Magnifica Humanitas and the EU AI Act

The EU AI Act in brief, for readers who need it

The EU AI Act is the European Union's comprehensive regulation on artificial intelligence. It was formally adopted in 2024 and entered into force on August 1, 2024. Its obligations apply in phases.

The phasing matters because it explains why August 2026 is a pivot point. The regulation's first phase, beginning February 2, 2025, brought into force prohibitions on certain AI practices (manipulative AI, social scoring, untargeted scraping for facial recognition databases, real-time biometric identification in public spaces with limited law enforcement exceptions) and obligations on AI literacy for providers and deployers. Violations of the prohibitions already carry penalties of up to €35 million or 7% of global annual turnover.

The second phase, beginning August 2, 2025, brought into force obligations on general-purpose AI model providers (the operators of foundation models including GPT, Claude, Gemini, and others). These obligations include transparency on training data, copyright compliance policies, and systemic risk assessment for the largest models. The European AI Office became operational. Member states designated their national competent authorities.

The third phase, beginning August 2, 2026, is the main enforcement phase. The regulation's high-risk AI system obligations under Annex III enter into application: AI used in critical infrastructure, education, employment decisions, essential services, law enforcement, migration, justice administration, and democratic processes. The transparency rules of Article 50 apply, including machine-readable marking of AI-generated content. The European Commission's supervision and enforcement powers against general-purpose AI providers begin. This is the date most companies operating in the EU have been planning their compliance against.

The fourth and final phase, beginning August 2, 2027, brings into application the obligations on providers of general-purpose AI models released before August 2025 and full applicability of the regulation for AI systems embedded in regulated products.

Several member states have also passed national AI legislation that implements or supplements the EU framework. Italy's Law No. 132/2025, in force since October 10, 2025, is one example. The regulatory landscape European AI actors navigate in 2026 is therefore both the EU-level AI Act and an emerging set of national laws.

This page assumes basic familiarity with the regulation beyond this summary. For readers who want the comprehensive treatment, the European Commission's AI Act Service Desk provides the authoritative implementation timeline and explanatory materials at ai-act-service-desk.ec.europa.eu.

Where the encyclical and the regulation align

The alignment is broader than press coverage will typically convey. Six major areas of substantive alignment matter for understanding how the two documents work together.

Prohibition of the most dangerous uses. Both documents identify certain AI applications as categorically impermissible. The EU AI Act prohibits, among other things, AI that uses subliminal techniques or purposeful manipulation to materially distort behavior, AI that exploits the vulnerabilities of specific groups, social scoring systems that lead to detrimental treatment, untargeted scraping of facial images for biometric databases, and emotion recognition in workplaces and educational institutions. Magnifica Humanitas names manipulation of public discourse (¶132-138), exploitation of vulnerable users in the attention economy (¶170), and various forms of relational simulation as moral lines that should not be crossed. The lists are not identical but they overlap substantially.

Transparency requirements. Article 50 of the EU AI Act requires that AI systems interacting with humans be identifiable as AI, that AI-generated content be machine-readably marked, and that deep fakes be clearly disclosed. Magnifica Humanitas develops the same direction in paragraphs 132-138 on truth as a common good, and the 2026 World Communications Day message (Pope Leo XIV's first such message, released a week before the encyclical) makes the deepfake concern more explicit. The regulatory move toward mandatory transparency is consistent with what the encyclical asks for morally.

Human oversight and accountability. The EU AI Act's high-risk system requirements include human oversight obligations, accountability chains that prevent the diffusion of responsibility into “the machine,” and the right to lodge complaints. Magnifica Humanitas paragraph 200, on autonomous lethal weapons, articulates the same requirement in its most absolute form: decisions must remain under “effective, self-aware and responsible human control.” Paragraph 199 specifies the criteria: personal responsibility identifiable and verifiable, judgment timeframes respected, civilians identified and protected. The regulatory framework's operational accountability requirements and the encyclical's anthropological framing point in the same direction.

Protection of vulnerable users. The EU AI Act's prohibitions on AI exploiting vulnerabilities of specific groups, on emotion recognition in workplaces and schools, and on the use of AI in ways that disadvantage children align with Magnifica Humanitas paragraph 141's specific concern about children and young people in digital environments. The regulation protects through legal prohibition; the encyclical asks for protection through moral discernment and pastoral care. The two protect from different directions toward the same end.

Robust legal frameworks as positive value. The EU AI Act exists. Magnifica Humanitas paragraph 106 explicitly calls for this kind of thing: “robust legal frameworks, independent oversight, informed users and a political system that does not abdicate its responsibility.” The encyclical does not name the EU AI Act, but the kind of framework the encyclical calls for is precisely what the EU AI Act attempts. The relationship is one of substantial alignment in direction.

International cooperation. The EU AI Act is, among other things, an attempt to set a regulatory standard with extraterritorial reach. Magnifica Humanitas paragraph 200 calls for “an international framework to curb the technological arms race” specifically on autonomous weapons, and the encyclical's broader argument for the civilization of love (¶210-228) calls for “a shift from the 'culture of power' to a genuine 'culture of negotiation.'” The encyclical's preference for multilateral and dialogical approaches to global AI governance aligns with the EU AI Act's posture as a framework intended to influence global standards.

Where the encyclical asks for more than the regulation requires

The alignment is real. It is not complete. Three areas matter for understanding what the encyclical adds.

Lethal autonomous weapons and the categorical prohibition. The EU AI Act's Article 2 explicitly excludes from the regulation's scope AI “systems specifically developed and put into service for military, defence or national security purposes.” This exclusion was politically necessary for the regulation to be adopted; defense remains a competence reserved primarily to member states under EU law. The practical effect is that the EU AI Act does not regulate the development or deployment of lethal autonomous weapons systems. Magnifica Humanitas paragraph 198 takes a categorical position that the regulation does not: “No algorithm can make war morally acceptable.” The encyclical declares the delegation of lethal decisions to AI to be categorically impermissible. The regulation does not. The encyclical's position is therefore broader in subject matter (it applies to military AI) and stronger in modal claim (categorical impermissibility rather than regulated compliance). For the full reading, see Magnifica Humanitas on War.

Labor conditions and supply chain transparency. The EU AI Act addresses AI in employment decisions and protects against AI-based exploitation in workplaces. It does not, however, reach the labor conditions of those who produce the data and components that AI systems depend on. The data labelers in Kenya and the Philippines, the content moderators reviewing traumatic material, the miners extracting rare earths and cobalt under conditions the encyclical names as akin to slavery: none of these workers is reached by the regulation, because the regulation's scope is the deployment of AI systems in the EU, not the global supply chains that make AI systems possible. Magnifica Humanitas paragraphs 173-179 develop precisely this point at length. Paragraph 179 calls for transparency, ethical due diligence, and protection of workers across the supply chain. The encyclical's labor argument extends beyond the regulation's reach. For the full reading, see New Forms of Slavery.

The anthropological argument and the level the law cannot reach. The third divergence is the most important and the most difficult to describe. The EU AI Act is a regulation. It addresses what AI systems may and may not do, what providers and deployers must and must not provide, what penalties attach to violations. It cannot, by the nature of what regulation is, ask anything of the human heart. Magnifica Humanitas paragraph 130 does ask something of the human heart. It invokes Augustine's two cities and two loves and tells readers that “the construction of Babel or the rebuilding of Jerusalem begins within each one of us.” This is not a request for regulatory compliance. It is a request for self-examination and conversion. The regulation cannot make this request because regulations cannot make this kind of request. The encyclical can. The encyclical's anthropological argument is at a level that the regulatory framework cannot operate at, and it is the level the encyclical considers most fundamental. For the full reading, see Two Cities and Two Loves.

What the regulation requires that the encyclical does not

The mapping needs to go in both directions. There are places where the EU AI Act requires specific operational things that the encyclical does not address. This is not a defect in the encyclical. The encyclical is not a compliance framework; it is a moral framework. But for actors who must satisfy both, it is worth being clear about what only the regulation provides.

Operational compliance specifics. The EU AI Act prescribes specific obligations for providers of high-risk AI systems: risk management systems, data governance and management practices, technical documentation, record-keeping, transparency information for deployers, human oversight measures, accuracy/robustness/cybersecurity requirements, quality management systems, conformity assessment, and registration in the EU database. Magnifica Humanitas does not address these as operational matters. The encyclical's call for transparency and accountability is at the level of principle; the regulation translates that principle into specific operational requirements with audit trails and penalties.

Specific risk classifications. The EU AI Act defines high-risk AI systems by category (Annex III), distinguishes them from limited-risk and minimal-risk systems, and prescribes different obligations for each. Magnifica Humanitas does not provide a comparable taxonomy. The encyclical's framework allows readers to identify particular AI applications as more or less concerning, but the regulation's categorical scheme is doing different work and is more operationally precise.

Enforcement mechanisms. The EU AI Act establishes the European AI Office, requires member states to designate competent national authorities, sets specific penalty levels for different violation categories, creates an EU database for high-risk AI systems, and provides specific complaint mechanisms. Magnifica Humanitas calls for “robust legal frameworks, independent oversight, informed users and a political system that does not abdicate its responsibility,” but the encyclical does not specify how these should be structured. The regulation is the institutional implementation; the encyclical is the moral imperative.

The two documents therefore do not substitute for each other. An EU AI actor cannot satisfy the regulation by reading the encyclical, and a Catholic institution cannot satisfy the encyclical's moral framework by complying with the regulation. The relationship is complementary in both directions.

What this means for different actors

The mapping translates differently for different audiences. Four cases worth treating specifically.

For European policymakers. The encyclical provides moral grounding for what the regulation attempts to do, and pushes beyond what the regulation reaches. The categorical position on autonomous weapons creates pressure for member states with significant defense AI programs to consider the encyclical's argument when forming national positions on weapons-related multilateral discussions. The labor argument in paragraphs 173-179 creates pressure for due diligence legislation that reaches AI supply chains in ways the AI Act does not. The encyclical is, for policymakers, a moral reference point that operates alongside the regulatory framework rather than within it.

For Catholic institutions in Europe. Universities, hospitals, dioceses, schools, and other Catholic-affiliated institutions deploying AI in the EU are subject to the regulation's compliance requirements. They are also being asked, by their tradition and now by an encyclical, to engage AI through the framework Magnifica Humanitas articulates. The two requirements operate together. The institutions that handle the convergence best will treat regulatory compliance as the floor and the encyclical's framework as the ongoing formation work. The framework here is not in tension with the regulation; it asks for more than the regulation requires, in particular at the level of the institutions' own internal practices of deliberation and discernment.

For AI companies operating in the EU. Companies face the August 2, 2026 enforcement deadline and need to be ready. The encyclical does not change that. The encyclical does, however, address AI developers directly in paragraph 111 and frames the work as “human participation in the divine act of creation” with corresponding “ethical and spiritual responsibility.” For Catholic AI developers and for companies whose leadership engages Catholic moral frameworks, the encyclical creates a second set of expectations alongside the regulation. The expectations are mostly compatible: a company that takes paragraph 111's framing seriously will probably exceed the regulation's requirements rather than struggle to meet them. For the full audience-scoped read, see Magnifica Humanitas for Developers.

For non-Catholic policy and industry actors. Catholic teaching's institutional reach in Europe is substantial. The encyclical is a document that European policymakers, civil society organizations, and industry leaders will encounter as part of the moral conversation around AI regulation, whether or not they share its theological premises. The mapping above is offered as a contribution to that conversation, on the assumption that an honest understanding of what the encyclical does and does not say is in everyone's interest.

Open questions and unresolved tensions

The mapping cannot be neat. Three areas of unresolved tension are worth flagging honestly.

Defense industry exclusion. The EU AI Act's Article 2 exclusion of military AI was politically necessary but it leaves a substantive gap. European defense industries are developing AI capabilities, including capabilities that approach the autonomous lethal weapons the encyclical declares categorically impermissible. The EU has no regulatory mechanism through which the encyclical's position translates into binding constraint on European defense AI. The encyclical's call in paragraph 200 for an international framework is doing work the EU AI Act cannot do. How European member states with significant defense AI programs (France, Germany, Italy among others) will respond to the encyclical's position is an open question with substantial consequences.

The proposed delays. Industry pressure to delay the August 2, 2026 enforcement phase has been visible through 2026. The European Commission has rejected calls for blanket two-year delays, signaling commitment to the timeline, but specific provisions are being discussed for refinement. The encyclical's position does not change whether enforcement is delayed; the moral framework would apply equally to a delayed regulation. But delays would matter for how the encyclical's call for robust legal frameworks lands in practice. If enforcement weakens, the encyclical's moral pressure becomes proportionally more important. This is not the encyclical's stated intent, but it is one of the practical consequences of how the two documents relate.

The global dimension. The EU AI Act is a regional regulation with extraterritorial effect. Magnifica Humanitas is a universal moral framework with no jurisdictional limit. The encyclical's reach is broader by design; the regulation's reach is constrained by design. How European institutions implement the encyclical's framework when interacting with AI providers based outside the EU, and how non-European Catholic institutions implement it without a comparable regulatory backstop, are questions the encyclical's reception will work out over time. The early reception is too recent to give clear signals.

Where this sits in the wider conversation

The relationship between Magnifica Humanitas and the EU AI Act is a particular case of a more general pattern. Catholic social teaching has long worked alongside, and sometimes ahead of, specific legal frameworks. Rerum Novarum in 1891 addressed industrial labor conditions that legal frameworks were just beginning to engage. Mater et Magistra in 1961 addressed post-war development conditions that international institutions were still constructing. Laudato Si' in 2015 addressed environmental conditions that climate regulation was incomplete on. In each case the magisterial document and the legal frameworks operated together: the legal frameworks doing institutional work the magisterium could not do, the magisterium doing moral work the legal frameworks could not do.

Magnifica Humanitas and the EU AI Act fit this pattern. The pairing matters because both documents matter. AI policy in Europe over the next several years will be shaped by both, and the actors who understand both will be better placed to engage either.

For the encyclical's broader argument and structure, see the section-by-section summary. For how Pope Leo XIV's papacy is building this teaching, see the compendium of major statements. For the relationship between Pope Leo XIV's teaching and Pope Francis's foundation, including Francis's 2024 G7 address that paralleled in some ways the present moment, see Pope Leo XIV and Pope Francis on AI.