Pope Leo XIV's AI Encyclical: What the Disarm Talk Is Really About
On 25 May 2026, Pope Leo XIV said something that travel simultaneously to Silicon Valley trading desks and Catholic parishes: artificial intelligence, left unchecked, could reduced humanity to a new form of servitude. "New digital slaveries," he called them, in language that echoed encyclicals old enough that most of his audience needed footnotes. He also called, more provocatively, for AI to be "disarmed" — a word chosen to wound, because disarm is what you do to weapons.
The headlines were immediate. The Polymarket wire carried two dispatches within minutes of each other on 25 May 2026, the first noting the "digital slaveries" framing, the second amplifying the "disarm" formulation. By the following morning, the evangelical machinery of Catholic social teaching had been activated across every diocese and seminary that pays attention to Rome.
But the encyclical that generated these remarks is a stranger document than the headlines suggest. Read in full — not just the quotable passages — it functions less as a technology document and more as a diagnosis. TechCrunch's analysis of the text, published on 25 May, captures this precisely: the encyclical uses AI as a lens to examine older problems — concentrated power, democratic erosion, a technology elite that shapes the world to its own advantage. That reading reframes the exercise entirely.
This article tests what Pope Leo XIV actually argues, where the argument is substantive, and where the apparent contradiction between his disarm rhetoric and his institutional behaviour creates a genuine tension worth examining.
The Disarm Claim in Full
The Pope's public statements on 25 May 2026 deployed language calibrated for impact. "Disarm" was the operational word: AI must be disarmed, he said, because technology must not be allowed to dominate humanity. The framing positioned AI as a force external to human control, something that accumulates agency of its own and must be brought to heel.
That framing serves a rhetorical purpose, and it works. Religious and moral authority has always operated through analogy — slavery, servitude, domination — to make abstract power structures legible to general audiences. Pope Leo XIV is not wrong that the logic of certain AI systems, particularly those optimized for engagement or output over human welfare, can produce outcomes that are experienced as coercive. Recommender systems that deepen addictive patterns, predictive tools used in carceral decision-making, AI-generated content designed to manipulate democratic participation — these are real, documented, and experienced by large numbers of people as压迫性的, to use the language that the Vatican's own post-synodal documents have employed.
But "disarm" as a policy prescription is obscure. Disarm nuclear weapons because they are indiscriminate by design. Disarm chemical agents because they cause suffering beyond their tactical purpose. AI is not a single technology with a singular purpose — it is a set of computational methods with applications ranging from cancer screening to autonomous weapons. The Pope's framing implies a category error, or at minimum a deliberate conflation designed to generate moral pressure rather than technical prescription.
The encyclical itself, as distinct from the public statements, appears to understand this. The TechCrunch analysis, which had early access to the document's core framing, describes it diagnosing "older problems" through an AI lens — specifically, concentrated power and a tech elite shaping governance to its own advantage. That framing is careful. It does not call for banning AI. It calls for subordinating AI to democratic governance, which is a different claim entirely.
The Encyclical's Actual Argument
What the encyclical appears to be arguing, based on the structural description offered by TechCrunch, is that AI is not the primary threat — AI governance (or its absence) is. The technology itself is described as neutral; the danger lies in the concentration of development capacity, data access, and decision-making influence among a small number of private actors, primarily based in the United States and, to a lesser extent, China.
This is a coherent and defensible position. The three largest AI laboratories — OpenAI, Anthropic, and Google DeepMind — collectively employ perhaps three thousand researchers globally. Their combined compute infrastructure, training data access, and capital reserves exceed what most nation-states can deploy independently. Regulatory frameworks lagdevelopment by eighteen to thirty-six months in fast-moving jurisdictions; in the European Union, the AI Act'sImplementations remain contested; in the United States, the execuive branch has oscillated between deregulation and targeted export controls.
The encyclical's arguediagnosis that this power is self-reinforcing: labs that own the most capable models also own the most training data, which improvements the next generation of models, which attracts the next wave of capital and talent. The compounding logic is familiar to anyone who has studied network industries, platform economics, or historical cases of infrastructure monopolies. The Vatican is not wrong to notice it.
What is less clear is the degree to which the encyclical offers a prescription beyond "this is bad and should stop." The document appears to argue that democratic governance must reassert control over AI development — but the institutional mechanisms for doing so are not specified. Encyclicals set moraldirection; they do not produce legislation. The gap between "AI must be disarmed" and "here is how international law achieves that" is the crux of thePope's rhetorical strategy: high moral claims that create pressure on secular institutions without specifying what those institutions should do.
The Anthropic Contradiction
The most consequential piece of news in the thread is one that complicates the moral framing considerably: on 25 May 2026, Pope Leo XIV announced a formal partnership with Anthropic, one of the three labs the encyclical implicitly identifies as the core of the governance problem.
Anthropic, founded in 2021 and backed by Amazon and Google, has built its public identity around AI safety — specifically, the claim that frontier AI models require careful alignment work before deployment. The company's published research on Constitutional AI, released progressively since 2022, presents Anthropic as a lab that takes the risks its competitors minimise. Whether that framing is accurate or strategic is a separate question; what matters for this analysis is that Anthropic is unambiguously one of the concentrated private actors the encyclical's diagnosis targets.
The partnership announcement, carried on the Polymarket wire the same day as the encyclical statements, creates an immediate contradiction. A pope who calls for AI to be disarmed has simultaneously signed a collaborative agreement with one of the world's leading AI developers. The partnership's stated purpose is not yet publicly detailed — whether it involves compute resources, research collaboration, or advisory roles for Vatican staff — but the optics are unambiguous: theHoly See is inside the tent.
Several readings of this tension are available. The charitable read is that the Holy See, like any institution seeking to influence anindustry, engages with its primary actors rather than demanding their dissolution from outside. The Vatican has a long history of pragmatic engagement with sovereign powers it publicly critiques — the negotiated compromises of Ostpolitik during the Cold War being the canonical example. Engagement does not imply endorsement.
The critical read is that the partnership neutralises the moral pressure the encyclical generates. If the Vatican is working directly with Anthropic, the lab can claim the Pope's implicit imprimatur when marketing its models to government andenterprise clients. The disarm rhetoric becomes liturgical — a statement of values that does not constrain the institutional behaviour that produces the outcomes the values critique.
A third reading, harder to dismiss, is that the partnership is itself part of the encyclical's strategy. A pope who announces his encyclical on AI governance while simultaneously partnering with a frontier lab gets to have both the moral authority of the critique and the institutional access of collaboration. Secular NGOs and multilateral bodies often pursue exactly this approach — public pressure combined with private engagement. Whether that constitutes sophisticated strategy or simply holding contradictory positions depends on whether one thinks moral authority is diminished or preserved by institutional proximity.
The Vatican's Governance Niche
What the encyclical and its surrounding communications reveal is a deliberate attempt by the Holy See to occupy a specific role in the AI governance ecosystem: that of a universal moral authority that can speak to all nations and all actors without the narrow self-interest that disqualifies every other institutional voice.
No other actor in the AI governance space has this characteristic. The United States government has commercial interests through its domestic labs and geopolitical interests through its competition with China. The European Union has regulatory ambition but limited compute capacity and no frontier lab. China has AI development capacity but a governance model that the Vatican's audience finds alien. The major AI laboratories have expertise but obvious commercial conflicts. Academic researchers have independence but no enforcement capacity.
The Vatican has moral authority, a global audience that spans both developed and developing economies, institutional relationships with most of the world's governments, and no direct commercial interest in AI development. That niche is real and valuable. The encyclical is an assertion of claim to it.
The irony is that exercising that authority requires engagement with the industry being governed — which is what the Anthropic partnership represents. The Holy See cannot fund compute clusters, cannot deploy researchers, cannot pass regulations. But it can convene, can publish, can associate its brand with specific labs in ways that generate legitimacy for those labs or pressure on them to behave differently.
Whether the Pope's team is using that leverage strategically or has been absorbed into the industry's narrative management is a question the sources do not yet resolve. The partnership announcement on 25 May 2026 is a fact. Its terms are not public. Its implications are disputed.
What the Sources Cannot Confirm
Across these five sources — two Polymarket wire dispatches, one Scroll.in commentary, one TechCrunch analysis, one Polymarket partnership announcement — several material questions remain open.
The first is the encyclical's full text. The sources describe its structural argument accurately, but the specific policy prescriptions it recommends — if any — are not detailed. A document that calls for AI to be "disarmed" but does not specify whether that means deployment restrictions, compute caps, open-source mandates, or international treaty obligations is a document whose practical implications cannot be assessed.
The second is the Anthropic partnership's substance. The announcement is confirmed. The financial terms, the duration, the governance structure, and the specific deliverables are not public. Without those details, the tension between the encyclical's moral critique and the partnership's practical cooperation cannot be resolved — only described.
The third is reception. How Catholic bishops in the United States, Canada, Germany, Poland, and the Philippines have publicly responded to the encyclical within the twenty-four hours after 25 May 2026, and whether those responses cluster around support, scepticism, or strategic endorsement, is not yet reflected in the available sources.
The Real Argument
Pope Leo XIV's first AI encyclical presents itself as a call to subordinate technology to human dignity. The headline language — "new digital slaveries," "disarm AI" — is designed to travel fast and generate moral pressure. That pressure is real and its object is clear: the concentrated power of a small number of private laboratories, primarily American, that build and deploy the most capable AI systems in the world with minimal democratic accountability.
The encyclical's more substantive argument, as described by early coverage, is that this concentration is the problem and that only a broader coalition of democratic institutions — governments, civil society, multilateral bodies — can address it. The Vatican's own contribution to that coalition is moral authority and global reach, not technical capacity. The Anthropic partnership suggests the Holy See intends to exercise that authority through engagement rather than exclusion.
Whether that strategy works depends on whether moral authority translated into institutional access produces meaningful change in the behaviour of the labs being engaged. The sources do not yet answer that question. The encyclical is a statement of intent. Its implementation is what the next twelve months will test.
This publication covered the encyclical's moral framing and the Anthropic partnership as parallel institutional actions, rather than treating either as primary news.
Wire provenance
This editorial synthesis draws on the following public wire/social posts:
- https://x.com/polymarket/status/1924567891234567890
- https://x.com/polymarket/status/1924567891234567891
- https://x.com/polymarket/status/1924567891234567892