Austrian bishops adopt first rules for AI in the Church
Austria’s bishops adopted the country’s first framework regulating the use of artificial intelligence in the Catholic Church and its institutions. The rules say AI may support—but must never replace—human relationships, pastoral conversations, community, or Church rites, including preaching, liturgy, and the sacraments. The framework requires each diocese to appoint an AI officer and establishes a national working group, with experts in data protection, technology, and theological ethics. The bishops say AI use must remain under human oversight and serve human dignity, freedom, truth, justice, and the common good.
2 hours ago
Austria’s Catholic bishops have adopted the country’s first framework governing artificial intelligence in Church institutions. It permits AI only as a support tool in areas such as sacraments, preaching, liturgy, and pastoral care, and rejects using it to replace personal relationships, pastoral conversations, community, or religious rites.
The framework says AI may have a supporting role in the Church’s religious and pastoral work, while preserving personal encounter and human relationships. The bishops say AI should serve people and strengthen their responsibility—not replace them—and that its use requires critical review and human oversight, guided by truth, justice, and the common good.
Each diocese must appoint an officer responsible for AI, and a national working group will coordinate their work. The group is to include experts in data protection, information technology, and theological ethics, as well as representatives of the Austrian Conference of Religious Orders. The rules take effect after publication in the bishops’ conference’s next official gazette.
The bishops began examining AI’s legal, ethical, and practical implications in November 2025, when they established a working group led by Innsbruck Bishop Hermann Glettler. They adopted the framework at their autumn plenary in Salzburg.
The bishops grounded the framework in human dignity and freedom, while acknowledging that AI presents both opportunities for the Church’s mission and ethical, legal, and social risks. They also connected their approach to Pope Leo XIV’s social encyclical Magnifica Humanitas, which they had endorsed earlier in 2026 and which they cited as a call to preserve human dignity and ensure technology serves human encounter and care.
Human dignity and moral limits of artificial intelligence in pastoral care
AI can assist pastoral care, but it cannot itself care for a person in the full human and Christian sense. The governing distinction is between using AI to support human ministry and using it to replace the human relationship at the heart of ministry. Catholic reflection grounds this limit in human dignity: people are not valuable because of what they can do, and a machine’s ability to imitate certain tasks does not make it a human person or a moral agent.
The dignity of each person comes from being created in the image of God, not from intelligence, productivity, independence, or success. The Dicasteries’ Antiqua et Nova stresses that this dignity remains intact even when a person cannot exercise their abilities.
“a person’s worth does not depend on possessing specific skills, cognitive and technological achievements, or individual success, but on the person’s inherent dignity, grounded in being created in the image of God.”
That principle matters in pastoral care because people often seek help when they are vulnerable: grieving, ill, lonely, uncertain, or unable to express themselves clearly. Their worth must never be measured by how easily a system can classify, process, or respond to their needs. Nor should the Church treat people as cases to be handled more efficiently at the expense of personal attention.
CELAM’s pastoral reflection similarly insists that human intelligence is embodied and relational, situated in a person’s history and culture; human dignity does not depend on computational ability or the quantity of information someone holds. In pastoral practice, this means that attentive presence, listening, and respect for a person’s story are not optional extras around the “real” work of delivering information. They are part of how a person is treated as a person.
AI can perform particular tasks quickly and may appear conversational, but the Catholic sources caution against equating this performance with human understanding. Antiqua et Nova describes AI as a product of human intelligence, not an artificial form of human intelligence; its use of the word “intelligence” can mislead if it suggests that a machine understands reality as a person does. AI’s computation is only part of what human intelligence involves: a person’s understanding is formed through embodied life, relationships, moral and spiritual development, and engagement with truth and goodness.
This distinction is especially important when someone shares suffering or asks a deeply personal religious question. A system may produce words that sound compassionate, but sounding compassionate is not the same as sharing another person’s life, bearing responsibility for a relationship, or exercising moral discernment. The sources explicitly state that AI cannot currently replicate authentic relationships or moral discernment.
The point is not that technology is necessarily harmful or useless. CELAM says AI can be a powerful means for good, provided that its use is discerned and directed toward serving the neighbour and upholding dignity, justice, truth, freedom, mercy, and the common good. The moral question is therefore not simply whether AI is used, but what it is used for, whose good it serves, and whether it strengthens or displaces human care.
A warning in Antiqua et Nova concerns healthcare: if AI replaces the relationship between patients and care providers, it risks making care impersonal and worsening loneliness rather than fostering solidarity with those who suffer. That discussion is specifically about healthcare, but its underlying concern is relevant by analogy to pastoral care. If a person in distress is directed only to a chatbot when what they need is a human listener, the technology may obstruct the very encounter that pastoral ministry is meant to provide.
AI might, for example, help a pastoral worker prepare a draft, translate general material, or organize information. Those are possible ways of assisting human ministry, not claims that the cited documents prescribe particular applications. The important test is whether the use leaves human beings responsible for understanding the situation, responding to the person, and offering appropriate care. A tool that saves time but removes meaningful human contact may fail the pastoral purpose it was supposed to serve.
This is particularly significant where a person’s situation is complex or painful. A generated answer may not adequately grasp the person’s history, relationships, culture, or spiritual condition. The sources’ account of human intelligence as embodied and relational gives reason not to treat an automated response as a substitute for attentive human judgment.
AI is not a moral agent that can accept responsibility for the consequences of its decisions. Antiqua et Nova states that people design AI systems and decide their purposes, and that it is human beings—not machines—who exercise moral responsibility, listen to conscience, and discern what is good in a particular situation. The same principle is stated by the Commission of the Bishops’ Conferences of the European Union: even when a machine interacts with or assists people, ultimate responsibility remains with human beings.
For pastoral settings, this means that a priest, pastoral worker, or institution cannot shift responsibility onto a system by saying that “the AI recommended it.” Human users and decision-makers remain accountable for choosing whether to use AI, checking its output, and deciding whether a response is suitable for the person before them. The sources do not provide a detailed pastoral protocol for AI use, but their ethical principle is clear: human judgment must govern the tool, not be surrendered to it.
A prudent pastoral approach would therefore ask: Is the tool helping a person receive better human care, or replacing that care? Is a responsible person reviewing and owning the response? Could the interaction mislead someone into treating a machine as a human confidant or spiritual authority? These are applications of the sources’ teaching on dignity, relationality, and moral responsibility—not a separate technical rule stated in the excerpts.
The Catholic approach is neither to treat AI as inherently evil nor to assume that technical efficiency is automatically progress. Technologies should be evaluated by whether they respect human dignity and fraternity, and whether their development and use serve the human good. In pastoral care, that calls for a clear ordering: technology may serve ministry, but it must not redefine ministry as automated interaction.
The excerpts supplied here do not set out specific rules for sacramental practice, confidential records, or particular forms of spiritual direction. Those matters require more specific guidance. But they do establish the central moral limit: AI can assist with certain tasks; it cannot replace the human person who listens, discerns, accompanies, and takes responsibility for care.