From AI Dressing Pope Francis in a Puffer Jacket to Pope Leo Addressing AI
A few years ago, the internet briefly convinced itself that Pope Francis was walking around in a white puffer jacket. The image was fake, of course, but it was also a warning shot. Artificial intelligence had reached the point where it could fool millions of people with something as simple as a coat. Now, his successor, Pope Leo XIV, has taken the issue on directly in Magnifica Humanitas, an encyclical on safeguarding the human person in the age of artificial intelligence.
Although the encyclical is not written for courts, it lands directly in the middle of the conversation the justice system is now having about artificial intelligence. Pope Leo does not treat technology as inherently evil. He recognizes that it can “heal, connect, educate, and protect our common home…” But he also warns that technology is never neutral in practice because it reflects the choices of those who design it, finance it, regulate it, and use it. The concern, then, is not whether courts should use technology. They should. The concern is whether the justice system will use AI in a way that protects human dignity or in a way that slowly reduces people to data, predictions, rankings, and outputs.
Courts cannot treat that distinction as academic. The justice system is built on a social contract that depends on human responsibility. People may disagree with judges. They may be angry at rulings, frustrated by delay, or convinced that the law should be different. But the legitimacy of the system rests on a basic understanding that human beings, bound by law, oath, conscience, reason, and public accountability, are making the decisions. It is not a promise that judges will always be right. It is a promise that decisions affecting liberty, custody, property, and other constitutional rights will be made by human beings who can be questioned, reviewed and corrected.
For courts, AI cannot be reduced to whether the technology is impressive or efficient. We did not agree to be judged by a machine. We did not agree that the most serious decisions in our public life would be resolved by an automated system trained on data we cannot see, applying weights we cannot test, influenced by design choices we did not make, and producing outputs no one can fully explain. We certainly did not agree that the moral responsibility of judgment could be transferred to a tool simply because the tool is fast, confident, or useful.
None of this means AI has no place in the justice system. It does. AI can help organize records, summarize pleadings, identify procedural issues, check citations, translate language, improve access, and support judges, lawyers, clerks, and litigants who are overwhelmed by the volume and complexity of modern legal work. Used carefully, it may become one of the most useful tools courts have ever had, but support is not decision, assistance is not judgment, and efficiency is not legitimacy. The line matters because once the tool begins to decide, the social contract begins to change.
Pope Leo’s warning names the deeper risk. He cautions that we are handing consequential decisions to automated systems that do not know compassion, mercy, and forgiveness. The danger is not only that AI may hallucinate cases or produce false quotations, although that risk is real and courts are already seeing it. The deeper danger is that we begin to accept a system in which human responsibility becomes blurred. The model suggested it, the system ranked it, the tool flagged it, the algorithm recommended it. Those phrases may sound modern, but they are not enough in a justice system because someone must still be responsible for the decision that follows.
The human being must go first and then AI may enter the workflow after that, but it cannot become the first voice in the room and it certainly cannot become the last. The white page still matters because it forces the human mind to engage before the machine begins to frame the issue, and that moment of engagement is not wasted time. It is part of the discipline that keeps responsibility attached to the person who must ultimately answer for the work.
The social contract behind the justice system was never built on the assumption that judges are perfect. It was built around the reality that judges are human, and because they are human, they can be wrong. That is why the system requires reasons, preserves records, allows parties to object, permits review, and provides a path for errors to be corrected. Those safeguards are not technical details. They are part of the bargain. They only work when a responsible human being remains at the center of the decision and can answer for the judgment that has been made.
An AI system cannot take an oath, exercise conscience, show mercy, or understand the weight of a sentence, the fear of a parent, the dignity of a victim, the credibility of a witness, or the public trust carried by a court’s signature. It can produce language about those things, sometimes very persuasive language, but it cannot bear responsibility for them. Allowing AI to decide, then, is not merely a questionable workflow choice.
Courts should use technology, modernize, reduce delay, improve access, and use every responsible tool available to help the people who come before them. But they must do so without forgetting what makes a court a court. The robe is not a symbol of speed or efficiency. It is a symbol of authority, neutrality, and justice, and those ideals require a human being who remains responsible for the judgment. AI can help us rebuild the justice system, but it cannot tell us what justice requires, what dignity demands, or what judgment should cost the person entrusted to make it.
That is not the deal.
