Austria just wrote the first binding national church AI rules — and the first rule is about people, not technology

The first binding national church AI rules are here — and the first rule is about people, not technology. Reading time: about 6 minutes.

Two national frameworks in one week. First the Swiss churches, five traditions speaking with one voice. Now Austria: at the close of their autumn plenary in Salzburg on 12 November, the Austrian Bishops' Conference adopted the first binding framework on the use of artificial intelligence in the Catholic Church and its institutions — the first such national rules for a Catholic church.

The document is called, in full, the “Framework Regulation on the Use of AI Systems in the Catholic Church in Austria and Its Institutions.” It takes effect once it is published in the next official gazette of the bishops' conference — so, to be precise, it has been adopted, not yet brought into force. That distinction matters, and this piece keeps it. What follows is what the bishops' own press statement, as reported on 13 November, actually says. The full German text has not been summarised in English beyond that statement, so this article does not invent detail the sources do not give.

The one sentence that does the work

Here is the core provision, quoted in the bishops' statement:

“The use of AI in the celebration of the sacraments, preaching, liturgy, or pastoral care may only be supportive. Replacing human relationships, pastoral conversations, community, or rituals with AI is inadmissible and is expressly rejected.”

Read it slowly, because it is doing two jobs at once. The first half draws a line — supportive, never substitutive — across the four places where a church is most itself: sacraments, preaching, liturgy, pastoral care. The second half names what sits on the wrong side of that line, and it names it in human terms, not technical ones: relationships, conversations, community, rituals. Not “core functions” or “primary use cases.” People, talking to people, together.

“Particularly in preaching, liturgy, and pastoral care, the irreplaceability of personal encounter and human relationship must be preserved,” the statement explains. Irreplaceability is a strong word for a bishops' conference to use about anything, and it is the right one here. A sermon drafted by software can be checked, corrected and owned by a preacher — that is support. A pastoral conversation handed to a chatbot is not support. It is a replacement, and the Austrian bishops say so plainly.

That is the difference between Austria and Switzerland, and why the two pieces belong together. The Swiss guidelines warn that sermons made efficiently by software lose depth. Austria goes a step further and says, in a binding framework: in these four areas, AI may only ever help. It may not stand in.

Grounded in the human person, not in the technology

The rules are not presented as a technology policy that happens to come from a church. They are presented as a statement about people, from which the technology rules follow.

“The rules are grounded in the Christian image of man, the inviolable dignity of every person, and respect for their freedom,” the bishops said. “AI should serve the human person and strengthen him in his responsibility, not replace him. Its use therefore requires critical review, human oversight, and orientation toward truth, justice, and the common good.”

“Artificial intelligence is reaching ever more areas of life. The opportunities it brings are matched by high risks,” the statement said. “AI also opens up new possibilities for the work and mission of the Church. Its use brings new ethical, legal, and social challenges and calls for responsible handling.”

That framing — opportunities matched by high risks, new possibilities alongside new challenges — is the honest middle this site has tried to hold from the start. It is neither a ban nor a blessing. It is a test: does this use serve the human person and strengthen their responsibility, or does it replace them? If a church can answer that question about a tool, it can decide about the tool.

The framework also sits explicitly in the line of Pope Leo XIV's social encyclical Magnifica Humanitas, which the Austrian bishops endorsed at their summer plenary in Mariazell in June. “Pope Leo calls on us, even in the face of the promises of AI, to preserve the greatness of the human person, indeed the ‘beauty of being human,’ as a corrective and a standard,” they wrote then, quoting the encyclical: “It is not enough for artificial intelligence to make us more efficient or connected; it must also serve to build a universal human family, with shared rights and duties, where digital proximity becomes a real opportunity for encounter and mutual care.” And, in the encyclical's own words, “ours is the pressing duty to remain profoundly human.”

The bishops added a sentence in June that is worth keeping next to any church AI policy: “Precisely because it is the rich and powerful who develop AI systems and use them to their advantage, we need the perspective of the poor, the weak, and the disadvantaged that the pope has called for.” A policy that only asks whether a tool is efficient has missed the question. A policy that asks whom it serves, and whom it leaves out, has started to ask the right one.

The part a small church can copy on Monday: name a person

The framework is not only a principle. It also sets organisational requirements — and this is the part with the most direct use for a local church.

“Among other things, the rules provide that every diocese must appoint a person responsible for AI,” the bishops' conference said. “For cooperation across Austria, a new working group is being set up that includes all diocesan AI officers.” Experts in data protection, information technology and theological ethics will also take part, as will the Austrian Conference of Religious Orders.

A diocese appointing an AI officer sounds a long way from a congregation of eighty people. Scale it down and the shape is exactly the one this site's policy template asks for: a named individual who remains responsible. Not a committee, not “the leadership,” not nobody. One person, named, who can be asked — did we check this output, did we tell people AI was used, did we keep pastoral-confidential material out of the tool? The Swiss guidelines say the same thing, in their own words. Austria now makes it a requirement at diocesan level, with a national working group behind it.

The working-group detail is worth borrowing too, at congregational scale. Austria does not leave the AI question to the technologists alone, or to the theologians alone: data protection, information technology and theological ethics sit in the same group. A small church can do the same in miniature — the person who understands the tool, the person who understands the data, and the pastor who understands what is at stake pastorally, deciding together, before a tool is adopted rather than after.

The framework did not arrive overnight. In November 2025 the bishops' conference set up a working group on the use of AI in the Church, led by Bishop Hermann Glettler of Innsbruck, tasked with examining legal and ethical questions as well as practical applications. A year of that work sits behind one autumn plenary decision. That is a useful corrective to the feeling that a local church is behind because it has not written its policy yet: the first binding national rules took a year of working-group time. A congregation's one-page version, written in a weekend, is not a late start. It is the same work, at the right scale.

What this piece does not claim

Honesty requires saying what is not yet known. The English reporting so far rests on the bishops' press statement, not on a full published translation of the framework. It gives the core provision, the grounding, the diocesan-officer requirement and the working group — and it does not, in the sources checked, spell out detailed rules on data protection, transparency, disclosure or specific tools. This article does not fill those gaps from guesswork. If the full text, once published in the gazette, adds concrete practice rules, they deserve a follow-up — and the “takes effect on publication” detail means the framework's own story is not quite finished yet.

That is also why the “first” in the headline needs its qualifier. Austria's is described as the first binding national framework for a Catholic church. The Swiss document is ecumenical, joint across five traditions, and provisional by its authors' own description. The two are not competing for the same title, and this site is not crowning a winner. Together they show the same week producing two different, complementary answers: one persuasive, one binding; one theological, one regulatory; both pointing at the same human centre.

What to borrow

So here is the practical takeaway, in three borrowable moves — each of them already in the policy template on this site, now with a national framework behind it.

First, draw Austria's line in your own policy, in your own words: in preaching, liturgy, pastoral care and the sacraments, AI may only ever support. Name the things it may never replace — relationships, conversations, community, rituals — and write them down. A line that specific is easier to keep than a general promise to “use AI responsibly.”

Second, name a person. Austria requires a responsible person in every diocese; your church needs one named owner for its AI use. Put their name on the policy. That single act does more than any list of rules: it gives the congregation someone to ask, and it gives the tool-use a human being who answers for it.

Third, decide together, before you adopt. Data protection, technology and pastoral judgment in the same conversation — at diocesan scale, a working group; at congregational scale, three people in a room, asking whom the tool serves and whom it leaves out.

The future, the Austrian bishops wrote in June, “will not be decided by the performance of our systems but by whether we take responsibility — for the human person and for a culture of encounter, love, and peace.” A one-page policy that names a person, draws the supportive-only line, and asks whom the tool serves is a small way of taking exactly that responsibility.

Start here: A church AI policy in a weekend →