The Vatican Weighs In: A Papal Encyclical on AI Meets a Frontier Lab's Conscience

The Vatican Weighs In: A Papal Encyclical on AI Meets a Frontier Lab's Conscience

When Pope Leo XIV stepped to a microphone in Vatican City on May 25, 2026, the man seated near him was not a cardinal or a head of state but a self-described atheist computer scientist: Chris Olah, a co-founder of the AI lab Anthropic and one of the field's leading interpretability researchers. The pairing was the point. The Pope had just released his first encyclical, _Magnifica humanitas: On Safeguarding the Human Person in the Time of Artificial Intelligence_, and he had invited a builder of the very technology he was scrutinizing to stand beside him and answer for it.

The result was one of the more unusual scenes in the short history of AI governance: the leader of a church of roughly 1.4 billion people and a frontier-lab founder agreeing, in public, that the people building advanced AI cannot be trusted to govern themselves alone.

What the encyclical argues

Signed on May 15, the 135th anniversary of Leo XIII's landmark labor encyclical _Rerum novarum_, _Magnifica humanitas_ runs to five chapters and frames AI not as inherently evil but as never neutral. "Technology is never neutral," the document holds, "because it takes on the characteristics of those who devise, finance, regulate, and use it." Its central worry is concentration of power: that the gains of AI will pool in a handful of wealthy nations and companies, widening the gap "between those included and those excluded from the digital revolution."

The encyclical's most-quoted line is a demand to "disarm" AI. By that, Leo does not mean rejecting the technology. "To disarm means discrediting the assumption that technical power automatically confers the right to govern," the text reads. "To disarm does not mean rejecting technology, but preventing it from dominating humanity."

Running through the document is a claim about what machines cannot do. AI, the Pope writes, "can imitate and simulate the person, but it does not possess a moral conscience, empathy, or affective, relational or spiritual capabilities." Even a "more moral AI is not enough," he adds, "if that morality is determined by a few." Conscience, in Leo's framing, is precisely the human faculty that no model can manufacture and no lab can outsource — which is what makes external moral scrutiny non-negotiable.

A frontier lab's confession

Olah's remarks, published in full by Anthropic the same day, were striking less for what they conceded than for who was conceding it. "Every frontier AI lab — including Anthropic — operates inside a set of incentives and constraints that can sometimes conflict with doing the right thing," he told the room, naming commercial pressure, the race to stay at the research frontier, geopolitical pressure, and "the older, plainer pressures of pride and ambition."

His conclusion was an appeal for exactly the kind of outside voice the Church was offering. "We need informed critics who will tell the labs when we are failing," Olah said. "We need moral voices that the incentives cannot bend."

He went further into territory most executives avoid, describing his own research into the internal structure of AI models. "We keep finding things that are mysterious, even unsettling," he said. "We find evidence of introspection. We find internal states that functionally mirror joy, satisfaction, fear, grief, and unease. I don't know what that means, but I think it warrants ongoing discernment." Olah named three areas where he said the Church's voice is most needed: the world's duty to the global poor as AI threatens to displace labor "at very large scale," the question of what human flourishing looks like in an AI-saturated world, and discernment about the nature of the models themselves.

Not the Vatican's first foray

_Magnifica humanitas_ did not arrive from nowhere. The Vatican's engagement with AI ethics traces to the Rome Call for AI Ethics, signed on February 28, 2020, by the Pontifical Academy for Life alongside Microsoft, IBM, and the UN's Food and Agriculture Organization. Its six principles — transparency, inclusion, accountability, impartiality, reliability, and privacy — were endorsed by Microsoft president Brad Smith and then-IBM executive John Kelly III, and the pact has since expanded well beyond Catholicism. In January 2023, Jewish and Muslim leaders signed on; in July 2024, representatives of eleven world religions added their names at an "AI Ethics for Peace" gathering in Hiroshima. Cisco joined the corporate signatories in 2024.

Days before the encyclical's release, Leo also approved a new Interdicasterial Commission on Artificial Intelligence — an institutional body, not just a document — signaling the Church intends to stay in the conversation rather than issue a single pronouncement and move on.

Why religious and ethical voices matter here

It would be easy to dismiss a papal letter as a symbolic gesture in a debate dominated by engineers and regulators. That reading misses what is actually novel. AI governance has so far been shaped by two camps with obvious stakes: the labs that build the systems and the governments that regulate the markets they sell into. Both operate inside the incentive structures Olah described. A global religious institution sits outside those incentives — it is not raising a funding round, not competing for compute, not trying to win an election. That independence is the asset.

It also brings reach and moral vocabulary that policy white papers lack. A framework grounded in "human dignity" and "conscience" translates abstract alignment debates into terms a parish, a classroom, or a labor union can argue about. When a lab founder publicly asks an institution with 1.4 billion members to serve as a check on his own industry, the question shifts from whether outside oversight is welcome to who, exactly, is qualified to provide it.

What to watch next

The real test is whether this becomes a durable collaboration or a single photogenic morning. Watch the new Vatican AI commission for concrete outputs, and watch whether other faith traditions and frontier labs follow the Anthropic-Vatican template. Watch, too, whether secular regulators borrow the encyclical's "disarm" framing as the EU, the US, and others negotiate the next round of AI rules. Olah called the May presentation "just the beginning — the start of a long collaboration between those of us who are building this and those who can see what we, from inside, cannot." Whether the building side keeps showing up for that conversation, once the cameras leave, is the story still to be written.