When the Church falls victim to artificial intelligence

In January 2024, in a long speech addressed to the media and communicators, Pope Francis expressed a number of concerns regarding artificial intelligence, targeting in particular the problem of “deepfakes,” a technology that “distorts our relationship with others and with reality.”

“I've been the object of it myself,” he noted.

In fact, a year earlier, a doctored photo of Francis in a long white jacket, inspired by the Balenciaga fashion label, had been one of the first media successes in AI image generation. The photo was shared several hundred thousand times.

We now know that it was created by a certain Pablo Xavier, 31, after the Chicago resident ingested hallucinogenic mushrooms. The individual claimed to have no intention of harming the Pontiff, but merely wished to deceive and entertain users of an Internet forum.

Creating fake images of the Pope even became a trend for several weeks, prompting the New York Times, in an article published on April 8, 2023, to question why the Pope had become the “star of A.I.-generated photos.”

The verdict: It was more a sign of appreciation for the Pope, coupled with a desire to create a form of comic incongruity with his well-known status, than a desire to harm him.

But since then, more problematic misappropriations of Pope Francis' image have not been slow in coming.

This is the case, for example, when a false image is taken as real information, for example to overestimate the presence of young people at the World Youth Day in Lisbon. And the problem can also have a political dimension, as when a photo was circulated showing the Pope with a rainbow flag on his shoulders, to give the impression that he was celebrating “Pride Month,” an event promoting the LGBT cause.

These hijackings don't just affect the Pope. Last May, Joan-Enric Vives, bishop of the diocese of Urgell in Spain and Andorra, denounced identity theft after discovering a video posted on Instagram. In this clip, generated by artificial intelligence, he was seen inviting people to buy shares in an oil company. For a time, the diocese considered legal action.

In January, Cardinal Carlos Aguiar Retes, archbishop of Mexico City, also sounded the alarm following the broadcast of a video in which he seemingly claimed to have been cured by a miraculous drug. The archdiocese of the Mexican capital immediately warned the population against this deception.

But artificial intelligence is not limited to image generation. It also concerns the creation of text, and on this point the experience of the first “artificial intelligence priest” is particularly enlightening. Launched last April by Catholic Answers, a non-profit dependent on the Diocese of San Diego in the USA, “Father Justin” was an interface offering to engage users in conversation as if a priest were answering questions.

Sometimes presenting itself as a priest living in Assisi, Italy, this bot sporting a Roman collar was quickly defrocked after inappropriate responses. For example, it authorized a user to baptize his child with an energy drink, and gave absolution to another who confessed his sins.

After receiving numerous complaints, “Father Justin” finally became plain Justin, a “lay theologian” now sporting a simple suit jacket. Catholic Answers was keen to warn its users: its AI is strictly for educational purposes and is “not a substitute for real human interaction.”