WHAT THE ENCYCLICAL ACTUALLY SAYS
The substance of Magnifica Humanitas is more specific than the headlines suggest. The document is not a blanket condemnation of AI, nor is it a naive endorsement of technological progress dressed in theological language. It is a structured argument about power, transparency, and the conditions under which any technology can be considered to serve human dignity rather than undermine it.
On regulation, the encyclical is explicit: AI systems that affect decisions about employment, healthcare, criminal justice, or access to public services must be subject to meaningful human oversight, and that oversight must be governed by law rather than left to the voluntary commitments of the companies building those systems. The Pope calls for ownership and control of AI training data not to be concentrated in private hands — a position that, if taken seriously, would represent a more radical restructuring of the current AI industry than anything proposed by the EU AI Act or the US executive orders that have preceded it. Data is the feedstock of frontier AI; who controls it determines who controls the systems trained on it.
On warfare, the document is unambiguous in a way that few secular governance frameworks have been willing to be. Autonomous weapons systems — those capable of selecting and engaging targets without meaningful human authorisation at the point of decision — are characterised as incompatible with the Catholic tradition of just war doctrine. This is not a new position for the Church, but embedding it within an encyclical gives it a different weight. Encyclicals are not opinion pieces; they are formal expressions of the Church's teaching, and they carry authority for approximately 1.4 billion Catholics worldwide. Leo XIV is not writing a policy paper. He is issuing a moral directive to a community that spans every country where AI is being developed and deployed, including the engineers and executives building the systems in question.
On workers, the document draws its most direct historical parallel. The Pope argues that AI automation poses a threat to labour dignity that is analogous to but structurally different from the threat posed by industrial machinery. Industrial displacement destroyed specific categories of physical work; AI automation threatens the cognitive work that previously offered displaced workers a path to economic participation. The encyclical calls on governments to invest in transition infrastructure — retraining, social protection, and the design of economic systems that distribute AI-generated productivity gains broadly rather than concentrating them at the frontier of capital ownership. This is a political economy argument, not just a pastoral one, and it puts the Church's weight behind a set of distributional demands that no major AI company has voluntarily embraced.
WHY ANTHROPIC CHOSE THE VATICAN OVER THE WHITE HOUSE
The Washington Post framed yesterday's ceremony as Anthropic "aligning with the Vatican over the White House," and that framing, while blunt, captures something real. Anthropic has been in a sustained and public dispute with the Trump administration over the use of its technology in military and intelligence operations. The company's refusal to accept certain Pentagon contract terms — specifically those that would allow its models to be used in lethal autonomous weapon systems without the human oversight provisions the company considers non-negotiable — has put it in an adversarial posture with the current US government at a moment when most of its competitors are competing for defence contracts.
Standing alongside Chris Olah at the Vatican was not a neutral act. It was a deliberate signal about which legitimating authority Anthropic is choosing to cultivate. Regulatory approval from governments is necessary but not sufficient for an AI company that has positioned safety as its core identity. Moral legitimacy — the sense that the company's values and the values of the broader society it claims to serve are actually aligned — is a separate and arguably more durable asset, particularly in a period when governments are demonstrating that their AI governance positions are fluid and politically contingent.
Anthropic's involvement in the encyclical process reportedly stretched back months. The Vatican's decade-long effort to engage Silicon Valley in dialogue was not primarily a theology project; it was an institutional recognition that the Church's authority on questions of human dignity and social justice would become relevant again in the AI era, and that building relationships with the companies making the decisions now was preferable to condemning them retrospectively. Olah's presence at the ceremony was described by the National Catholic Reporter as a call for a "church-tech ethics partnership" — a phrase that would have been unintelligible five years ago and now describes something that both parties clearly see as strategically valuable.
The calculus is not complicated. Anthropic operates in a political environment where the dominant US government position on AI is that accelerated development is a national security imperative and that restrictions on capability are a form of competitive self-harm. In that environment, associating with an institution that frames AI ethics not as a technical concern but as a matter of fundamental human dignity gives Anthropic a platform for its positions that does not depend on Washington's approval. The 1.4 billion Catholics who now have a pope who has issued a formal teaching on AI represent a constituency that can shift public opinion and legislative priorities in ways that no corporate communications strategy can replicate.
THE CHURCH AS A GOVERNANCE ACTOR
The most significant dimension of yesterday's events is the least frequently analysed one: what it means when the Catholic Church enters the AI governance arena as an actor with institutional weight, not merely as a commentator. Encyclicals have historically shaped policy in ways that are not always visible in the immediate aftermath. Rerum Novarum, the document Leo XIV explicitly invokes, was published in 1891. Its influence on the development of labour law, social insurance, and the political economy of the twentieth century was substantial and long-running. It did not produce immediate legislative change; it produced a durable framework for thinking about the obligations of industrial actors that shaped debate for decades.
The Vatican is not a regulatory body. It cannot fine AI companies, impose compliance requirements, or block model deployments. What it can do is set terms for moral seriousness on questions where the existing regulatory frameworks are still being constructed. The EU AI Act addresses specific risk categories; it does not make a claim about the fundamental relationship between technological development and human dignity. Magnifica Humanitas does make that claim, and the institutions that choose to engage with it — as Anthropic has, as the Vatican's new cross-department AI commission will — will shape the vocabulary of the debate in ways that eventually filter into legislation.
There is also a direct organisational consequence to consider. The Vatican AI commission established in advance of the encyclical's publication will now have a formal mandate to examine AI's effects across the full range of the Church's global operations, from hospitals and schools to development organisations and refugee services. That is not a small footprint. The Catholic Church operates one of the largest non-governmental healthcare and education systems in the world. Its experience deploying, assessing, and critically evaluating AI in high-stakes human-facing contexts will produce institutional knowledge that regulators and companies alike will eventually find it useful to consult. The Church is not joining the AI governance conversation from the outside; it is building infrastructure for sustained participation from within.
WHAT THIS MEANS IF YOU ARE BUILDING AI PRODUCTS
The immediate practical implication for most AI product teams is that the governance landscape is more complex than it was six months ago. The EU AI Act and various national regulatory frameworks were already creating compliance requirements; the encyclical adds a different layer — not a legal one, but a reputational and institutional one that will matter in markets where the Church has significant influence over public opinion, purchasing decisions, and political priorities.
For companies selling AI systems into healthcare, education, or social services — sectors where Catholic institutions are significant buyers and operators — the encyclical's specific positions on human oversight, data ownership, and worker protection will increasingly be relevant to procurement decisions. An AI product that cannot demonstrate meaningful human oversight of consequential decisions, or that relies on opaque training data practices, faces headwinds in those markets that would not have existed before yesterday. This is the way institutional moral authority actually operates in practice: not through direct regulatory intervention but through the accumulation of stated positions that become purchasing criteria, donor requirements, and political pressure points.
The deeper implication is about the terms on which AI's legitimacy will be contested over the next decade. Regulatory approval proves legal compliance. Benchmark performance proves capability. Neither proves that a system is actually aligned with the values of the people it affects. The Church's entry into this space is a signal that the legitimacy question — who decides that AI is serving human flourishing rather than undermining it, and on what basis — is not going to be answered solely by engineers, economists, or governments. The institutions that have spent centuries developing frameworks for thinking about human dignity, distributive justice, and the limits of power are arriving at the table. The companies that engage with that arrival seriously will be better positioned than those that dismiss it as a communications problem.