Pope Leo's texts confirmed their human authorship by utilizing AI detection tools.

      To counter allegations that artificial intelligence assisted in crafting his messages, an anti-AI Pope turned to an unexpected resource: AI itself. An Australian company has declared that the writings of Pope Leo XIV are entirely human-generated, utilizing a range of detection tools developed for this very technology.

      AI has become a central theme for Pope Leo XIV. His inaugural encyclical, Magnifica Humanitas, released in May, cautioned against the potential for technology to overpower humanity. Recently, a compilation of his thoughts on education and AI, titled Maps of Hope, has been verified as authored by humans.

      The uncomfortable backdrop to this certification involves a contentious incident. Following the release of the encyclical, AI researcher Linch Zhang claimed in a widely circulated critique that some sections bore signs of AI influence. He noted the frequent occurrence of em dashes, the term "genuinely," and rhetorical trios, suggesting that Anthropic’s Claude might be at fault.

      The AI detection tool Pangram flagged certain excerpts. The Verge posed a pertinent question: Did the Pope utilize AI while discussing the hazards of AI? Zhang clarified that it was likely Vatican officials who may have done so, not Leo himself. The Vatican has never verified this assertion, nor does it have conclusive proof.

      This sets the stage for the new certification. Proudly Human, led by Australia’s former chief scientist Alan Finkel, endorsed Maps of Hope as a human-authored work and conferred its trust mark upon it. The book was created by the Australian Catholic University in collaboration with the Vatican’s Dicastery for Culture and Education and includes a foreword by Cardinal José Tolentino de Mendonça.

      Finkel describes the certification as the essence of their initiative. “With the Pope’s focus firmly on the issue of AI and human authorship, this signifies that it’s no longer a niche debate,” he remarked. “It’s the central question of our time.”

      The certification process involves both paperwork and technology. Authors confirm their identity and sign a declaration asserting that AI did not compose the first draft or any part of the text. Proudly Human then tests the text with “a suite of four or five” AI detectors, alongside its own averaging algorithm, ultimately determining that the collection is “essentially human.” Thus, to certify the human aspect, it relies on machine analysis.

      Finkel argues that, in the absence of laws requiring writers to disclose the use of AI, readers have the right to know the source of a text. "I actually think it’s a fundamental right,” he stated. This is particularly crucial for religious leaders, as congregations anticipate authentic wisdom from the individual delivering the sermon.

      This approach reflects a rising demand for confirmation that a person—not a program—completed the writing. However, AI detection remains an uncertain science; tools often disagree and can produce misleading results.

      The Atlantic has cautioned against over-relying on detection tools like Pangram, as this could lead to baseless accusations. Some human authors may be flagged simply because AI has adapted to write in their style. Therefore, a certificate's reliability hinges on the accuracy of the detection tools, which are still imperfect.

      This incident illustrates the ongoing debate surrounding AI. As machine-generated text inundates the internet, the origin of the writing is becoming a matter of trust and commercial significance. Proudly Human markets its certification as the equivalent of an organic or fair-trade label for human writing. The fact that the test case involves the Pope, with an AI detection tool as the adjudicator, only emphasizes this issue further.