In this issue — 4 parts
Part 1 of 4
The Accountability Question Nobody Is Answering
We Are Quietly Letting Machines Make Decisions That Belong to People
Most of the conversation about artificial intelligence in business is about capability: what the new model can do, how much faster it works, how many people it can replace. Almost none of it is about a far more important question. When an AI system makes a decision that changes someone's life, who is responsible for it? When a model denies a loan, screens out a job applicant, flags a patient as low priority, prices a customer out of coverage, or recommends a target, where does the human accountability sit? In the rush to deploy, the honest answer at most organizations is: nobody is quite sure.
That evasion is one of the oldest in Scripture. When God asks Cain where his brother is, Cain answers with a question of his own: "Am I my brother's keeper?" (Genesis 4:9). It is the first recorded attempt to dodge accountability for a human life, and it failed. "The algorithm decided" is the same move in modern dress, an attempt to make responsibility disappear into a system. Magnifica Humanitas refuses it, and so should any organization that takes its own conduct seriously.
On May 25, 2026, that uncomfortable question was placed at the center of the world stage by an unlikely source. Pope Leo XIV released the first encyclical of his pontificate, Magnifica Humanitas ("Magnificent Humanity"), subtitled On Safeguarding the Human Person in the Time of Artificial Intelligence. An encyclical is the most authoritative form of papal teaching, a letter addressed not only to the world's roughly 1.3 billion Catholics but to "all people of goodwill." This one runs to more than 35,000 words across five chapters. It is the most prominent statement any global moral authority has yet made about AI, and it is fundamentally a document about governance, accountability, and human dignity, the same concerns that sit at the heart of how every serious enterprise should be deploying this technology.
We are writing about it because it is not a religious curiosity to be filed away. Within seventy-two hours of its release it had been publicly endorsed by the Vice President of the United States, attacked as an "anti-American screed" by influential figures in Silicon Valley, and turned into a front-page fault line in the American debate over how, and whether, to govern AI. When the questions a papal encyclical raises are the same questions your board, your regulators, and your customers are starting to ask, the smart move is to understand the argument and get ahead of it.
And the argument matters to us specifically, because it is the argument AuthorityGate was built to answer. Our entire premise is that AI should be deployed for good, with humans accountable for consequential decisions and a verifiable gate between what a machine proposes and what an organization actually does. Magnifica Humanitas makes that case from the highest pulpit in the world. This issue explains what it actually says, separates the substance from the political noise, and translates it into concrete governance steps any organization can take.
Why a Pope Is Writing About Software
The choice of name is itself the thesis. Cardinal Robert Prevost took the name Leo XIV in deliberate echo of Leo XIII, who in 1891 wrote Rerum Novarum, the landmark encyclical on the rights and dignity of workers during the upheaval of the Industrial Revolution. That document gave rise to a whole tradition of "Catholic social teaching" that has shaped labor law, the idea of a living wage, and corporate responsibility far beyond the Church. Leo XIV signed Magnifica Humanitas on May 15, 2026, the anniversary of Rerum Novarum, drawing an explicit line: the Industrial Revolution reorganized human labor and demanded a moral response, and the AI revolution is doing the same thing now.
The document's deeper roots run all the way back to the opening pages of the Bible. Its title, "Magnificent Humanity," is a claim about what a human being is: a creature made "in the image of God" (Genesis 1:27), with a worth that no system can confer and none can revoke. The Psalmist marvels at the same thing, "what is man that you are mindful of him… yet you have made him a little lower than the angels and crowned him with glory and honor" (Psalm 8:4-5). That conviction, that the human person carries an irreducible dignity, is the load-bearing wall under everything the encyclical says about AI. Strip it out and "optimize the human" starts to sound reasonable; keep it, and the machine is permanently put in its place: a tool, never the measure.
That framing should land for any executive. The last time a general-purpose technology this powerful swept through the economy, the organizations that treated workers as interchangeable inputs eventually faced regulation, unionization, reputational damage, and legal liability. The ones that got ahead of the human question, that built dignity and accountability into how the technology was used, fared far better over the long run. Leo's wager is that AI is the same kind of inflection point, and that the choices organizations make right now about human accountability will define which side of history they end up on.
Crucially, this is not a document that says "no" to technology. Leo is explicit that the choice is not between accepting or rejecting AI. The choice is about how it is built, who controls it, and whether it serves people or subordinates them. That is precisely the posture a mature enterprise needs: not fear, not blind adoption, but governance.
The central image of Magnifica Humanitas: the human person, not the machine, is the measure of every system. Technology that affects people's "rights, opportunities, status and freedom" must remain in service of the person and accountable to a human.
Why This Matters to You
The encyclical's core demand is the one regulators, courts, and customers are converging on independently: that a clearly identifiable human must remain accountable for consequential decisions, and that "it is not permissible to entrust lethal or otherwise irreversible decisions to artificial systems." If your organization cannot today name who is responsible for a given AI-driven decision, or show the chain of responsibility behind it, you are exposed, ethically, reputationally, and increasingly legally.
This is no longer a fringe concern. The United States is roughly 20% Catholic, the document will echo through sermons and op-eds for a year, and it has already split the technology and policy worlds. AI accountability has moved from an engineering footnote to a board-level governance question. The organizations that can demonstrate human oversight will earn trust; the ones that cannot will spend the next several years explaining themselves.
Coming up in Part 2 — exactly what Pope Leo said, distilled into six themes that read less like theology and more like a governance manifesto.
Part 2 of 4
Six Things Pope Leo Actually Said
What you missed: a machine just made a life-altering call and nobody could say who was responsible — so a Pope sat down to answer the question your board is about to ask.
Six Themes That Map Straight onto Governance
Strip away the theology, which is genuinely substantial, and Magnifica Humanitas reads in places like a governance manifesto. It applies the five principles of Catholic social teaching, the common good, the universal destination of goods, subsidiarity, solidarity, and social justice, to artificial intelligence. Below are the six themes that matter most to anyone deploying AI, in plain terms.
The encyclical's pivotal claim (¶9): technology "takes on the characteristics of those who devise, finance, regulate and use it." Four parties shape every system, and each carries a share of responsibility for what it does.
Technology Is Never Neutral
The encyclical opens by demolishing the most common excuse in technology: that a tool is just a tool, and only its use is good or bad. In its ninth paragraph, Leo writes that "technology is never neutral, because it takes on the characteristics of those who devise, finance, regulate and use it." Among the AI builders and investors who have studied the document closely, this single line is repeatedly singled out as its most important sentence. A system carries the priorities, blind spots, and incentives of the people and companies that build and fund it. Pretending otherwise is how organizations avoid responsibility for what their systems actually do.
Read slowly, it assigns responsibility to four parties, whoever devises a system, finances it, regulates it, and uses it (see above), and puts a pointed question to each: are we building, funding, and using tools that help people, or ones that quietly harm them? For governance, the consequence is decisive. If technology is never neutral, "the algorithm decided" is never an acceptable answer; someone chose the data, the objective, the thresholds, and the deployment. Accountability cannot be outsourced to the math, an organization owns the behavior of the systems it deploys, full stop.
Governance translation: "The model did it" is not a defense. Whoever devised, financed, configured, and deployed the system carries its consequences. Build your accountability model on that assumption.
A Human Must Remain Accountable for Consequential Decisions
This is the heart of the document and its most quoted line: "Moral judgment cannot be reduced to calculation," and therefore "it is not permissible to entrust lethal or otherwise irreversible decisions to artificial systems." Leo insists that "responsibility must be clearly defined at every stage" and that there must always be "the possibility of identifying who must account for decisions." Accountability, he warns, must never be "collapsed into the machine."
Note the precise wording: lethal or otherwise irreversible. The principle is not "humans must approve everything," which would be impractical and is not what the encyclical says. It is that the more consequential and irreversible a decision, the more clearly a named human must own it. An AI can draft, sort, recommend, and accelerate; but for decisions that cannot be taken back, ending a life, terminating an employee, denying critical care, executing an irreversible financial action, a person must be the one who decides and answers for it.
Scripture frames this as the difference between calculation and wisdom. When Solomon is made king, he does not ask for processing power; he asks God for "an understanding heart to judge your people, that I may discern between good and evil" (1 Kings 3:9). That is precisely the faculty the encyclical says no machine possesses, "no computational system," Leo writes, "can create a heart that gives itself, or a conscience that discerns good from evil." A model can compute an output, but it cannot bear the weight of a decision. And Scripture insists that weight is unavoidable: "I have set before you life and death… therefore choose life" (Deuteronomy 30:19). Choosing is a human act, and so is answering for the choice.
This maps almost exactly onto the risk-tiered approach that good governance already uses. Low-stakes, reversible decisions can be heavily automated. High-stakes, irreversible ones require a human in the loop with real authority, not a rubber stamp. The encyclical gives that engineering instinct a moral spine.
Governance translation: Tier your decisions by reversibility and stakes. The more irreversible the outcome, the more explicit and authoritative the human accountability must be. Keep the chain of responsibility intact and named.
AI Affects Real Rights, Not Just Outputs
Leo punctures the idea that AI is a "purely technical matter." "The use of AI is never a purely technical matter," he writes; "when it enters processes that affect people's lives, it touches on rights, opportunities, status and freedom." The moment a model is deciding who gets hired, who gets a loan, who gets flagged, or who gets care, it has left the realm of pure engineering and entered the realm of rights.
The encyclical grounds this in a robust definition of human dignity: every person has worth "simply by virtue of existing, of having been willed, created and loved by God." This is the imago Dei of Genesis 1:27, "God created man in his own image", and it is the reason, in this tradition, that a person can never be reduced to a data point or a productivity score. You do not have to share the theology to grasp the governance consequence: dignity is not earned by usefulness or output, so a system that quietly ranks people by their value to the organization is doing something morally serious that demands oversight. As the prophet Micah compresses the whole ethic, the requirement is "to do justice, and to love mercy, and to walk humbly" (Micah 6:8), a standard of justice and humility that a purely optimizing system will not supply on its own.
The encyclical also pins the responsibility squarely on the people who build these systems. In a later passage on the duties of developers, it is unambiguous that encoding bias into a model is not a neutral technical choice but a moral failure, because that bias then scales to everyone the system touches. This is not abstract: independent AI-safety researchers have documented that today's leading models can carry measurable, systematic biases, including implicit valuations that weight some human lives or belief systems above others. Those values did not arise from nowhere; someone's choices, in the data and the training, put them there. A model that silently discriminates is the clearest possible case of a system "touching on rights" while no one is accountable for the harm.
Governance translation: Classify any AI use that touches a person's rights, opportunities, status, or freedom as high-impact by default, and govern it accordingly, with bias testing, transparency, appeal rights, and human review.
Truth Is a Common Good Worth Protecting
The encyclical is sharp on disinformation. "Tools that could foster dialogue and participation," Leo writes, "are often used to construct distorted narratives and blur the boundaries between truth and falsehood." Generative AI makes it trivially cheap to manufacture convincing falsehoods at scale, and a society that loses its shared grip on what is true loses its capacity to reason together and to govern itself.
The moral line here is ancient and blunt: "You shall not bear false witness against your neighbor" (Exodus 20:16). The Eighth Commandment did not anticipate synthetic media, but it names exactly what synthetic media industrializes, the manufacture of convincing lies about real people. Against that, the tradition holds out a promise that doubles as a design goal: "the truth will set you free" (John 8:32). An information ecosystem worth building is one that bends toward truth, not one engineered to blur it for engagement.
For organizations, this is both an external risk and an internal discipline. Externally, synthetic media and AI-generated fraud are now operational threats. Internally, an organization that deploys AI to manipulate, to dark-pattern customers, to fabricate reviews, to blur what is human-written and what is machine-generated, is corroding the same common good the encyclical defends, and accumulating reputational and regulatory risk while it does so. Provenance, disclosure, and honesty about what is AI-generated are becoming table stakes.
Governance translation: Treat truthfulness as a control objective. Disclose AI-generated content, preserve provenance, and prohibit deceptive uses, both to protect the public and to protect yourself.
Work Should Be Centered on the Person, Not Only Performance
Returning to the Rerum Novarum theme, Leo observes that automation "is rapidly transforming the very structure of work" and insists that systems must be "centered on the human person and not solely on performance." He goes further than abstract principle, naming concrete exploitation in the AI supply chain: the child labor in mineral extraction that feeds the hardware, and the underpaid, often invisible workers who label and moderate data to train the models. He warns, too, of "social control made possible by algorithmic systems" and the "massive collection of data."
This reaches back to the garden, where the human being is placed "to till it and to keep it" (Genesis 2:15): work as a form of stewardship and dignity, not mere output. Jesus sharpens the priority in a line that could be the encyclical's motto for automation, "The Sabbath was made for man, and not man for the Sabbath" (Mark 2:27). Systems are made for people, not people for the systems. And the warning against pure performance is older still: "what shall it profit a man, if he shall gain the whole world, and lose his own soul?" (Mark 8:36). An organization can win every efficiency metric and still hollow out the people it depends on.
The encyclical's second chapter confronts the question the AI-and-jobs debate usually dodges: what is a human actually worth? If a person's value is only their output, automation's logic is brutal, replace them the moment the machine is cheaper. But if a person has inherent value simply by existing, "the robot does it better" is not sufficient reason to discard them. That is the line between treating a workforce as a cost to optimize away and as people to whom the organization owes something, and the encyclical insists on the latter.
The governance lesson is that AI's human cost does not stop at your own employees. It extends through your supply chain and into the workers whose labor and data the technology depends on. Responsible deployment means asking not only "does this make us more productive?" but "what does this do to the people involved, ours and others'?"
Governance translation: Measure AI initiatives on human outcomes (augmentation, dignity, safety), not performance alone. Extend due diligence to the data and hardware supply chain, and limit surveillance of your own workforce.
Power Must Not Be Concentrated in a Few Hands
Leo is pointed about who actually controls AI. "The main drivers of development are private, often transnational, parties," he notes, which makes democratic governance "even more challenging." He invokes the principle of the universal destination of goods, arguing it should extend even to "patents, algorithms, and digital platforms," and warns against a world where a small group controls a general-purpose technology that affects everyone. His most striking formulation is the call to "disarm AI," which he is careful to define: "To disarm does not mean rejecting technology, but preventing it from dominating humanity."
This theme reaches its sharpest point in the encyclical's fifth chapter, on power and peace, where Leo condemns "the culture of power," calls for a ban on lethal autonomous weapons, and argues the "just war" theory is "now outdated." "No algorithm can make war morally acceptable," he writes. Scripture refuses to equate capability with authority: "Some trust in chariots, and some in horses: but we will remember the name of the Lord our God" (Psalm 20:7), the chariot being the apex military technology of its day. For most organizations the warfare application is not directly operational, but the underlying principle, that the ability to build something powerful does not confer the right to govern others with it, is a bracing corrective to the "move fast" reflex.
Governance translation: Resist single-vendor lock-in and unaccountable concentration. Insist on transparency, portability, and the ability to inspect and override the systems you depend on.
The thread running through all six themes: a human being is not a score. Dignity comes "simply by virtue of existing," so a person can never be reduced to the number a system assigns them.
Coming up in Part 3 — the single choice every organization now faces, and why a document about software ethics ignited a political firestorm in seventy-two hours.
Part 3 of 4
Babel or Jerusalem, and Why It Became a Political Firestorm
What you missed: six themes that turn an encyclical into a governance manifesto — technology is never neutral, a human must stay accountable, and AI touches real rights, not just outputs.
"The primary choice is not between a 'yes' or 'no' to technology, but rather between constructing Babel or rebuilding Jerusalem; between a power that claims to dominate the heavens and a people who work together… to rebuild the walls of fraternal coexistence."
Pope Leo XIV, Magnifica Humanitas (2026)
The Choice in Front of Every Organization
The encyclical's organizing image is a choice between two ways of building, and both come straight from Scripture. The first is the Tower of Babel (Genesis 11:1-9), where a unified humanity resolves, "let us build us a city and a tower, whose top may reach unto heaven; and let us make us a name." Babel is technology in the service of self-glorification and concentrated power, an impressive monument built to reach the heavens and overawe, and it ends in confusion and scattering. The second is Jerusalem, and specifically Nehemiah's account of rebuilding its broken walls (Nehemiah 2-6): a humbler, communal project in which ordinary people each repair the section in front of their own house, working "with all their heart" for the common good. Leo's own words frame the fork as "between a power that claims to dominate the heavens and a people who work together… to rebuild the walls of fraternal coexistence."
Leo is not romanticizing the past or rejecting progress; both Babel and Jerusalem are images of building. The question is what you build, for whom, and under whose control. The Babel impulse, "let us make us a name", is the impulse to scale and dominate for its own sake. The Jerusalem pattern is many hands, clear responsibility for one's own portion, and a shared end.
That Nehemiah pattern, build together, each repairing the wall in front of their own house, has long served faith-driven founders and builders as a practical blueprint, which is part of why it is striking to see the same biblical build-language now sitting at the heart of a papal encyclical on AI. And Scripture carries the image to its destination: the book of Revelation closes not with a tower clawing at heaven but with the New Jerusalem coming down, a city given and shared rather than seized (Revelation 21:2). The arc runs from Babel's grasping to Jerusalem's gift, and it is the arc every organization gets to choose its place in.
That same fork sits in front of every organization adopting AI. You can deploy it as a tool of pure efficiency and control, optimizing people out of the loop and treating accountability as friction. Or you can deploy it to augment human judgment, with people clearly accountable and dignity built into the design. The two roads look similar at the start and diverge sharply over time.
The encyclical's central image: the path toward "Babel" (concentrated, impersonal, dominating power) versus the path toward "Jerusalem" (human-scaled, accountable, built in communion). Both are technology; only one keeps the person at the center.
Two Ways to Deploy the Same Technology
- Decisions attributed to "the algorithm"; no one is clearly responsible.
- Consequential, irreversible actions automated to cut cost and friction.
- People measured purely by performance; dignity treated as overhead.
- Opaque, single-vendor systems no one inside can inspect or override.
- Truth and provenance sacrificed to engagement and persuasion.
- Every consequential decision has a named, accountable human owner.
- Irreversible actions require genuine human judgment, not a rubber stamp.
- AI augments people; human outcomes are measured alongside efficiency.
- Systems are transparent, inspectable, portable, and overridable.
- AI-generated content is disclosed; truthfulness is a design goal.
The technology is identical on both sides. The difference is governance: who is accountable, what stays human, and whether the person remains the point.
By The Numbers
35K+
Words Across Five Chapters
1.3B
Catholics Addressed Worldwide
1st
Encyclical of Leo XIV's Papacy
5
Social-Teaching Principles Applied to AI
The Stakes for Organizations
Organizations that let automated systems make consequential, hard-to-reverse decisions without clear human accountability face eroding stakeholder trust, dehumanized customer and employee outcomes, mounting regulatory and reputational exposure, and the strategic risk of building on a foundation that society, regulators, and now the world's largest moral institution are actively rejecting.
Why This Became a Political Firestorm in Seventy-Two Hours
A document about software ethics does not normally make the political pages. This one did, because it landed in a live Washington fight over whether and how to regulate AI, and because of who showed up to support it. At the May 25 presentation, alongside Cardinal Pietro Parolin and theologians, sat Christopher Olah, a co-founder of the AI company Anthropic who leads its interpretability team. Politico read the optics bluntly: the Pope had effectively chosen "Team Anthropic," the lab that brands itself on AI safety as a middle path between unrestricted acceleration and outright halt, and the encyclical struck a similar note.
The reaction split along the fault lines of the American AI debate. Microsoft's president Brad Smith welcomed it, noting that religions share "a starting point of humanity first," whereas much of tech "start[s] with the technology they're creating, and then think[s] about its impact on people second." On the other side, David Sacks, the influential former White House AI czar, warned the Pope wanted to "hand governments sweeping power over AI development in the name of safety"; former Trump AI adviser Dean Ball dismissed it as "a pretty weak document" amounting to "a deeply anti-American screed in favor of technocratic regulation." The document had touched a nerve.
No one embodies the tension more than Vice President JD Vance, the highest-ranking Catholic in American government, a longtime ally of Silicon Valley figures, and a defender of the administration's light-touch AI stance. Pressed by reporters before the release, he predicted he would "agree with some" of it and disagree with other parts. After reading it, he told NBC News on May 26 that "what I read of it sounds very profound, and the sort of thing that you would expect and hope from a leader of the Church." Two days later, in a commencement address at the United States Air Force Academy, he leaned into the part he could fully embrace, telling graduates that "decisions over life and death must be made by humans and not machines," and that they should "use technology to make you better, but never submit to it."
Strip away the partisanship and notice what survived it: even AI's most enthusiastic political champions endorsed the encyclical's central governance claim, that a human must remain accountable for irreversible decisions. The fight is only over how much regulation should enforce it. The principle itself has near-universal assent, and it is exactly the part that translates into how you run your AI program.
Where the Risk Lands: Principles to Governance Exposure
The following maps each of the encyclical's core principles to the concrete governance exposure an organization faces if it ignores that principle, and how urgent it is to address.
| Principle at Stake | Priority | Governance Exposure If Ignored |
|---|---|---|
| Human accountability for irreversible decisions | Critical | Liability with no clear owner, regulatory penalties under emerging AI-accountability rules, and reputational damage when "the algorithm decided" becomes the public explanation for a harmful outcome. |
| Decisions touching rights, status, freedom | Critical | Discrimination claims, denial-of-service to protected groups, and loss of license to operate when hiring, lending, or care decisions are made by unexamined models. |
| Technology is never neutral | High | A culture that treats systems as blameless tools fails to assign ownership, leaving harms unaddressed until they become crises. |
| Truth as a common good | High | Brand and legal exposure from undisclosed AI content, synthetic-media fraud, and deceptive practices that regulators and customers increasingly punish. |
| Work centered on the person | High | Workforce distrust, attrition, and supply-chain scrutiny over data-labeling labor and surveillance practices. |
| Avoiding concentration of power | Medium | Strategic risk from single-vendor lock-in: no ability to inspect, port, or override critical systems you do not control. |
Why This Reached the Vatican: The Race to Deploy Without a Compass
Why did the head of a two-thousand-year-old institution make AI the subject of his first major teaching? The same dynamic we write about in every issue: the technology is advancing far faster than the governance around it. Organizations are deploying AI into consequential workflows faster than they can answer basic questions about responsibility, fairness, and oversight. The capability races ahead; the compass is left behind. The encyclical's diagnosis is that this is not merely a technical lag but a moral one, systems optimized for performance, engagement, or cost, with human dignity treated as a constraint to minimize rather than the point of the exercise. Leo's response is not to slow the technology but to reverse the order: person first, then technology.
The encouraging part, and the reason this is an opportunity rather than just a warning, is that almost everything the encyclical asks for is achievable through governance, not by abandoning AI. You do not have to choose between capability and conscience; you have to build the structures that keep humans accountable while the machines do the work. That is a solved problem in principle. What is missing in most organizations is the discipline and the framework to implement it.
Coming up in Part 4 — the part you can act on: six concrete ways to put the human person at the center of your AI, plus a governance checklist to score yourself against.
Part 4 of 4
Six Ways to Put the Human Person at the Center
What you missed: two roads — "Babel" (optimization without accountability) and "Jerusalem" (augmentation with it) — and a 72-hour firestorm in which even AI's loudest champions endorsed the core principle: a human must own irreversible decisions.
Six Ways to Put the Human Person at the Center of Your AI
These six steps translate the encyclical's principles into concrete governance practice. None of them require slowing your AI adoption. They require deploying it in the "Jerusalem" mode, with accountability and dignity built in rather than bolted on later.
The encyclical's accountability principle in practice: the machine proposes, but a named, accountable human decides any consequential or irreversible action, and the chain of responsibility is never "collapsed into the machine."
Name an accountable human for every consequential decision
For each AI-influenced decision in your organization, you should be able to answer a single question instantly: who is accountable for this outcome? Maintain an explicit decision register that records, for every model in production, what it decides, how reversible those decisions are, and the named role that owns the result. This directly implements the encyclical's demand for "the possibility of identifying who must account for decisions."
The test is simple and unforgiving: if something goes wrong, can you name a person, not a system, who is responsible? If the only answer is "the model," you have a governance gap that the encyclical, your regulators, and eventually your lawyers will all flag.
Tier decisions by reversibility, and keep the irreversible ones human
The encyclical's line, "it is not permissible to entrust lethal or otherwise irreversible decisions to artificial systems," is a governance design principle. Classify your AI use cases along two axes: how consequential the outcome is, and how reversible it is. Reversible, low-stakes decisions can be fully automated. Irreversible or high-stakes ones, terminating employment, denying care, executing an irrevocable transaction, must route to a human with genuine authority to decline.
The danger is the rubber stamp: a "human in the loop" who approves whatever the model suggests because the queue is long and the model is usually right. Real accountability means the human has the information, the time, and the mandate to say no, and is measured on the quality of that judgment, not just throughput.
Treat any decision affecting a person's rights as high-impact by default
When AI "enters processes that affect people's lives," the encyclical says, "it touches on rights, opportunities, status and freedom." Build that trigger into your governance: any model that influences who gets hired, paid, promoted, served, insured, or flagged is automatically subject to your strictest controls, bias testing, documentation, transparency to the affected person, and a route to appeal to a human.
This is also where dignity becomes concrete. A person told "no" by an automated system, with no explanation and no human to appeal to, has been treated as an object to be sorted rather than a person owed an account. Designing in transparency and appeal is not just ethical; it is how you avoid the discrimination claims and regulatory action that opaque decisioning invites.
Make truthfulness and provenance a control objective
The encyclical warns against tools used "to construct distorted narratives and blur the boundaries between truth and falsehood." Operationally, that means two commitments. First, disclose AI-generated content to the people who receive it, and preserve provenance so you can always show what was machine-made. Second, prohibit deceptive uses of AI in your own operations, no fabricated reviews, no synthetic personas posing as real people, no dark patterns dressed up as personalization.
This protects the public, but it also protects you. The reputational and legal cost of being caught deceiving customers with AI is rising fast, and "we disclosed and kept provenance" is the difference between a defensible program and a scandal.
Measure AI on human outcomes, not performance alone
Systems must be "centered on the human person and not solely on performance." Add human-impact measures to every AI business case alongside the efficiency numbers: Does this augment people or simply replace judgment? What happens to the workers whose roles it changes? Does it increase or decrease the surveillance your own staff experience? Extend the same scrutiny down your supply chain to the data-labeling and content-moderation labor the technology depends on.
Two plain questions, used by seasoned investors to size up a business, capture most of this: Do you love your customer? (After someone uses your product, do they genuinely thrive, or are they worse off?) And do you love your employees? (Can the people who build your company earn a living wage and retire well?) An AI initiative that improves your metrics while failing both questions is exactly what the encyclical warns against.
And this is not charity at the expense of returns. There is growing evidence, including multi-year backtests across the Fortune 500 and Russell 2000, that companies which treat customers and employees well deliver competitive or superior long-term returns. Organizations that frame AI as augmentation rather than pure replacement likewise see better adoption, retention, and trust. The encyclical's "person first" principle, sound change management, and durable shareholder value point in the same direction.
Keep the power to inspect, override, and walk away
The encyclical's warning against concentrated, unaccountable power has a direct enterprise analogue: do not build your critical operations on a black box you cannot inspect, a single vendor you cannot leave, or an automated process you cannot override. Insist on transparency into how the systems you depend on behave, retain the ability to take a human back into the loop at any time, and avoid lock-in that would make accountability impossible.
"Disarming AI," in Leo's sense of "preventing it from dominating humanity," at the organizational level simply means never ceding so much control to a system that you can no longer answer for what it does. The off-switch, the audit trail, and the exit option are governance controls, not afterthoughts.
Governance Checklist
Does your AI program satisfy the accountability principles Magnifica Humanitas calls for?
Most organizations currently lack the controls marked with ✗. Closing even two or three of these gaps moves you decisively from the "Babel" column to the "Jerusalem" one.
This Is What AuthorityGate Was Built For
Magnifica Humanitas articulates, from the world's most prominent moral platform, the exact conviction AuthorityGate is built on: AI should be deployed for good, with humans accountable for consequential decisions. Our 8-gate model operationalizes that conviction. Gate 1 (Pre-Validation) classifies each use by stakes and reversibility before anything ships. Gate 7 (SME Approval) keeps a named, accountable human in control of consequential and irreversible actions, never "collapsing accountability into the machine." Gate 4 (Security Scan) and the framework's transparency requirements keep systems inspectable, and Gate 8 (Recovery Plan) preserves the ability to override and reverse.
The encyclical names the destination, a person-first, accountable use of AI. AuthorityGate is the road to it: the concrete checkpoints that turn "humans must stay accountable" from an aspiration into a verifiable, auditable practice.
The Bottom Line
When the leader of 1.3 billion people devotes the first major teaching of his papacy to artificial intelligence, and when the Vice President of the United States, Microsoft's president, and Silicon Valley's sharpest critics all feel compelled to respond within days, AI accountability has unambiguously become a governance question, not just an engineering one. Magnifica Humanitas did not invent the principle that humans must remain responsible for consequential decisions. It gave that principle the loudest possible voice.
The encyclical's genius is that it refuses the false choice between fearing AI and surrendering to it. It asks instead the only question that matters for the long run: will this technology be built and used in a way that keeps the human person at the center, with someone always accountable for what it does? Every organization deploying AI is answering that question right now, whether deliberately or by default.
The choice is, in the end, the one at the close of the Sermon on the Mount: the house built on rock, which stands when the storm comes, versus the house built on sand, which "fell: and great was the fall of it" (Matthew 7:24-27). An AI program built on accountability, dignity, and truth is built on rock. One built on speed and opacity, with no one answerable for what the system does, is built on sand, and the floods, regulatory, reputational, and moral, are already rising. Build on rock.
Our position has not changed; it has been vindicated. AI is a profound good when it augments human judgment and keeps people accountable, and a profound risk when it displaces both. The work is to build the governance that secures the first outcome and forecloses the second. That is what we are focused on, and it is exactly what this moment calls for.
This article is part of our AI governance newsletter series. Subscribe to receive complete analyses with principle-to-risk mappings, governance checklists, and actionable recommendations.
