The EU AI Act and Catholic Teaching

What the AI Act actually does

The EU AI Act is Regulation (EU) 2024/1689, adopted in 2024 and entered into force on August 1 of that year. It is the world's first comprehensive law on artificial intelligence. Its central move is to sort AI systems into risk tiers and apply different rules to each tier.

Prohibited AI practices sit at the top of the risk pyramid. These are uses of AI the Act bans outright. The list includes AI that uses subliminal techniques to manipulate behavior in harmful ways, AI that exploits the vulnerabilities of specific groups such as children or the elderly, social scoring by public authorities, certain forms of real-time biometric identification in public spaces, predictive policing based purely on profiling, untargeted scraping of facial images from the internet to build recognition databases, and AI systems that infer emotions in workplaces or educational settings. The prohibitions began applying in February 2025.

High-risk AI systems are the central category. These are AI systems used in domains the Act treats as carrying significant risk to safety, fundamental rights, or the rule of law. The list includes AI in critical infrastructure, education and vocational training, employment and worker management, access to essential services, law enforcement, migration and border control, and the administration of justice. Providers of high-risk systems must meet a substantial list of obligations: risk management, data governance, technical documentation, transparency, human oversight, accuracy, robustness, and cybersecurity. These obligations take effect on August 2, 2026.

General-purpose AI models, the underlying large models that power many downstream AI applications, are governed by a separate set of rules. Obligations include technical documentation, transparency about training data, copyright compliance measures, and, for the most capable models classified as carrying "systemic risk," additional risk identification and mitigation requirements. These rules have applied since August 2, 2025.

Limited-risk and minimal-risk AI systems face lighter obligations or none at all. Chatbots must disclose that users are talking to an AI. Synthetic media must be labeled. Beyond that, most AI systems used in ordinary commercial contexts are governed by general consumer protection law rather than specific AI regulation.

The Act creates a new EU AI Office within the European Commission to enforce the rules for general-purpose AI models, alongside national competent authorities in each Member State who handle enforcement for AI systems in their jurisdictions. Penalties for the most serious violations reach up to 7 percent of global annual turnover or 35 million euros, whichever is higher.

That is the skeleton. The flesh of the Act is in the implementing acts, technical standards, and codes of practice that the Commission and the AI Office have been publishing since 2024, and that will continue to evolve.

Where the Catholic framework and the AI Act converge

A careful reader of Antiqua et Nova, the 2025 Vatican doctrinal note on AI, and of the AI Act will notice that the two documents, despite different sources, different methods, and different audiences, share a remarkable amount of substantive ground.

Protection of the vulnerable. The AI Act's prohibition on AI that exploits the vulnerabilities of specific groups, including children, the elderly, and persons with disabilities, maps almost directly onto the Catholic preferential option for the poor and vulnerable. The framing is different. The Catholic framework grounds this in the dignity of the person and the duty of solidarity. The AI Act grounds it in fundamental rights and consumer protection. The conclusion is the same: certain uses of AI against the vulnerable are off the table.

Restrictions on manipulation. The Act's prohibition on subliminal manipulation tracks closely with the Catholic moral tradition's long-standing concern about practices that undermine the rational agency of the person. Antiqua et Nova warned about AI's capacity to manipulate users emotionally and psychologically. The AI Act draws a legal line in roughly the same place the Catholic moral tradition draws a theological one.

Human oversight of consequential decisions. The Act requires meaningful human oversight of high-risk AI systems, including the right to a human review of decisions that significantly affect a person. This is the legal expression of the Catholic insistence that moral acts cannot be delegated to machines, an argument that Antiqua et Nova made explicitly with respect to lethal autonomous weapons and that the encyclical is expected to extend more broadly.

Transparency and truthfulness. The Act's requirements that synthetic content be labeled, that users be told when they are interacting with AI, and that providers disclose substantive information about their systems align with Catholic teaching on the moral seriousness of truth-telling and the right of persons to accurate information about the world they are acting in.

Concern about biometric surveillance. The Act's restrictions on real-time biometric identification in public spaces, while controversial in their final form, reflect a concern the Catholic tradition shares: that pervasive surveillance changes the relationship between persons and the state in ways that compromise human dignity and freedom.

None of this is to claim that the AI Act is a Catholic document, because it is not. It is to observe that the moral conclusions a careful natural-law analysis would reach overlap substantially with the conclusions a careful fundamental-rights analysis has reached. This convergence is itself a useful thing for Catholic readers to see.

Where they diverge, and what the Church brings that the law does not

The convergences are real. So are the gaps, and the gaps are the place where the Catholic contribution is most distinctive.

AI in human relationships. The AI Act does not address AI companions, romantic chatbots, AI-mediated grief processing, or the use of AI to simulate friendship and intimacy. These uses sit largely outside the law's risk framework, partly because they are not obviously commercial in the way the Act expects regulated AI to be, and partly because the harms they cause are diffuse and long-term rather than acute. Antiqua et Nova called the use of AI to simulate human relationships a grave ethical violation, and Magnifica Humanitas is expected to develop this teaching in depth. The Catholic framework treats the relational use of AI as morally serious in a way the AI Act does not.

Lethal autonomous weapons. The AI Act explicitly excludes AI systems developed or used exclusively for military or defense purposes from its scope. This is a deliberate political choice that reflects the EU's limited competence in defense matters. The Catholic position, articulated by Pope Francis at the G7 in 2024 and reiterated in Antiqua et Nova, is that lethal autonomous weapons should be prohibited outright. The encyclical is expected to make this prohibition a central magisterial demand. The AI Act will not.

The dignity foundation. The AI Act is built on a foundation of fundamental rights as enumerated in the EU Charter of Fundamental Rights. This is a legal foundation, not a metaphysical one. It works well enough most of the time. Where it fails is where the question becomes: why do these rights apply to this entity at all? The Catholic framework provides an answer to that question, grounded in the imago Dei and in a theological anthropology of the person. The AI Act assumes the dignity of persons; the Catholic framework explains it. Both are needed, and the explanation matters most precisely when the cases get hard, which they will.

Formation and the long term. The AI Act is, by design, a regulatory instrument. It addresses what providers and deployers must do, what users have the right to expect, and what penalties apply when the rules are broken. It does not address what AI does to human beings over the long course of their formation, to the institutions that form persons across the life span, or to the cultural conditions in which moral character is built. These questions are at the heart of what an encyclical is for. The AI Act and Magnifica Humanitas are not in competition. They are working on different timescales and different problems.

What the encyclical changes about the AI Act conversation

Magnifica Humanitas does not amend the AI Act. The encyclical has no legal force in the European legal order, and the Catholic Church does not legislate for Europe. What the encyclical does change is the moral vocabulary in which the AI Act is interpreted, implemented, contested, and amended over the next decade.

The AI Act is not a closed text. The Commission is publishing implementing acts and guidance documents continuously. National authorities are working out enforcement priorities. Courts will, in time, decide cases that establish how the Act actually constrains AI in practice. Industry is lobbying for narrower interpretations of particular provisions. Civil society is lobbying for broader ones. Member States are exercising the discretion the Act gives them. In all of this, the moral vocabulary used to make the case for one interpretation or another matters. The vocabulary of fundamental rights, which the Act itself uses, is one register. The vocabulary of human dignity, grounded in something deeper than positive law, is another. The encyclical is about to supply that second vocabulary with the magisterial weight of a papal teaching document, in Europe, at the exact moment the AI Act becomes live.

Three changes seem likely.

First, the encyclical will give Catholic policymakers, academics, journalists, and civil-society actors a framework they can cite without apology in public debates about AI. The same way Laudato Si changed how Catholic voices participated in climate debates after 2015, Magnifica Humanitas is positioned to change how Catholic voices participate in AI debates after 2026.

Second, the encyclical will sharpen the gaps in the AI Act in a way that creates pressure for future amendments. The Act will be reviewed, and proposed amendments will be debated. The encyclical's treatment of AI in human relationships, of AI companions, of lethal autonomous weapons, and of the formation of children will sit on the table during those debates whether the Commission likes it or not.

Third, the encyclical will give Catholic institutions, Catholic schools and universities, Catholic hospitals, Catholic dioceses, Catholic charities, a basis on which to make their own decisions about AI procurement and use that goes beyond legal compliance. The AI Act tells these institutions what they must not do. The encyclical will help them work out what they should do, which is a different and longer question.

What Catholics should actually do about this

This page closes with a practical question, because the convergence of the encyclical and the AI Act is the kind of moment where Catholic readers reasonably want to know what is being asked of them.

For most Catholic readers, the AI Act will affect daily life indirectly, through the AI products they encounter as users and consumers. The most important practical thing to do is to know that the Act exists, to know that certain practices are prohibited, and to use the rights the Act gives, such as the right to know when an AI is being used in a decision that affects you and the right to a human review of certain automated decisions.

For Catholics who work in technology, particularly those employed by providers and deployers of AI in Europe or for European customers, the AI Act is a serious matter that requires actual legal compliance. The encyclical's framework should inform how that compliance is approached, not as a minimum to be cleared but as a floor on a larger moral obligation. The questions to ask are whether the system you are building or deploying makes the human persons who interact with it more themselves or less themselves, whether it serves the vulnerable or extracts from them, whether it builds up the bonds between persons or replaces those bonds with simulations. The AI Act will not ask these questions of you. The encyclical does.

For Catholics in policy, civil society, or journalism, the encyclical is an invitation to participate in a conversation that is going to happen with or without Catholic voices. Better with than without. The next ten years of AI governance will be shaped by who shows up and who has something useful to say. The Catholic tradition has a great deal to say. The challenge is to say it in language that is honest about its foundations and accessible to people who do not share them.

For Catholic institutions, the moment to develop an AI policy grounded in Catholic teaching is now, before the August 2026 enforcement deadline and before AI tools are deployed in ways that will be hard to unwind later. The encyclical will help. Institutional discernment, of the kind that Catholic universities and hospitals have done with other technologies, is the work ahead.

None of this requires being against AI, and Catholic teaching is not against AI. It requires being for the human person, and being clear about what that means in a moment when the technology is moving faster than the moral conversation about it. The encyclical is the Church's contribution to slowing that gap down. The AI Act is Europe's. They are not the same project, and they are not at odds.