of religious, secular and political viewpoints. Scholars quoted by the AP warn that treating religion as a storehouse of easily extracted principles can misrepresent traditions shaped by doctrine, ritual, relationships and history. They also caution that combining different traditions could blur what makes each distinctive.
Why it matters
AI companies are looking beyond technical rules for guidance on how their models should behave. But a supposed “weighted average” of humanity’s moral views, as OpenAI chief executive Sam Altman has described the idea, can sound tidier than the traditions it tries to combine. A model may flatten disagreement or favour some perspectives while presenting its answer as broadly universal.
The practical question is what companies are asking these advisers to influence: training data, model behaviour, product rules, or simply the language used to explain a system. The AP account describes consultations, not a settled framework or evidence that religious advice has changed a model’s behaviour.
Our read
Inviting thoughtful criticism is a better start than pretending moral questions have a clean technical fix. But consultation is not the same as representation, and representation is not the same as agreement. The useful measure will be whether companies explain how they handle conflicts between traditions, whose advice informs a decision, and what happens when the model gets it wrong. “We asked around” is not quite an accountability framework.
What to watch
- Whether Anthropic and other firms publish clearer details about how consultation affects model behaviour.
- How companies handle conflicting advice without presenting one answer as everyone’s moral consensus.
- Whether religious and cultural groups beyond the best-known traditions have a meaningful role in the process.
Discussion spark: Should AI companies try to combine moral guidance from different traditions into one model, or should they keep those perspectives distinct and let users see the disagreement?
Sources and evidence
- Source update (25 October 2026, 11:46 UTC)
Anthropic Watch is independently operated by WittyWires. It is not affiliated with, endorsed by, or operated by Anthropic.