VATICAN CITY: Pope Leo XIV released his first encyclical on May 25, 2026, and did something no previous pope has done: he stood at a press conference and presented the document himself. Seated next to him on the Vatican dais was Chris Olah, the 33-year-old Canadian co-founder of Anthropic and head of the company’s mechanistic interpretability research, according to the National Catholic Reporter’s on-the-ground recap and a primary-source transcript of Olah’s remarks published by Anthropic.
The encyclical, titled “Magnifica Humanitas: On Safeguarding the Human Person in the Time of Artificial Intelligence,” calls for “robust regulation” of frontier AI, prescribes “disarmament” of the most powerful systems, and warns that AI threatens to “dominate humanity” if its gains are not “shared globally,” per Vatican News. The Vatican itself issued a quiet caveat afterward: the encyclical should not be read as a “church blessing” of Anthropic.
Why did Anthropic, not OpenAI or Google, get the Vatican slot?
The short answer is that Anthropic has been working on this for years. Religion News Service traces the relationship to 2016, when Olah first began corresponding with Vatican officials interested in machine-learning interpretability. The company hosted two private gatherings of Christian leaders at its San Francisco headquarters in March and April 2026, and credits three Catholic thinkers — including Bishop Paul Tighe, secretary of the Vatican’s Dicastery for Culture and Education — as contributors to the Claude Constitution.
That groundwork bought Anthropic something OpenAI and Google have not built: a working relationship with one of the few institutions on earth that is older than the modern nation-state and that frames moral authority on technology in language regulators and reporters are willing to repeat. In his Vatican remarks, Olah said frontier AI labs operate under “incentives and constraints that can conflict with doing the right thing” — commercial pressure, geopolitical pressure, the race to stay at the research frontier — and that outside critics not bound by those incentives are “enormously important” if the technology is going to go well.
What does the encyclical actually say about AI?
The document runs 235 pages and is the first papal encyclical fully dedicated to artificial intelligence. Its core argument, per CNN’s coverage, is that AI must be subordinated to “the human person” — meaning regulation is required, autonomous weapons systems should be banned, and the economic gains of AI should not concentrate in “a handful of wealthy nations.” Leo also calls for what the document terms “disarmament” of the most powerful frontier systems, a striking word choice that maps cleanly onto Anthropic’s longstanding public framing of advanced AI as a national-security category.
What the encyclical does not do is endorse any specific company, regulation, or technical approach. It is a moral framework, not a policy paper. But the framework happens to align almost point-for-point with Anthropic’s stated public positions on AI safety, frontier model evaluations, and the need for binding government oversight.
How does the Vatican appearance fit Anthropic’s broader regulation playbook?
This is the part founders should pay attention to. The Vatican stage is one move in a pattern. Anthropic donated roughly $20 million to a political group backing pro-regulation AI candidates ahead of the 2026 midterms. It was the most prominent corporate backer of California’s SB53, the first transparency law aimed at frontier AI developers, which Governor Newsom signed in September 2025. It is now publicly aligning with the Catholic Church on a global ethics framework that calls for “robust regulation” — at exactly the moment the company is reportedly set to close a $30 billion round at a $900 billion valuation, surpassing OpenAI as the world’s most valuable AI startup, per Bloomberg reporting.
The pattern echoes Anthropic’s recent run of institutional partnerships, including its $200 million health partnership with the Gates Foundation in May and its $100 billion-class AWS expansion with Amazon. Each move trades short-term optionality for long-term legitimacy with the institutions that decide what AI is allowed to do.
The David Sacks counter-argument
The political backdrop to the Vatican appearance is open warfare with the White House. David Sacks, the Trump administration’s AI and crypto czar, has spent months publicly accusing Anthropic of “a sophisticated regulatory capture strategy based on fear-mongering,” writing on X that the company is “principally responsible for the state regulatory frenzy that is damaging the startup ecosystem.” Sacks has also said that, on its current trajectory, “Anthropic will be the most powerful monopoly ever created in human history.”
The Sacks read is the one Anthropic’s critics will lean on: that the company is using safety language and institutional partnerships to raise the regulatory cost of entry for everyone else, locking in the lead it already has. The Anthropic read is that frontier AI is a different category of technology that requires outside moral checks, and that the alternative — racing on commercial incentives alone — is worse. Both reads can be true at the same time. The question for founders is which one matters more for their own product roadmap.
What this means for founders
For builders, the Vatican appearance is a signal about how the AI policy fight will be argued for the next 18 months. The frontier labs that win the regulatory framing get to set the cost of compliance for everyone downstream. Anthropic is making a deliberate bet that “AI safety” wins that framing fight at the global level, and it is willing to spend brand capital, founder time, and political capital to win it. Founders building on Claude should expect that bet to show up in pricing, in feature gating, and in the rules they have to follow about model evaluations, agent permissions, and government reporting.
The closer milestone to watch is the round itself. If Anthropic closes at $900 billion this week, the company will sit at a valuation that is hard to read as anything other than a structural lead — and the Vatican photo will read, in retrospect, as one of the more sophisticated brand-positioning plays in tech history. The next signal will be whether OpenAI, Google DeepMind, or xAI try to build their own institutional counterweights, or whether they cede the moral-legitimacy lane to the company that got there first.
