Sebastien Rousseau
Get in touch ›

Magnifica Humanitas: An AI-Industry Reading of the First Papal Encyclical on Artificial Intelligence

On 25 May 2026, Pope Leo XIV released the first papal encyclical devoted to artificial intelligence. Read from inside the technology industry — not as doctrine but as an ethical framework — it echoes every moral anxiety that accompanied the train, the automobile, nuclear power, and the internet. And it lands, in the end, on hope.

20 min read

Magnifica Humanitas: An AI-Industry Reading of the First Papal Encyclical on Artificial Intelligence

On 25 May 2026, Pope Leo XIV released Magnifica Humanitas, the first papal encyclical devoted to artificial intelligence (Vatican News). Read from inside the technology industry — not as doctrine, but as an unusually coherent ethical framework for the AI and quantum era — it echoes the moral anxieties that accompanied the train, the automobile, nuclear power, and the internet, and lands on hope rather than fear.

Executive Summary / Key Takeaways

  • On 25 May 2026, Pope Leo XIV released Magnifica Humanitas ("Magnificent Humanity"), a 42,300-word encyclical subtitled On Safeguarding the Human Person in the Time of Artificial Intelligence — the first papal encyclical dedicated to AI (Vatican News), signed on 15 May, the 135th anniversary of Leo XIII's Rerum Novarum (1891), and presented at the Vatican alongside an Anthropic co-founder (NCR).
  • Read objectively from the technology field, its central claim is one most serious engineers already hold: technology is never neutral — it "takes on the characteristics of those who devise, finance, regulate and use it." This is not a religious assertion. It is a design observation.
  • The document deliberately parallels Rerum Novarum, which addressed the Industrial Revolution. That parallel is the key to reading it correctly: every transformative technology — the train, the automobile, nuclear power, the internet — arrived alongside moral anxiety, and in each case the anxiety was neither pure hysteria nor pure obstruction but a signal that governance had to catch up with capability.
  • Its sharpest contribution is a reframing of the core question. The choice, it argues, "is not between a 'yes' or 'no' to technology" but between building Babel (uniformity, profit-idolatry, the person reduced to "data and performance") and rebuilding Jerusalem (communion, shared responsibility, plural voices building together). For the industry, that maps cleanly onto the difference between extractive AI and humane AI.
  • Its warning about transhumanism and posthumanism — the temptation to treat human limits (illness, ageing, vulnerability) as defects to optimise away — connects directly to the themes explored in the May 2026 piece on Lucy on this site: the seductive fantasy of knowledge migrating from flesh to machine.
  • The honest conclusion is hopeful, not alarmist. The encyclical's deepest argument is that the future of AI is not weather, happening to people, but architecture, made by people — being designed in every architecture review, training run, and decision about what to optimise and what to refuse. That is not a warning. It is an invitation.

A Note on How to Read This #

This is not a theological reading. The lens is engineering — payments infrastructure, post-quantum cryptography, agentic systems, the work Magnifica Humanitas is concerned about. The encyclical is addressed, in its own words, "to all the Catholic faithful, to all Christians and to men and women of goodwill" (Vatican.va), which is the warrant for the secular reading that follows.

Read as ethical reasoning rather than doctrine, the document is coherent — more so than most of what the industry produces about itself, more honest about the concentration of power than most regulatory white papers, and more hopeful than the doom-laden discourse that has dominated AI commentary since 2023.

What the Encyclical Actually Says #

The framing device is two biblical images, and they are worth understanding even for a wholly secular reader because they do real analytical work. The first is the Tower of Babel: a single language, a single technology, a single direction, built — in the encyclical's reading — on "pride and the claim to self-sufficiency," a project that "sacrifices human dignity for efficiency." The second is the rebuilding of the walls of Jerusalem under Nehemiah: a project that "rebuilds relationships before rebuilding with stones," where the work is distributed across the whole community and diversity becomes a resource rather than a threat.

The encyclical's pivotal move is to argue that the real choice facing the AI era "is not between a 'yes' or 'no' to technology, but rather between constructing Babel or rebuilding Jerusalem." This is a more sophisticated framing than the accelerationist-versus-doomer binary the industry has been stuck in. It refuses both the techno-utopian position (that more capability is automatically good) and the reflexive-rejection position (that the technology is inherently corrupting). Instead it locates the moral weight where it actually sits: in how the thing is built, financed, governed, and used.

From there, the document is structurally disciplined. It establishes that technology has been "a profoundly human reality, linked to the autonomy and freedom of man" since the beginning, and that it has, over centuries, significantly improved human living conditions — this is not a reactionary text. It then makes its central observation, the one that should arrest any honest engineer: in practice, technology is never neutral, because it takes on the characteristics of those who devise, finance, regulate and use it. It warns about a specific structural shift in power — that the main drivers of development are now private, often transnational, parties whose resources and capacity to intervene surpass those of many governments — and it warns about a specific anthropological danger: the reduction of the person to "data and performance," the treatment of human limits as defects to be engineered away. Its later chapters address truth as a common good, the dignity of work in a time of automation, and — with notable force — the use of AI in warfare and autonomous weapons (TIME, CNN).

That is the architecture. Reading it from the engineering side comes next.

Technology Is Never Neutral: A Design Observation, Not a Sermon #

The single most important sentence in the encyclical, for an engineer, is the claim that technology "takes on the characteristics of those who devise, finance, regulate and use it." Strip away the context and this is simply true, and it is true in a way that the field has been slowly, expensively relearning for a decade.

A recommendation algorithm optimised for engagement takes on the characteristics of the metric it was given; it does not "decide" to radicalise anyone, but it will find that outrage retains attention and it will serve more of it. A credit model trained on historically biased lending data takes on the characteristics of that history; it does not "intend" to discriminate, but it will. A facial-recognition system takes on the characteristics of the dataset it was trained on and the institution that deploys it — benign in a photo app, something else entirely in a surveillance context. The entire discipline of AI safety, AI alignment, and responsible-AI engineering is, in a sense, an elaborate restatement of the encyclical's sentence: the values of the makers and the deployers are inscribed into the artefact, whether or not anyone intended them to be.

This is why the framing matters more than the conclusions. The industry's most common rhetorical defence — "the technology is just a tool, it's neutral, it depends how you use it" — is precisely the position the encyclical, and a decade of hard engineering experience, refutes. Tools are not neutral. They carry the fingerprints of their makers in their defaults, their training data, their optimisation targets, their access models, and the business models that fund them. Recognising this is not anti-technology. It is the precondition for building technology well.

The Pattern: Train, Automobile, Nuclear, Internet #

What gives the encyclical its real interpretive power is its explicit anchoring in Rerum Novarum, Leo XIII's 1891 encyclical on the Industrial Revolution. By signing Magnifica Humanitas on the 135th anniversary of that document, Leo XIV is making a deliberate historical claim: that AI is "another industrial revolution," and that the right response is neither to halt it nor to worship it, but to build the social and ethical scaffolding that lets its benefits reach everyone and its harms fall on no one.

This is the correct frame, and it is worth extending, because the history is genuinely reassuring once you look at it clearly. Every transformative technology has arrived trailing a cloud of moral anxiety, and in each case the anxiety carried real signal.

When the railway spread across Britain in the 1830s and 1840s, respectable medical opinion held that the human body could not survive speeds of 30 miles per hour — that passengers would suffocate, or that their organs would be damaged. There were sermons about the impiety of such speed. The anxiety was, in its literal form, nonsense. But underneath it was a real and legitimate concern about disruption: the railways did upend local economies, did concentrate capital in new ways, did require entirely new bodies of safety law written in the blood of early accidents. The hysteria was wrong; the underlying instinct that this changes everything and we are not ready was right.

When the automobile arrived, the pattern repeated. The UK's Locomotive Acts — the "Red Flag Acts" — required a person to walk ahead of every motor vehicle carrying a red flag. Mocked now as absurd obstruction, they were a clumsy first attempt to answer a genuine question: what happens when private individuals can move heavy machinery through shared public space at lethal speed? It took decades to build the answer — licensing, traffic law, road design, seat belts, crash standards. The cars did kill people. The governance did, eventually, catch up. We did not ban the automobile, and we did not let it run unrestrained; we civilised it.

Nuclear technology is the hardest case and the most instructive. Here the anxiety was not hysterical at all — the technology really could end civilisation, and twice in 1945 it demonstrated its capacity to erase cities. But even here the story is not one of pure catastrophe. The same physics gave us a carbon-free energy source that, per unit of energy delivered, has killed fewer people than almost any alternative; the non-proliferation regime, for all its fragility, has held for eighty years against confident predictions that it would not. Nuclear is the case the encyclical's chapter on weapons and AI is most clearly thinking about, and it is the right cautionary tale: a technology whose destructive and creative potentials are both maximal, and which we have survived only by building — imperfectly, contestedly, but really — the international scaffolding to govern it.

And the internet — the one most of us lived through. The early anxieties (that it would rot children's minds, destroy real human connection, drown truth in noise) were dismissed as moral panic by the technologists, myself among them. With two decades of hindsight, some of those anxieties look less like panic and more like under-reaction: the attention economy, algorithmic polarisation, the collapse of a shared factual baseline, the documented mental-health effects on adolescents. The internet delivered staggering goods — access to the totality of human knowledge, the collapse of distance, the democratisation of voice. It also did real damage that we waved away precisely because the techno-optimist frame told us that progress was self-justifying. The lesson is not "the pessimists were right." It is that the anxiety was data, and we should have read it instead of dismissing it.

This is the pattern Magnifica Humanitas is pointing at, and it is why the document should not be read as Luddism. In every case, the moral anxiety that accompanied the technology was neither pure hysteria nor pure wisdom. It was a signal that capability had outrun governance, and that the gap had to be closed by deliberate human effort. The encyclical is, in effect, the early moral anxiety of the AI era — and the historical record says that the right response to such anxiety is not to dismiss it as religious technophobia, but to read it for the signal it carries and to close the governance gap faster than we did last time.

Where This Touches Financial Engineering #

Two of the encyclical's specific concerns land directly on the engineering of financial systems, and both deserve an engineer's response rather than a believer's.

The first is the concentration of power in private, transnational hands. The encyclical's observation that the principal drivers of this technology are now private actors whose capacity exceeds that of many states is not a theological claim; it is an accurate description of the 2026 AI landscape, and it is precisely the concern that runs through the regulatory architecture the industry is currently scrambling to meet — the EU AI Act's high-risk obligations enforceable from 2 August 2026, DORA's third-party concentration-risk provisions, and the sovereign-cloud and sovereign-control-plane movements covered in the cloud architecture and agentic engineering pieces on this site. The encyclical and the EU AI Act are, remarkably, worried about the same thing: that capability without accountability concentrates power in ways that escape democratic governance. One says it in the language of social doctrine and the other in the language of conformity assessments. The diagnosis is identical.

The second is quantum computing, which the encyclical does not name but whose ethical shape it precisely anticipates. The premise of the post-quantum migration covered in Securing the Ledger is that a capability now under construction — a cryptographically relevant quantum computer — will, when it arrives, retroactively expose everything encrypted under today's assumptions. "Harvest now, decrypt later" is, in a secular register, exactly the kind of power-without-wisdom the encyclical warns against: a technology whose consequences arrive on a timeline that outruns preparation, in the hands of those with the resources to wield it first. The encyclical's insistence on asking "where are we going?" before the succession of emergencies dictates the path is, for a security engineer, good practice rendered as moral philosophy.

The Transhumanist Temptation, and the Lesson of Lucy #

The encyclical reserves its most searching critique for transhumanism and posthumanism — the narratives that treat human limits (illness, ageing, suffering, vulnerability) not as constitutive of the human condition but as engineering defects awaiting an upgrade. Its claim, in paragraph after paragraph, is that human beings often flourish through their limitations, and that an AI which tempts us to "escape limitation through optimisation" rather than supporting "openness and communion" has misunderstood what a person is for.

This is the seduction examined in the May 2026 piece on Luc Besson's Lucy on this site. The film's fantasy — that consciousness could be progressively unlocked until it migrated, complete, from flesh onto a flash drive — is the purest possible expression of the posthumanist dream the encyclical is worried about: knowledge without a knower, intelligence without embodiment, the human reduced to extractable information. That piece argued the fantasy is seductive precisely because it flatters discomfort with limits, and that the more interesting truth is the opposite one — that the limits are not the bug, they are a large part of where the meaning lives. The encyclical reaches the same conclusion by a different road. A film critic's reading of a science-fiction thriller and a papal encyclical arriving at the same place is itself worth noticing: it suggests the insight is not parochial to either frame.

This is the point at which the secular reader and the encyclical can shake hands without anyone having to convert. Believing that a human being is more than a dataset does not require believing in a soul. It only requires noticing that the most meaningful things in a life — love, grief, the slow earning of wisdom, the closeness that vulnerability makes possible — are not optimisation problems, and that an industry which treats them as optimisation problems will build tools that make people efficient and lonely. The encyclical names this with unusual precision.

A Message of Hope: Architecture, Not Weather #

The dominant register of AI commentary since 2023 has been fear. The encyclical, for all its warnings, is finally not a fearful document — its closing movement is called "the song of hope" (Ascension Press), and the hope is worth reading with engineering precision.

The choice between "building Babel and rebuilding Jerusalem" is not a prophecy about which future the technology will deliver. It is a statement that the future is being built, and that the people building it have agency over its shape. This is something every engineer knows in their hands even if they have never put it in these words: the system does what it is designed to do. The recommendation engine optimises for what it is told to optimise for. The model embodies the values trained into it. The agent acts within the boundaries it is given. AI is not weather, happening to people. It is architecture, made by people — and in the encyclical's image, the families of Nehemiah, each assigned a section of the wall.

That reframing dissolves the false fatalism in much of the current discourse. The question "will AI be good or bad for humanity?" is badly formed, because it treats AI as an autonomous force with its own trajectory. The honest question is "what is being built, by whom, under what constraints?" — and that is a question answered every day, in design reviews and architecture decisions and the quiet choices about what to optimise and what to refuse.

The historical record is, on balance, reassuring. The train was civilised. The automobile was civilised. Nuclear power, against the odds, has been held — the non-proliferation regime, for all its fragility, has now lasted eighty years against confident predictions that it would not. The internet's harms are being slowly, belatedly addressed. In every case the good was preserved and the harm was reduced not because the technology arrived pre-civilised, and not because anxious people stopped it, but because builders and citizens took responsibility for the shape of the thing. AI and quantum computing are the present generation's section of the wall.

What This Means by Reader Position #

The encyclical's implications differ depending on the role.

Technology leaders and founders. The "technology is never neutral" claim is a governance mandate, not a philosophical aside. The values, incentives, and business model behind a system are part of its specification, and pretending otherwise is no longer credible to regulators, the public, or the engineers inside the firm. Institutions that internalise this will treat ethics as an architectural concern, designed in from the start, rather than a public-relations layer applied afterwards.

Engineers and researchers. The encyclical is, unexpectedly, on the side of the field's most important internal argument: that how a thing is built matters as much as whether it works. Pushing for the more accountable, more auditable, more humane design against the pressure to ship the extractive one is the work the encyclical describes as rebuilding Jerusalem. The case for that work now has Catholic social doctrine alongside the EU AI Act, DORA, and ten years of post-incident learning behind it.

Policymakers and regulators. The encyclical and the EU AI Act are describing the same risk from different vocabularies. The convergence is an opportunity: the moral framing can build the public legitimacy that technical regulation, on its own, struggles to achieve. "Concentration of unaccountable power" is an abstraction; "Babel" is a story, and stories move people to act.

The wider public — the encyclical's "men and women of goodwill". The encyclical's sharpest line for the non-specialist reader is that most people are "watching and waiting, observing from afar and merely hoping for the best." That posture is not neutral: where infrastructure is concerned, abstention is itself a choice about who gets to design the defaults.

Conclusion #

Magnifica Humanitas will be read by most of the technology industry, if it is read at all, as a religious document with limited bearing on the engineering. That would be a mistake. Read as ethical reasoning rather than doctrine, it is among the clearest statements yet made of the thing the industry most needs to hear and most resists hearing: that the artefacts being built carry the values of their makers whether the makers admit it or not, that capability has once again outrun governance, and that the gap will be closed only by deliberate human choice. It says this in the company of Rerum Novarum and the long history of technologies — the train, the automobile, nuclear power, the internet — that arrived trailing moral anxiety and were, in the end, neither stopped nor worshipped but civilised by people who took responsibility for them.

It lands on hope. The future of AI and quantum computing is not a forecast; it is, in the encyclical's phrase, "the construction site of our time." On the evidence of history, the gap between capability and governance is again closable — not perfectly, not without cost, but closable, by the same mechanism that closed it for the train, the automobile, nuclear, and the internet: builders and citizens taking responsibility for the shape of the thing. That is the right note to start from.

Questions? Answers.

What is Magnifica Humanitas, in plain terms?

It is the first encyclical (the most authoritative form of papal teaching document) devoted entirely to artificial intelligence, released by Pope Leo XIV on 25 May 2026. At roughly 42,300 words, subtitled On Safeguarding the Human Person in the Time of Artificial Intelligence, it situates AI within the Catholic Church's "social doctrine" — the same body of teaching that began with Leo XIII's 1891 encyclical on the Industrial Revolution, Rerum Novarum, on whose 135th anniversary this document was signed. It is addressed not only to Catholics but explicitly to "all men and women of goodwill," which is the warrant for reading it, as this article does, as a secular ethical argument and not only a religious one.

Why should anyone in the technology industry care what a papal encyclical says about AI?

Because it is one of the most coherent and institutionally durable ethical frameworks anyone has produced for this technology, and because its central claims are correct on their own terms. "Technology is never neutral" is a design observation that the entire discipline of AI safety implicitly endorses. The warning about power concentrating in private transnational actors is an accurate description of the 2026 landscape that the EU AI Act and DORA are independently trying to address. You do not have to share the theology to find the analysis useful — and the encyclical's authors explicitly invite exactly that kind of reading.

Isn't comparing AI to the train, automobile, nuclear, and internet just a way of minimising the risk?

The opposite, actually. The historical pattern shows that the moral anxiety accompanying each transformative technology carried real signal — it marked the gap between what the technology could do and what our governance was ready for. In some cases (the train) the literal fears were nonsense but the instinct that disruption was coming was right. In others (nuclear, and arguably the internet) the anxiety was under-stated, and we paid for dismissing it. The lesson is not that AI is harmless because we survived the others. It is that we survived the others by taking the anxiety seriously and closing the governance gap deliberately — and that we should do so faster this time.

How does this connect to your earlier article on Lucy?

The encyclical's critique of transhumanism — the treatment of human limits as defects to be optimised away — is the same critique the Lucy piece made about that film's fantasy of consciousness migrating from flesh to machine. Both arrive at the conclusion that a human being is more than extractable information, and that the limits we are tempted to engineer away are a large part of where meaning actually lives. A film critic's reading and a papal encyclical reaching the same conclusion by different roads is a sign the insight is robust rather than parochial.

Where do the encyclical and the EU AI Act actually agree?

On the concentration of power. The encyclical's observation that "the principal drivers of this technology are now private, often transnational, parties whose resources and capacity to intervene surpass those of many governments" is, almost word for word, the structural anxiety underlying the EU AI Act's high-risk obligations, DORA's third-party concentration-risk provisions, and the sovereign-cloud movement now reshaping financial-infrastructure procurement. The encyclical reaches its conclusion through social doctrine; the regulation reaches it through conformity assessments. The diagnosis is identical.

So is the overall message optimistic or pessimistic about AI?

Optimistic, but not naïvely so. The argument is that the future of AI and quantum computing is not predetermined — it is being built, right now, by human choices — and that history shows we are capable of preserving the good of a powerful technology while reducing its harm, provided we take responsibility for its shape rather than treating it as an autonomous force. Fear treats AI as weather. Hope treats it as architecture. The honest reading of both the encyclical and the engineering is that it is architecture, and that the people building it can choose to build it well. That is a genuinely hopeful conclusion, and it is earned rather than wished for.

References #

Last reviewed .