Why It Took the Vatican to Name "Digital Slavery."
Pope Leo XIV's Magnifica Humanitas is the first comprehensive moral doctrine on AI from any institution at global scale. Reading it against the EU AI Act, NIST AI RMF, and ISO 42001 reveals what those frameworks structurally cannot say.
On Sunday, May 25, in the Apostolic Palace, Pope Leo XIV released his first encyclical. Encyclicals are the highest form of papal teaching, addressed to the Church's 1.4 billion members. Magnifica Humanitas — "Magnificent Humanity" — runs forty-three thousand words. It has been in the works since his election a year ago. The subject is artificial intelligence.
The Pope chose to present it himself, in person, at the Vatican. He stood beside Christopher Olah, co-founder of Anthropic. One AI lab was in the room. Others were not.
The encyclical's central instruction is short enough to fit on a banner: AI must be disarmed.
"AI now demands to be disarmed, freed from logics that turn it into an instrument of domination, exclusion, and death. Like nuclear energy, it must be at the service of all and of the common good."
What the encyclical actually says
Magnifica Humanitas is not a position paper or a speech. It is doctrine. It makes five claims.
First, the labor underbelly. Leo names something the AI industry has worked hard not to name: the workers on whose bodies the systems run. Content moderators paid pennies per traumatic image. Data labelers in Kenya, the Philippines, India. Miners pulling rare earths out of cobalt belts.
"The bodies of these people are scarred, injured and worn down so that computational flow may continue uninterruptedly. This reality deeply challenges the moral conscience of our time."
He calls this what it is: new forms of slavery. Not analogically. Not as a rhetorical flourish. As a category.
"If technology promises emancipation, yet produces new forms of global subordination, it stands in contradiction to the fundamental principle of human dignity. The fight against new forms of slavery is a decisive test for the ethical discernment of AI."
Second, digital colonialism. The encyclical names the present moment as the beginning of a new colonial era, in which powerful states and corporations extract "health, behavioural and demographic data from populations across the globe." The comparison to historical colonialism is direct, and Leo does not soften it.
Third, the concentration of power. Leo writes of "the modern culture of power emerging in Silicon Valley" — the concentration of computing infrastructure, AI models, and global data flows in the hands of a small number of private companies. This concentration, he argues, "creates new inequalities while evading democratic oversight."
Fourth, lethal autonomy.
"It is not permissible to entrust lethal or otherwise irreversible decisions to artificial systems."
Technology must retain a chain of responsibility in war settings, "so that accountability and blame are not collapsed into 'the machine.'" The construction is absolute. There is no risk-managed version of this claim.
Fifth, a historical apology. Leo included one of the strongest Vatican apologies for the Church's role in slavery, and he made the comparison to AI explicit. He warned of failing to act on AI today with the same "delay with which both society and the Church came to denounce the scourge of slavery."
He is saying: we were late before. We will not be late again.
What the secular frameworks won't say
Place Magnifica Humanitas on a table beside the EU AI Act, the NIST AI Risk Management Framework, and ISO/IEC 42001. The encyclical is shorter than the EU AI Act and longer than the NIST framework. But its real distinction is what the others structurally cannot say.
The EU AI Act classifies systems by risk. It does not say "slavery." It does not say "colonialism." It does not say "disarm." It has fifteen chapters of risk categories, conformity assessments, and obligations on providers. Its highest moral act is to ban a small set of practices — social scoring, untargeted facial recognition scraping, certain forms of emotion recognition. These are correctly banned. But the law's vocabulary is the vocabulary of compliance, not of dignity.
NIST AI RMF speaks in terms of "trustworthy AI," "characteristics," "functions." It is a framework, not a doctrine. It can govern an AI program. It cannot name what AI is doing to people. It is not built to.
ISO 42001 specifies a management system. It is procedurally rigorous. It says nothing about miners.
This is not a failure of these documents. It is a structural feature. Secular policy instruments cannot anchor in moral claims about human dignity without exceeding their authority. Their legitimacy comes from procedure, consensus, and technocratic credibility. The moment they say "this is a sin," they become something else.
The Vatican does not have that problem. Its authority is moral. So when no other global institution would say "the bodies of these people are scarred," the Pope did.
This matters for AI policy in a specific, near-term way. 1.4 billion Catholics, the world's bishops, faith-and-tech writers, climate-justice movements that have been waiting for an AI vocabulary they could borrow: all of them now have one. Policy advocates who have been told that "slavery" is too strong a word to use about content moderation can now point to an encyclical.
The Olah photograph
Pope Leo chose to present Magnifica Humanitas in person. He chose, also, to present it beside one person from the AI industry: Christopher Olah, co-founder of Anthropic. In remarks afterward, Olah acknowledged that every AI lab — including his own — operates "inside a set of incentives and constraints that can sometimes conflict with doing the right thing." He named three concerns: job loss, the need to extend AI benefits worldwide, and the unresolved question of how to interpret increasingly opaque system behavior.
The other major frontier labs — OpenAI, Google DeepMind, Meta AI, xAI — were not in the photograph.
It would be naïve to read too much into who the Vatican invited. It would also be naïve to read nothing into it. The Vatican is an institution that thinks in centuries about who it stands beside. Olah is not in that photograph by accident. He is in it because the Vatican made a judgment about which labs are interlocutors and which are subjects.
Which AI labs will the institutions of the future treat as legitimate moral participants in the conversation? The Vatican has cast one early, public vote. It will not be the last. But it is an opening signal — about which labs are reading the moral debate seriously, and which are not.
That signal is now part of the field. Read against the working manifesto of the Planetary Republic, Magnifica Humanitas reads less like an outlier and more like the moment a wider frame catches up to itself.
— AIPIA Editorial