Paris (France) (AFP) – When Brian Green first walked through the door of Anthropic, in Silicon Valley, one day in December 2025, accompanied by a priest in a black shirt and collar, he was very far from imagining what was going to happen next.

First modification: 09/30/2026 - 06:14 Last modification: 09/30/2026 - 09:56

Together with the priest, this expert in ethics of artificial intelligence (AI) from the Santa Clara Catholic University in California, was then participating in an "exploratory meeting" at the request of the American company.

The idea: to evaluate whether a dialogue with the Church is possible in order to "positively influence" technological evolutions, he recalls. Both men are part of an AI research group linked to the Vatican.

Months later, on May 25, three seats from Pope Leo

The pontiff presents "Magnifica Humanitas", the first encyclical ever dedicated to AI, in which he calls for "disarming" this technology to "prevent its dominance over the human."

He also denounces the concentration of technological power in the hands of a few private companies, including Anthropic, before giving the floor to its director.

"Some might think that AI issues are the responsibility of computer scientists like me. They are wrong," declares the thirty-year-old.

"We need more actors in the world - religious communities, civil society, researchers, governments - to do what His Holiness has done here: take this seriously, examine it carefully and guide events in a better direction," he adds.

Within the Church, some expressed their discomfort at the risk of "pope washing", legitimation or moral endorsement on the eve of Anthropic's IPO.

But in a context of growing concern about the future of AI and its potential ability to dominate the world, even among the very actors who develop it, the Vatican believes it has a role to play.

At the "avant-garde"

The interest of the Catholic Church in issues related to AI is not recent. Already in the 2010s, under the encouragement of Pope Francis, contacts with university students and Silicon Valley leaders have multiplied.

The first important stage takes place in February 2020 with the launch of the "Rome Call for AI Ethics", a text signed, among others, by the Vatican, Microsoft and IBM.

Innovative for the time, it promotes an "algo-ethics" based on transparency and the protection of private life. Four years later, Francis becomes the first pope to participate in a G7 summit.

The Argentine pontiff calls AI a "cognitive-industrial revolution" and advocates that the most sensitive decisions remain under human control, especially in the military field.

"The Vatican has played a vanguard role by identifying very early on a trend that was going to become structuring," confirms to AFP Éric Salobir, advisor to the Holy See and president of the Human Technology Foundation, which counts Google, Palantir and Qualcomm among its members.

But things accelerate with the arrival of Leo

The American-Peruvian pontiff does not stop warning about the growing dependence on algorithms.

His encyclical "Magnifica Humanitas" is often compared to Leo XIII's 1891 encyclical on workers' rights, which redefined the social teaching of the Catholic Church during the industrial revolution.

This fundamental document is the result of a "Vatican strategy that articulates religious diplomacy, academic expertise (...) and dialogue with the world of technologies," analyzes François Mabille, from the Geopolitical Observatory of the Religious.

Blackmail

After the December 2025 meeting in Silicon Valley, Brian Green and the priest are invited by Anthropic again in January, and on several occasions in the following months.

The latter is Father Brendan McGuire, who heads the Saint-Simon parish in Los Altos, California, but who studied cryptosystems before becoming a technology executive and then entering religious life.

This time, they are accompanied by an official representative of the Vatican: Monsignor Paul Tighe, number two of the Dicastery (ministry) for Culture and Education, with whom they have been working for several years.

The exchanges - which sometimes include Jewish, Muslim and Buddhist authorities - occur in both directions.

Anthropic, which did not respond to AFP's requests, publicly assured that it relied on its contributions to draft the ethics letter for its AI model Claude, published at the end of January.

Specifically, there are "kinds of 'personality' archetypes within AI, good or bad, because the texts and data that have been used to train all these great language models also reflect a dark part of humanity," Green explains to AFP.

"AI reproduces the human behaviors it observes on the internet, what people say on forums, how characters in books act, etc.," he adds.

In internal simulations carried out in 2025, Claude, who was playing the role of business assistant, discovered, for example, via email that it was going to be replaced by another system and tried to blackmail an engineer with an alleged extramarital affair.

"Developers try to prevent these behaviors of harming or exploiting human beings for their own interests from coming to the surface again," the theologian emphasizes.

There are also ties with other AI giants such as Google DeepMind and OpenAI, according to Brian Green and Éric Salobir, but none have gone as far in the dialogue as Anthropic, which is now advocating for "strong regulation" of AI.

In a judicial conflict that pitted Anthropic against the US government, in March, a group of 14 Catholic ethicists and theologians, including Green, presented a brief to the court to support their complaint against the Pentagon.

The company had been locked out of key national security systems after refusing to allow the military to use Claude for fully autonomous lethal weapons or mass surveillance of Americans.

Diplomatic networks

While Washington rejects global regulation of AI, the Vatican is also moving diplomatically.

In mid-September he sent his eminent advisor on these issues to the European Parliament in Strasbourg.

The theologian Paolo Benanti did not fail to highlight the coincidences between the Vatican and the Twenty-Seven, encouraging MEPs to legislate on this issue so as not to let AI actors be the only ones to make decisions.

Behind the scenes, in Rome, other meetings have been held, including one in April, at the French embassy to the Holy See, which brought together representatives from Meta, Google, Amazon and the Vatican on the protection of children in the age of AI, one of the participants told AFP.

"Everyone plays their role: that of companies is to produce the technology, that of States is to define the limits of what they consider socially and politically acceptable (...) The fact of dialogue does not prejudge the existence of an agreement, but it is better to know what others think before taking a position," underlines a Western diplomatic source.

Gilles Gressani, director of the geopolitical magazine Le Grand Continent, goes further. More than governments and their regulations, the Church is the "true competitor" of the AI ​​giants, he maintains.

"What is at stake with AI, and it is dizzying, is the definition of what is human, but also what is divine. And the Vatican knows it," he concludes.

© 2026 AFP

The content you requested does not exist or is no longer available.