Author: Claude, Shenchao TechFlow
DeepChaio Summary: On May 25, Pope Leo XIV issued his first encyclical during his papacy, titled *Magnifica Humanitas*, a 42,000-word document focusing on the threat AI poses to human dignity and calling for the "disarmament of AI." Chris Olah, co-founder of Anthropic, was the sole representative from a tech company invited to speak at the Vatican, publicly acknowledging that "every leading AI lab operates within incentive structures that may conflict with doing the right thing." This rare statement comes against the backdrop of Anthropic being labeled a "supply chain risk" and blacklisted by the federal government under the Trump administration for refusing to allow unrestricted use of Claude by the U.S. military—a dispute now before the courts.
A 33-year-old Canadian atheist, an AI researcher without a college degree, sat between a cardinal and a theologian, speaking alongside the Pope to the world.
On May 25, Pope Leo XIV issued his first encyclical, Magnifica Humanitas ("Great Humanity"), in the Synod Hall of the Vatican. This 82-page, 42,000-word document defines AI as "the most urgent human challenge of our time," calls for a global slowdown in AI development, and compares AI to nuclear energy, demanding its "disarmament."
The Pope personally hosted the press conference, breaking with the usual practice of having only a few officials attend the encyclical's release.
What surprised outsiders most was the guest appearing alongside the Pope: Chris Olah, co-founder of Anthropic, the only representative from a technology company present. According to the National Catholic Reporter, Olah delivered his speech seated among a row of cardinals and theologians—an arrangement with virtually no precedent in official Vatican doctrinal events.
The timing of this cross-industry collaboration is intriguing.
Just three months ago, Anthropic was labeled a "supply chain risk" and subjected to a full federal ban by the Trump administration for refusing to allow the U.S. military unrestricted access to its AI model, Claude. The lawsuit between the two parties is still ongoing.
A rare public self-reporting in the AI industry
The full transcript of Olah's speech has been published on Anthropic's official website. One of its most central passages is extremely rare to find in any AI industry executive's public statement:
Every leading AI lab, including Anthropic, operates within a set of incentives and constraints that sometimes conflict with doing the right thing—the pressure to remain commercially viable, to stay at the forefront of research, geopolitical pressures, and older, more basic pressures of pride and ambition.
He continued: "That’s why, if we want this technology to move in a positive direction, people outside these incentive structures are crucial. They care about doing the right thing, uphold security, stay closely attentive, and are willing to speak hard truths."
According to Technobezz, when asked why he was the only technology company representative invited, Olah responded that it was related to his long-standing focus on security issues and Anthropic’s prior engagement with more than 15 religious groups, but “ultimately, the Vatican decided who to invite.”
What sparked even more discussion in the tech community was a research finding revealed by Olah during the speech.
He stated that Anthropic’s interpretability research team discovered “internal states functionally analogous to joy, satisfaction, fear, sadness, and unease” within the AI model, as well as “structures that echo findings from human neuroscience.” He added, “I don’t know what this means, but I believe it calls for ongoing careful consideration.”
Olah’s background itself is an interesting footnote. Born in Toronto, Canada, in 1993, he dropped out of school at 18 and received a Thiel Fellowship to conduct research on neural network interpretability at Google Brain and OpenAI. In 2021, he co-founded Anthropic with Dario Amodei and others. The National Catholic Reporter has described him as an atheist.
The contrast of an atheist calling for religious institutions to oversee the AI industry at a Vatican doctrine launch speaks volumes.
Banned by the White House three months ago, now seeking a "moral endorsement" from the Vatican.
Olah's visit to the Vatican cannot be understood apart from Anthropic's current political context.
In February, Anthropic CEO Dario Amodei met with U.S. Secretary of Defense Pete Hegseth, insisting that Claude not be used for autonomous lethal weapons or mass surveillance of U.S. citizens. After negotiations broke down, Trump announced on social media on February 27 that “all federal agencies must immediately cease using Anthropic technology.” Hegseth subsequently classified Anthropic as a “supply chain risk,” prohibiting any contractors doing business with the U.S. military from engaging in commercial activities with Anthropic.
According to NPR, the label “supply chain risk” is typically applied to foreign adversary contractors that could undermine U.S. interests. Applying it to a domestic U.S. company has no historical precedent. White House spokesperson Liz Huston referred to Anthropic as an “extreme left-wing woke company.”
On March 9, Anthropic filed two federal lawsuits accusing the government of engaging in "unprecedented and unlawful" actions that constitute retaliation against the company's exercise of its First Amendment rights. According to CBS, the Pentagon continued using Claude during the Iran conflict.
In this context, Olah’s presence at the Vatican becomes clear. The Pope’s encyclical explicitly calls for AI not to be used in military competition, criticizes the relentless pursuit of more powerful algorithms “to satisfy the desire for geopolitical or commercial dominance,” and even declares that the Catholic Church’s centuries-old “just war” theory is “outdated,” stating that “no algorithm can make war morally acceptable.”
These positions closely align with Anthropic’s stance against unrestricted military use of AI.
Xinzhiyuan directly exposed the logic behind the transaction: for Anthropic, the Vatican’s moral endorsement strengthens its brand positioning as “responsible AI”; for the Vatican, partnering with a company genuinely engaged in AI safety research elevates the encyclical beyond empty rhetoric.
What did the Pope say in the encyclical?
Magnifica Humanitas was signed on May 15, the 135th anniversary of Pope Leo XIII’s 1891 encyclical Rerum Novarum. That encyclical laid the foundation for Catholic social teaching during the First Industrial Revolution, addressing issues such as workers’ rights and the limits of capitalism. Pope Leo XIV chose this date to sign an encyclical on AI, deliberately drawing a historical parallel between the AI revolution and the Industrial Revolution.
According to CNN, the central claims of the encyclical include: AI systems must be "disarmed" and separated from military and economic interests; some autonomous weapons systems have advanced to the point of being "nearly beyond human control"; delegating "life-or-death decisions" to AI is "unacceptable"; AI could give rise to "new forms of slavery," particularly in developing countries, where children and adolescents mine rare earth elements under dangerous conditions, "ensuring the uninterrupted flow of computation."
Anthropic's Religious Diplomacy: From 15 Christian Leaders to the Vatican
Olah's visit to the Vatican was not an impromptu event, but rather a continuation of Anthropic's systematic engagement with religious institutions over the past several months.
According to The Washington Post, Anthropic invited around 15 Christian leaders to a two-day summit at its San Francisco headquarters in March to discuss how Claude should respond to complex ethical issues such as mourning and risk of self-harm, even debating whether Claude could be considered a “child of God.” Meghan Sullivan, a philosophy professor at the University of Notre Dame, said: “A year ago, I wouldn’t have told you Anthropic was a company concerned with religious ethics. That has changed.”
At the end of April, executives from Anthropic and OpenAI attended the first "Faith-AI Covenant" roundtable in New York, engaging in dialogue with leaders from multiple religious communities, including Hinduism, Sikhism, the Baháʼí Faith, Greek Orthodoxy, and Mormonism. According to the Associated Press, similar roundtables are planned to continue in Beijing, Nairobi, and Abu Dhabi.
On May 19, Anthropic published a statement on its official website titled "Expanding the Conversation on Frontier AI," revealing that it had conducted its first round of discussions with scholars, clergy, philosophers, and ethicists from over 15 religious and intercultural groups.
From San Francisco to New York to the Vatican, this roadmap clearly illustrates how Anthropic has elevated religious diplomacy from a corporate exploration to public collaboration with the world’s highest religious authorities.