Somewhere between the ancient question of how a good person should act and the new problem of how a machine should decide, the technology industry has begun knocking on the doors of religion. Companies building artificial intelligence are consulting theologians and religious traditions, hoping that centuries of reflection on duty, harm and mercy can give their systems a moral footing deeper than a corporate policy manual. Yet the traditions they turn to are plural and disputed, and the choice of which voice shapes a machine is itself a moral act that the companies are now making on behalf of e
Tech Firms Turn to Religion to Teach AI Morality, but Experts Question the Approach
Choosing which tradition shapes a model is itself a moral decision.
Mark
Why would an AI company go to religion at all? Isn't that an odd place to look?
Mimi
Less odd than it sounds. Religious traditions have long worked on questions of duty and harm, and the companies are consulting theologians and thinkers for that reason.
Luke
Careful, though. What we can confirm is the trend as AP describes it. We don't have named companies or named advisers here.
Mark
So what's the objection from the experts?
Mimi
That it's more complicated than it sounds. The moral frameworks inside religion are diverse and contested, so there's no single set of values to simply load in.
Luke
And that's the experts' view, not a measured result. Nobody in this material shows a model that behaves differently because of it.
Mark
Does choosing a tradition become its own moral choice?
Mimi
Yes, and the companies are the ones making it, which is part of why critics are uneasy.
Luke
That's a reasonable inference from the objection, but I'd flag it as inference. The reporting doesn't say how any firm decides.
Mark
What should a reader watch for next?
Mimi
Whether behavior actually changes in measurable ways, and how regulators react.
Luke
Agreed. Until there's evidence on either, this is a story about intent, not outcome.
Le Pouls
- Tech firms are reaching into centuries of religious moral thought, searching for a foundation sturdy enough to steady their AI systems.
- Critics warn that no single faith speaks with one voice, so every choice of tradition risks quietly picking winners in a contested moral debate.
- Companies are trying to translate ancient teachings on harm, duty and mercy into guidance a model can actually follow.
- The approach is still unproven, and whether it changes how AI behaves in practice, and how regulators respond, remains the open question.
Somewhere between the ancient question of how a good person should act and the new problem of how a machine should decide, the technology industry has begun knocking on the doors of religion. Companies building artificial intelligence are consulting theologians and religious traditions, hoping that centuries of reflection on duty, harm and mercy can give their systems a moral footing deeper than a corporate policy manual. Yet the traditions they turn to are plural and disputed, and the choice of which voice shapes a machine is itself a moral act that the companies are now making on behalf of everyone who asks them questions.
The Associated Press reports that artificial intelligence companies are increasingly turning to religious thinkers and traditions as they try to instill moral behavior in their models. The impulse is understandable. Faith traditions have spent centuries working through questions of duty, harm, mercy and obligation, and their frameworks offer a depth that engineers could hardly build from nothing. As AI systems take on questions about health, family and conflict, grounding them in something older than a corporate policy document carries a certain appeal.
Experts quoted in the coverage are not convinced it is a simple fix. Religious moral thought is diverse and frequently contested, both between faiths and within them. Deciding which tradition, interpretation or authority should guide a model is itself a moral choice, and it is one the companies are making.
The available summary does not say which firms are consulting which thinkers, how many are involved, or what has changed in any specific product. The story is best read as an account of a trend. Two things bear watching: whether these approaches produce measurable differences in how AI actually behaves, and how regulators respond once a company's moral inspiration begins shaping what millions of users are told.
The Associated Press headline says a good deal, and the material behind it says little more. Technology companies building artificial intelligence are turning to religion to help make their systems morally good. Some experts say the approach is not as simple as it sounds.
That is the extent of what the available reporting establishes, and it is worth stating plainly. The AP piece, published on October 25, 2026, describes AI firms increasingly consulting religious thinkers and traditions as they try to instill moral behavior in their models. Theologians are among those the companies are looking to for guidance on how an AI system should reason about right and wrong.
The appeal is easy to understand. Religious traditions have spent centuries working through questions of duty, harm, mercy and obligation, and they offer developed frameworks that engineers could not easily write from scratch. For companies whose products now answer questions on health, family and conflict, drawing on that inheritance is a way to ground their models in something older than a corporate policy document.
The experts quoted in the coverage push back on the premise that this is a straightforward fix. Their central objection, as the reporting frames it, is that encoding morality from religion is far more complicated than it sounds. Religious moral frameworks are diverse, and they are often contested, both between faiths and within them. Choosing which tradition, which interpretation or which authority should shape a model is itself a moral decision, and the companies are making it.
The summary of the piece does not name which companies are consulting which thinkers, how many are involved, or what changes have been made to any specific model. Those details are not in the material available here, so they cannot be reported with confidence. The account is best read as an analysis of a trend rather than a record of a particular deal or product decision.
Two things are worth watching from here. The first is whether these faith-informed approaches produce measurable differences in how AI systems actually behave, rather than simply appearing in company statements. The second is how regulators respond, since a company's choice of moral inspiration may draw scrutiny once it affects what millions of users are told.
Citations marquantes
Encoding morality from religion is far more complicated than it sounds, some experts argue.— Experts cited in AP reporting (paraphrase)
Envie de l'histoire complète ?
Lire l'original sur Associated Press ↗
Technology