Vatican Request for Information on the State of AI

Just two days before Christmas Eve, Transformer released an opinion piece entitled: “How the Catholic Church thinks about superintelligence.” It was written by Paolo Benanti, a Franciscan friar who served as an AI advisor to Pope Francis. Many in the AI safety community have been impatiently waiting for the Vatican to speak on frontier AI topics to see where the Holy Father stands on regulation and to what extent He addresses the transformative impacts of AI or even existential risks. Pope’s approval holds significant moral capital and therefore, offers potential bargaining power vis-à-vis national governments on AI regulation. While Benanti’s piece is an insider’s view, we’re still all waiting for an official, Holy See-stamped document.

Pope Leo XIV has been rising as a spokesperson of moral AI. He seems to never really ignore concerns from different interest groups (e.g. developers, kids’ safety advocates) but is quite selective in allocating His approval or time to official visits. It’s been interesting to watch the growing number of secular experts engaging publicly with the Pope’s AI positions (especially on X) and various advocates attempting to permeate Vatican structures, seeking approval of their agendas. For them, the stakes seem high because Pope’s discernment makes His judgement highly valued and of limited supply.

However, despite many public statements on AI, there isn’t yet a deeply fleshed out papal address on current frontier AI capabilities and their potential impacts. I think this is partly because the Church faces a significant challenge of balancing its imperative to speak eternal truths while responding to one of the fastest-evolving technologies humanity has seen.

There’s a fundamental dissonance between Vatican timelines and AI timelines. The former operates on a long-term deliberative scale, while the latter forces unprecedented changes on our lives in a matter of months. Interestingly, in November 2020, Pope Francis proved He was actually ahead of the AI boom by declaring the monthly prayer intention as: Artificial intelligence is at the heart of the epochal change we are experiencing. Robotics can make a better world possible if it is joined to the common good. Indeed, if technological progress increases inequalities, it is not true progress. Future advances should be oriented towards respecting the dignity of the person and of Creation. Let us pray that the progress of robotics and artificial intelligence may always serve humankind… we could say, may it “be human.”

Since 2020, the state of AI has changed dramatically – its capabilities and adoption have been rapidly increasing, making it an almost ubiquitous tool affecting our daily lives. If the Vatican wants to be at the forefront of the AI conversation now, it must first address two key challenges: unprecedented timelines and information asymmetry. I don’t think these challenges diminish the need for Church guidance. If anything, they make it more urgent, but they do require something different from the Vatican’s usual approach. Here, I especially mean a need for effective evidence-gathering mechanisms that are faster and broader in scope than the usual closed events/consultations timelines.

Challenges

Unprecedented timelines

In November 2022, OpenAI launched ChatGPT. In October 2025, it had approximately 800 million weekly active users. Overall, AI capabilities progress has sped up, growing almost twice as fast over 2024-2025 as it did over 2022-2023, with a 90% acceleration in April 2024.

In comparison, the Vatican has a track record of rather slower procedures, ones optimized for seeking truth and depth of analysis, rather than speed. For example, writing encyclicals takes time: John Paul II wrote 14 of them in his 27 years of papacy, Benedict XVI produced 3 in 8 years, and Francis 4 in 12 years. Relatedly, while the Manhattan Project took place in the early 1940s, the encyclical on the topic was only published in 1963 by Pope John XXIII (“Pacem in Terris”).

The Vatican prioritises deliberation in its teaching because its words carry weight: these are not documents that can be easily revoked by the next “administration” because there is only one administration – that of God. This responsibility requires a more careful approach and avoiding hasty judgements. The goal is to offer humanity a zoomed out perspective and a moral compass, not add another opinion to the pool of many. To do that in the context of AI, the Holy See should find ways to leverage the top experts to provide the Pope with the best available evidence on AI trajectories and risks, and support His informed discernment.

Information asymmetry

However, getting the most recent updates on AI capabilities and their potential effects is hard in the first place. Most likely, only a few hundred people globally have genuine insights into the frontier AI development landscape. This expertise is concentrated in private companies with financial incentives to control the narrative as they see fit. By the time models or information become public, they are in fact outdated in comparison to the real state of AI capabilities. The Vatican, like many other world leaders, has no independent AI capability assessment mechanism and faces the challenge of information asymmetry. Therefore, it needs novel, more appropriate evidence-gathering tools to pool the top AI expertise and provide support to the papal moral leadership.

To get more practical: when the Pope starts working on a potential encyclical related to AI (and He already may have actually started), one of the mechanisms to draw on could be a kind of Extraordinary Consultation on AI, that is, a systematic and ongoing expert Request for Information (RFI) on the state of AI.

There are a few factors that set this RFI apart from typical Vatican consultations:

• Scope and inclusivity: it would be broader and more systematic. An open RFI creates pathways for voices that might go unheard if they cannot travel to Rome or secure a meeting, such as communities in vulnerable contexts experiencing AI’s impacts differently to the developers’ predictions, former frontier lab employees with insider knowledge, domain experts (e.g. in education or healthcare) witnessing AI’s effects firsthand, and the many Catholics worldwide who have relevant expertise but no formal channel to contribute.
• Ongoing and adaptive: rather than a one-off consultation which could become outdated by the time the Vatican releases the final output, this would be a living resource in the run-up to Pope’s release of an encyclical or other key writing.
• Public accountability: making the consultation public creates crucial pressure dynamics. While we cannot rely on the good will of frontier AI labs to submit honest information, there is considerable symbolic cost of ignoring a papal request, especially for companies publicly claiming commitments to “responsible AI” or “AI for good.” The reputational cost of non-participation, or of being seen to provide misleading information to the Holy See, creates incentives for more honest engagement than might occur in private briefings.

Of course, the RFI may not solve the information asymmetry problem entirely, as companies will still likely hold back proprietary information, and some will attempt to shape the narrative to their advantage. But it would nevertheless represent a significant improvement over the current situation, where the Vatican’s understanding of AI risks lagging behind the rapidly moving frontier.

Lastly, I wanted to highlight that as I’m writing this, what’s registered as the most recent development on Wikipedia’s “Timeline of artificial intelligence” is: “Pope Leo XIV calls technologists to build AI machines that embody love, justice, and the sacred dignity of every human life at their core.” I wonder how we will respond to that call…

Special thanks to Elsie and Vesa for valuable feedback and accountability.