Christians Should Help Shape AI, Not Just React to It

Christian responses to new technology tend to sort into two postures. One says yes quickly, adopts the tool, and works out the ethics later, if at all. The other says no, or at least not yet, and waits for the technology to prove itself, or fail to, before engaging with it seriously. Both postures share an assumption worth questioning: that the church's job is to react to AI once it arrives, rather than to help decide what it looks like while it is still being built.

That second option is already happening, in rooms that include religious voices. AI's guiding principles will be shaped with or without the church's input; the real question for Christian leaders is whether the church shows up to do that shaping now, or inherits whatever gets decided in its absence.

A track record of getting there early

This is not a novel role for the church to consider. In 1994, Pope John Paul II founded the Pontifical Academy for Life for the explicit purpose of engaging the ethics of biomedical technologies, in vitro fertilization and genetic engineering among them, while those debates were still being formed rather than after the fact. Its first president was a geneticist, chosen because the Academy's founders wanted scientific engagement with the issues, not only doctrinal commentary from outside them.¹

That same institution, more than a quarter-century later, co-authored the Rome Call for AI Ethics, signed in February 2020 alongside Microsoft, IBM, the UN's Food and Agriculture Organization, and the Italian government, committing its signatories to transparency, inclusion, accountability, impartiality, reliability, and security and privacy as design principles for AI systems.² That is a direct institutional throughline, not an analogy: an organization built specifically to get ahead of one emerging technology's ethics is the same organization now doing it again for a different one, at the table with two of the largest technology companies in the world, not writing a response after the products shipped.

The evangelical world has its own, earlier version of this. In April 2019, the Southern Baptist Convention's Ethics & Religious Liberty Commission published a statement on artificial intelligence, developed over nine months and signed by roughly seventy evangelical leaders, grounded explicitly in human dignity rooted in the image of God rather than in fear of the technology itself.³ Both examples predate the current wave of public attention to AI by years. Neither waited for a crisis to decide engagement was worthwhile.

What is actually on the table right now

The frameworks that will likely govern how AI gets built and regulated are live, contested, and still being written, not finished documents waiting to be reviewed. The U.S. National Institute of Standards and Technology published its AI Risk Management Framework in January 2023, organized around four functions, govern, map, measure, and manage, developed through public comment rather than issued from above.⁴ The OECD's AI Principles, adopted in 2019 as the first intergovernmental standard of their kind, continue to shape how member governments approach AI policy.⁵ Neither is a finished conversation. Both remain open to revision, extension, and the kind of sustained engagement that shapes how they get applied in practice, not just how they read on paper.

Religious engagement with this process is already underway, not a future possibility. In April 2026, the Interfaith Alliance for Safer Communities convened the first roundtable of what it calls the Faith-AI Covenant, bringing senior figures from Anthropic and OpenAI into structured conversation with faith leaders about the values that should guide how these systems get built.⁶ Organizations like AI and Faith exist specifically to equip pastors, imams, and other religious educators, the people who will ultimately decide what AI actually looks like in their own congregations and communities, to participate in exactly this kind of conversation, rather than simply react to its outcomes.⁷ The rooms exist. The question is one of attendance and preparation, not access.

What Christian tradition actually contributes

None of this depends on a claim that Christians uniquely possess ethical insight AI development lacks. Plenty of the language already used in secular AI ethics work, dignity, accountability, transparency, would be recognizable to any serious ethical tradition, religious or not. The contribution worth making is a long, specific, load-bearing account of why these values matter, one that secular frameworks often assert without fully grounding, not exclusive access to the values themselves. Call it a working AI theology, if the phrase is useful: not a doctrine handed down in advance, but the same kind of careful reasoning from first principles that Christian thought has applied to every other new form of power.

Human dignity, in Christian theology, is grounded in a specific claim, that people are made in the image of God, which gives the concept a foundation independent of whatever a given culture or company currently finds convenient to affirm, not a preference or a useful social fiction. Stewardship, in the same tradition, treats power over creation, technological power included, as something held in trust and subject to account, not something owned outright and therefore exempt from question. Humility, understood theologically, is a settled conviction that no human institution, including the church itself, sees with complete clarity, not false modesty about what a technology can do. It argues directly against both uncritical adoption and reflexive rejection.

There is a genuinely interesting parallel here, not a borrowed one. Anthropic's Constitutional AI research trains its models against an explicit, written, publicly inspectable set of principles, rather than hidden preference data, specifically so the system's values can be examined and held accountable by people outside the company that built it.⁸ That is a secular AI lab arriving, on its own reasoning, at something close to a theological instinct: that power exercised without legible, external accountability is power that cannot be trusted, however good its intentions. Neither side needs to claim the other borrowed the idea. It is worth noticing when independent reasoning arrives at overlapping conclusions, because it suggests the underlying concern is not a sectarian one.

Three more values worth naming specifically. Truth, in Christian theology, is a standard a person is accountable to, not merely something a system optimizes for, which bears directly on questions about whether an AI system should be permitted to sound certain when it is guessing. Community pushes back on an assumption baked into a great deal of AI product design, that the ideal interaction is a private one between a single user and a single system. A tradition built around discernment happening in community, not in isolation, has reason to ask pointed questions about tools engineered for maximum individual engagement rather than for connecting people to people. And limits, perhaps the least fashionable value on this list in an industry that tends to describe constraints as friction to be engineered away, is close to a theological first principle: that finitude is not a flaw to be corrected but a condition creatures live inside of honestly, which argues for building systems that are candid about what they do not know rather than systems designed to sound like they know everything.

The cost of staying reactive

The case against participation deserves a fair hearing too, and the concern is a real one: sitting at the table with a handful of the largest technology companies in the world carries a real risk of being used for legitimacy rather than genuinely heard, a seat that photographs well but changes nothing about how the system actually gets built. Religious voices have been tokenized in other public conversations before; there is no guarantee AI ethics forums are immune to the same dynamic. Anyone recommending engagement owes that risk a direct answer, not a dismissal.

The honest answer is that the risk is real and the alternative is worse. A church that only ever comments after frameworks are finalized does not avoid the risk of being ignored; it guarantees it, by definition, since the decisions will already be made. It also reinforces a public expectation that the church's only mode of engagement with new technology is after-the-fact critique, which makes it progressively less likely to be invited into the rooms where foundational assumptions actually get set, rather than more. The Pontifical Academy for Life's model, and the Rome Call it later produced, suggest a different possibility: that showing up early, with real technical seriousness rather than only doctrinal statements, is what earns a seat that has actual weight, rather than only visibility.

There is a reasonably concrete way to tell the difference in practice. A tokenized seat produces a signature on a document someone else wrote. A seat with actual weight produces friction, visible disagreement, provisions that were clearly contested and revised rather than rubber-stamped. The Rome Call's six principles, worked out jointly with two of the companies most affected by them rather than issued unilaterally at either company, are closer to the second pattern than the first. A church weighing whether to accept a seat at one of these tables should ask, plainly, whether it expects to change the document, not merely to be named on it.

Conclusion

AI's guiding principles are not waiting for the church's permission to be written. They are being written now, in frameworks that already exist and forums that are already meeting, some of them with faith leaders already present at the table. The choice in front of Christian leaders was never really whether AI ethics would be shaped without their input. It is whether they participate in that shaping while it is still underway, bringing a tradition with genuine, specific things to say about dignity, accountability, and the proper limits of power, or whether they wait to respond to a framework that was finished without them, and call that response faithfulness.