When the Vatican released Pope Leo XIV’s Magnifica Humanitas on May 25, the document did something the AI debate often avoids. Before asking how powerful the systems are, it asked what they do to the person who has to live under them.
That is why the encyclical deserves a serious reading. Much of it is careful, even wise, about AI and human dignity. I read it with respect, and with growing frustration, because the Pope’s cure would deepen the disease he sees.
The encyclical is strongest when the Pope refuses to mistake the machine’s sentences for understanding. Software can calculate and write smoothly, but there is nobody inside it. A machine is never formed into a self that must answer before God and neighbor. Love and responsibility belong to persons.
To see the problem, you do not need a theology degree. It shows up in ordinary places: a lonely conversation with a chatbot, a classroom essay whose authorship is unclear, hiring software screening a woman before anyone reads her name. War pushes the question into darker territory, where code starts to compress decisions that should remain heavy on a conscience.
On this, the Pope deserves credit. He sees why the protection of children belongs in the same argument as surveillance and manipulated attention, because all of it changes the conditions under which people learn to choose.
The argument weakens when the Pope moves from diagnosis to cure. After making a strong case for human dignity, he turns toward a remedy that gives more discretion to governments and multilateral bodies, with large firms never far from the room.
What the Pope gets right
Technology stops being neutral when it starts sorting human lives. If a system helps decide whether someone gets work or credit, or whether she comes under watch and disappears from view, moral responsibility is already involved.
Much of the data and computing infrastructure sits in a few private hands. Market share is only part of the danger. Old human judgments can return dressed in machine language, so a score or ranking ends in denial with no person in sight and no obvious appeal. That is a moral problem.
The Pope also rejects the fantasy that a machine can become a moral agent. A machine may rank targets or produce a confidence score, but guilt and responsibility stay with human beings. Put lethal force behind software, and accountability starts to move out of reach.
His warning about autonomous weapons matters because lethal force must remain under accountable human control. The closer software moves to irreversible decisions, the more carefully we should slow it down.
No algorithm can make war morally acceptable.
He is right.
Where the cure goes wrong
The trouble starts when the Pope invokes subsidiarity and then moves away from it. Subsidiarity is one of Catholic social teaching’s best principles because it says power should sit close to the people whose lives it touches. Responsibility is learned close to home, first in families and churches, then through work and local life. Expertise becomes more dangerous when it moves too far from those places.
After defending that principle, the encyclical turns toward regulation from above. In the age of AI, the Pope writes,
it is no longer possible to rely solely on the ‘invisible hand’ of the market.
He then points toward law backed by public authority and international cooperation.
Some law belongs in any decent order. Serious defenders of markets believe in liability and in courts that can respond when harm occurs. Liberty does not let a company poison the well and call it innovation.
The danger begins when limited legal work becomes a global rulemaking project. At that point, the issue shifts from punishment for harm to advance permission to build. That favors the institutions already seated at the table. The Pope’s own warning should press against his cure:
A more moral AI is not enough if that morality is determined by a few.
The warning still applies when the few are regulators and diplomats instead of engineers and executives.
Rules like that are never written in the abstract. Powerful countries arrive first; permanent agencies know how to stay, and corporate legal departments can afford the time. By the time lobbyists are moving between Brussels and Washington, an entrepreneur in Dakar may have no representative there.
In Dakar and Kigali, people are testing practical tools without policy departments or lawyers on retainer. No rural school in Senegal can pause class while committees in Brussels or Geneva decide which models are acceptable. Giants can absorb rules built for giants, but the same rule can decide whether a small team close to the ground ever launches.
That is how regulation meant to check concentrated power ends up licensing it, especially for countries absent when the rules were written.
Who pays for control
The Pope is trying to speak for people with the least power, and I share that concern. His examples point toward a woman judged by a system she cannot see and a worker buried at the bottom of a supply chain. I would add the countries always told to wait their turn, because a cure built around permission would narrow their path.
AI may lower the cost of expertise far from elite institutions. Before class, a translated lesson can be modest and still useful; so can a nurse’s symptom check while the doctor is on the way. Sometimes the first answer is what keeps work moving until the expert arrives.
The World Bank’s recent work on small AI puts this in development language, with practical tools for ordinary devices and local conditions. It also notes the scale of the divide: 2.2 billion people remain offline, and low-income countries account for less than one percent of ChatGPT usage. Any agenda built around human dignity has to begin with that gap.
For many poor communities, access is still the immediate problem. They need cheaper ways to reach and adapt the technology before another authority tells them what is permitted.
The practical work starts with what people can actually use. Internet access has to be cheap enough to matter, and tools have to fit local languages and habits. People need enough know-how to turn them into work. Courts still matter, because fraud should be punished without smothering business.
The World Bank’s older work on digital development is useful here too. Technology pays off where a new business can challenge the old one and where a signed contract means something. It does far less good when every new tool needs a distant blessing before anyone poor can use it.
International redistribution can sound merciful, but power often moves upward before opportunity moves outward. The paperwork satisfies the ministry and the donor, while the contract still tends to land with someone connected. Ordinary people keep waiting for approval.
Better paperwork does not end dependency. Courts have to enforce contracts, and business owners need protection they can actually use. New businesses need room to start before gatekeepers decide whether they are safe.
Here the encyclical sometimes blurs protection with control. Protection gives people a remedy after harm and draws hard lines where power becomes dangerous. Control lets a political class decide in advance which technologies count as legitimate and whose gains deserve blessing.
In poor countries, that distinction is practical rather than theoretical, because permission is rarely neutral when the official holding the stamp also controls whether you get to work. Licenses and inspections can become payments or negotiations with power. The same wall rises in technology when scale depends less on customers than on ministries or foreign platforms.
Announcements about protection can leave people unfree. Freedom means people can trade, own what they build, and challenge abuse in court without asking a superior first.
A better AI agenda
Markets alone leave some moral questions unanswered. They need law because property has to be protected and contracts have to mean what they say. Competition law and liability matter too; business can turn predatory when nobody is answerable. Limits on violence are the floor.
Political power has the same problem when it becomes another monopoly dressed up as justice.
The harder question is which institutions preserve human dignity in practice.
A better agenda starts where dignity is directly attacked. Autonomous lethal force belongs behind a hard legal wall, along with state machinery that scores citizens or tracks bodies at scale. Children deserve protection from addiction loops. When an automated system denies credit, work, or public services, the person affected should get an explanation and a route of appeal. A company that knowingly deploys software that causes serious harm should face legal responsibility.
Once the law has drawn those lines, people need room to build. Open models should be able to compete with closed ones.
A founder in Nairobi or Manila should not need the legal budget of a global bank to use AI, and she may know the nearby farm or clinic better than any global board. Locked systems should give users a way out, while dominant platforms leave space for challengers and local institutions get room to experiment.
Distant planners cannot know which AI tools will matter in Bamako or Manila before people there try them.
The Pope worries, rightly, that AI could subject humanity to the rule of technique. A global approval machine would train people to wait for permission, even with ethical language and panels. Agency would move away from ordinary people and toward remote managers, leaving dignity weaker under a nicer name.
The old lesson of Pope Leo XIII
Pope Leo XIV chose his name with Pope Leo XIII in view. In 1891, Pope Leo XIII’s Rerum Novarum spoke to the industrial age. The fight over work and property carried a deeper moral question: whether human beings would be treated as persons or instruments. That tradition belongs in this debate.
The industrial age taught a different lesson. Human beings need institutions strong enough to defend them and open enough to leave room for business, invention, and voluntary cooperation. Family and faith have work to do here too.
The AI age needs that discipline. Law has to stop abuse without making every new tool wait for official approval. Human beings must remain responsible for human decisions, and when the largest firms try to close markets, the law should keep them open. Beyond those guardrails, people need the freedom to use tools that can help them escape dependency.
Pope Leo XIV sees the danger of a future controlled by a few powerful actors. But moving control into a smaller circle of regulators and diplomats would still leave the largest firms with seats at the table.
Magnifica Humanitas means humanity in its grandeur. For me, that grandeur includes people usually asked to receive rules made elsewhere.
A technology built for human dignity has to reach people close enough to the problem to see it clearly. In the world I care about, it is a phone in a field as much as a policy paper in Rome. It is also the nurse before morning and the teacher translating a lesson after midnight. Do not tell those people to wait while the powerful write rules for them.
Rome is right to warn the powerful. It should also trust the people who live closest to the problems.
Guardrails, not mandates. Structure, not absolute control from afar, created and driven by bureaucrats. As usual, Magatte and I agree!
What a clear, cogent, and balanced response. Thank you, Magatte