tl;dr: The pope calls for AI regulation, takes a stance against transhumanism (with some nuance), denies AI personhood (at least for current models), opposes entrusting lethal decisions to AI in war, and writes about the effects of AI on work. Superintelligence or AGI (artificial general intelligence) are not discussed and there is no mention of bioweapons. The encyclical’s framing comes from Catholic social thought. A technocratic paradigm of AI development motivated by power, efficiency, etc. at the expense of human dignity is an overarching concern.
This blog post examines Pope Leo XIV’s recent encyclical Magnifica humanitas, specifically from a Christian EA perspective (focusing on AI safety).
The bulk of this post consists of summaries of what the encyclical says about various topics like AI regulation, digital minds, subsidiarity, work, family, war, transhumanism, as well as what it doesn’t discuss: superintelligence, bioweapons, directly religious use of AI.
These summaries are not exhaustive: there may be additional themes related to the topics in Magnifica humanitas that I do not mention in the post. Feel free to only read the parts that interest you.
AI use note for the EA Forum version: I used Claude and ChatGPT for background research like asking where a certain topic is discussed in the encyclical and to get feedback on my draft. In maybe 2–5 places, I adopted wording that's close to the LLMs' suggestions.
Some key takeaways
- The encyclical mainly examines AI from the perspective of Catholic Social Doctrine and this heavily influences the things the pope is paying attention to. The encyclical also discusses much more than just AI, for example digital technology and war.
- The pope says it is “imperative to establish a shared framework — also at the international level — in order to curb the technological arms race”.
- Magnifica humanitas focuses on the types of harms that the current AIs are causing. This includes a wide range from effects on employment to distortion of human relationships and blurring the boundary between fact and fiction.
- The encyclical takes a stance against AI personhood. Lack of inner growth through experience and embodiment are central to the argument.
- The pope mostly does not discuss AI as an agent. The concerns are focused on what humans do with AI and AI’s societal effects.
- Relatedly, discussion of AGI and superintelligence is missing.
- The pope seems very worried about the concentration of power AI brings and highlights the need for dialogue, participation, and equity. However, there is no discussion of AI-powered stable totalitarianism in the lock-in sense.
- The tone is not anti-technology. The pope doesn’t consider AI intrinsically evil, but he also thinks the technology is never neutral.
What Magnifica humanitas is
One way to summarise the basic point of the encyclical is that we can use AI to either build a “holy city of coexistence and peace” or a modern Tower of Babel inimical to human flourishing. The pope calls Catholics and other Christians, as well as “all people of goodwill” to “not be afraid to get our hands dirty on the ‘construction site’ of our time.”
While there is overlap with the EA and EA adjacent AI worries, the pope is coming from a genuinely different perspective that only partially matches the concerns of the AI safety movement. He is especially concerned about the spiritual-moral level and the paradigms that cover the way people think about AI. Relationships, concentration of power, and the social and ideological (or spiritual) paradigms that govern AI development and use are primary points of interest in the document.
It goes without saying that this is a huge and dense document, so even an overview of what it says about some relevant topics is bound to be on the longer side. I’ll give a general intro and then focus on what the pope says about particular topics of interest.
This is a high-level official Catholic Church document, so its shelf-life is meant to be hundreds of years. The encyclical is laying groundwork for Catholic AI engagement, potentially for centuries to come. This likely causes reluctance to talk much about the details of the technology and its future developments – with a rapidly developing technology current details might be obsolete and future predictions could look embarrassing in just a few years, let alone decades or centuries.
The pope has a good grasp of some of the fundamentals of current AI development. He recognises the breakneck speed of progress (“any statement regarding AI risks becoming quickly outdated, given the remarkable pace at which these systems are developing”) and points out the black box nature of modern AI systems that results from them being trained instead of coded line by line.
Magnifica humanitas is a very theological document with lots of space given to explaining Catholic social thought, its history, and the connection with AI. Chapters 1 and 2 are meant to demonstrate that the encyclical is in line with the tradition of Catholic social teaching and they introduce it basic principles (the common good, the universal destination of goods, subsidiarity, solidarity, and social justice).
The more concrete AI-specific content mostly comes in chapters three to five. The rest of the post will mostly deal with material from these chapters.
The introductory part features a pair of biblical images that is sort of a structuring element the document keeps returning to: building the tower of Babel (Genesis 11:1–9) and the building of the walls of Jerusalem led by Nehemiah (Nehemiah 2–6). The tower of Babel represents a path of concentration of power led by few, and AI development driven by a technocratic paradigm motivated by power and efficiency at the expense of human dignity. Power concentration in various forms comes up repeatedly in the encyclical. Building the walls of Jerusalem is an effort characterised by subsidiary, a plurality of voices and visions, dialogue, and building for the common good. It represents the kind of approach that is in line with Catholic social thought, the kind of vision of using AI that’s conducive to human flourishing as understood in the Catholic tradition.
To close off the intro section, I want to mention that reliable AI detection tool Pangram flags parts of the encyclical’s text as AI generated, especially in the original Italian. Of course, in a document like this, it is the content that matters and the text has likely been checked several times by the pope and his team, so we can be confident they stand behind the thoughts expressed in the published text. Still, using AI to generate an encyclical on AI does feel a bit ironic.
What the encyclical says about various topics
Regulation
The pope calls for “adequate regulatory tools capable of upholding justice and curbing the distorting effects of technological power”, though he notes that “the issue is not limited to regulation”. He notes that rapid technological development may outpace the development of norms and institutions. The most interesting regulatory point in Magnifica humanitas from an AI Safety perspective is probably a short call to rein in the technological arms race in paragraph 200.
It is imperative to establish a shared framework — also at the international level — in order to curb the technological arms race and ensure robust protection for civilians and the infrastructures necessary for their survival.
Pope Leo’s predecessor Francis also called for “a binding international treaty that regulates the development and use of artificial intelligence in its many forms”, though not specifically in an arms race context. There is not much detail in either pope’s proposals, but the significant part is a call for international regulation. The Vatican has been involved in nuclear disarmament, so playing a diplomatic role in promoting AI treaties could be a possibility.
The pope also speaks of “[c]alling for prudence, rigorous evaluation and even, at times, a slower pace in adopting AI”. I find the mention of evaluations particularly interesting.
Subsidiarity vs a technocratic paradigm
There is a bundle of closely related themes running through the encyclical, connected with the Nehemiah and Babel visions. The document prominently discusses risks related to power concentration and homogeneity, as opposed to the pluralistic and subsidiarity. The spirit of Babel (my words, not Pope Leo’s) is apparent in “the pervasive technocratic paradigm in which we are immersed, and that is amplified by the digital revolution and AI, threatens to normalize an anti-human vision” where “the fullness of life is equated with having more, reducing weakness, eliminating uncertainty and exerting total control”. The pope points out that AI tends to amplify power concentration. There are also several calls to benefit sharing throughout the document. That AI should not only serve the interests of the few and a concern about people being left out is mentioned many times. The pope acknowledges that the labs are leading the development (as opposed to states) and that this presents challenges in terms of participation and power concentration.
Digital minds and the uniqueness of the human intellect
The pope appears to take a firm stance against AI personhood and for the uniqueness of humans. He writes that AI merely imitates certain functions of human intelligence but that AIs don’t grasp the ultimate meaning of situations or the text they produce. The main point is anthropological and ontological: by “intelligence”, the pope means something else than raw cognitive capability. He points out that AIs don’t have bodies and lack formative human experiences like work, love, and responsibility. They can create illusions of a relationship but “lack the affective, relational and spiritual perspective through which human beings grow in wisdom.” In the pope’s definition, intelligence includes embodiment and personal growth through experiences, which current AIs indeed lack.
The pope clearly denies the moral agency of current AIs: they don’t “have a moral conscience, since they do not judge good and evil, grasp the ultimate meaning of situations, or bear responsibility for consequences.” He doesn’t directly address the separate question of moral patienthood, but denying that AIs can form relationships or undergo experiences heavily points to this direction. The overall tone of the encyclical is predisposed against AI personhood.
However, my reading is that even though the document is inclined against the possibility of AI personhood, the possibility that some future AI systems could be intelligent in the sense the pope is talking about is not entirely ruled out. The encyclical does not answer the question of whether an embodied AI capable of gaining experience through interacting with humans or other AIs could be “truly intelligent”.
The environment
Paragraph 101 discusses the environmental effects of AI and calls for developing more sustainable technological solutions to reduce the environmental impact of AI. These concerns are commonly considered as blown out of proportion in EA-adjacent circles (Andy Masley is the semi-official AI water guy).
Epistemics
The pope is also worried about the effects of AI-generated content and digital media on truthseeking – things like AI disinformation and blurring the line between fact and fiction, and how this threatens democracy, plus negative effects on education. He also criticises “a culture of immediacy and hyper-stimulation, which gives rise to fatigue, boredom and apathy concerning the effort required for seeking the truth.”
AI values
The pope is very clear that AI is not morally neutral and that we must “examine how that system is designed and what vision of the human person and society is embedded in the data and models that guide it.” The call for plurality and participation is extended to the values AI systems are imbued with. The pope thinks “merely calling for the moralization of machines — so-called ‘alignment’ of AI with human values” is not enough if these moral frameworks are not discussed and subjected to shared standards. This would mean that the few who control AI will impose their vision on others.
Somewhat relatedly, there’s several paragraphs worth of discussion related to algorithmic bias (102–104) and the problem of externalising responsibility to AI systems with a “veneer of neutrality and objectivity”.
Transhumanism
The pope takes a stand against transhumanism and posthumanism, though with some room for nuance. He recognises that these are not unified ideologies, but rather an ideological archipelago connected by a shared assumption of “the central role of technology and the aspiration to transcend the limits of the human condition”. He sees these ideologies as “present[ing] in some centers of technological power” – I imagine he’s thinking about their prevalence in Silicon Valley and among people working in frontier AI.
The pope sees value in human limitations – they enable growth – and is worried about ideologies that see them merely as an obstacle to be removed. In paragraph 12 he writes that building for the common good means accepting the limits and weakness of humanity without considering them an error to be corrected.
Work
Labour questions are an integral part of Catholic social thought, so labour-related questions are prominent in Magnifica Humanitas. The encyclical calls for several policies to help people deal with job market effects of AI such as retraining and supporting career transitions.
Labour concerns combine with another integral concern of Catholic social thought, concern for the vulnerable. Magnifica Humanitas talks about people working in data labeling, model training (the pope is probably thinking of RLHF feedbackers here) and content moderators, as well as people doing physical work to provide resources for chip construction. This is very much in line with the encyclical’s general left-leaning AI-ethics-style tone.
Whether the encyclical truly grapples with a transformation of the economy created by AI is somewhat debatable: the pope does acknowledge the possibility of mass unemployment, but solutions such as retraining don’t seem to be applicable to a future where most human work is obsolete.
Family
Since working on the CFI AI risk report about a year ago, I’ve thought of AI effects on the family as a Catholic specialty, though in 2026 there’s been things happening in more protestant-coded US conservative space. The Catholic association came from the fact that at the time, the best treatment of this question was a Catholic book, Encountering Artificial Intelligence. The book discussed a wide range of risks like the effects of retreating to AI-generated virtual worlds, moral de-skilling, and AI’s impacts on intergenerational care. Magnifica humanitas focuses on the effects of family via labour, likely due to the Catholic social teaching framing, so the treatment feels narrow.
War
The pope observes that AI accelerates hybrid warfare like cyberattacks and influence campaigns and notes that defensive technologies can easily be repurposed for offense.
AI may lower the threshold to use force and may distance people from responsibility in war.
The pope is concerned about abdicating moral responsibility to AI. Moral judgement involves “conscience, personal responsibility and the recognition of the other as a person”, so it’s not permissible to “entrust lethal or otherwise irreversible decisions” to AI. The Catholic Church has an established position of condemning lethal autonomous weapons, so this isn’t new. Opposition to LAWs is also an established point of connection with the AI Safety space (think of Slaughterbots.)
Nuclear weapons are discussed, but not much in relation to AI.
The pope devotes a paragraph to “disarming” AI, by which means “freeing it from the mentality of ‘armed’ competition, which today is not limited simply to the military context, but is also an economic and cognitive phenomenon”. This can be read as supporting moving away from the current race dynamics, and I think a message like this would be very much in line with the general cooperative common-good shape of the pope’s vision.
What is not discussed
AGI and superintelligence
The terms AGI or superintelligence don’t appear in the encyclical.One reason may be that the pope doesn’t want to discuss technologies that are currently hypothetical. But superhuman intelligence is something that AI companies are explicitly working towards, so I would have expected the encyclical to at least mention the topic. The race towards AGI/ASI also seems like a prime example of a small group of people trying to force something on the rest of humanity in pursuit of their own vision. Perhaps the paragraph dismissing AI as not truly intelligent was meant to preclude these concerns, but it doesn’t read like it was aimed at this.
Biological weapons or other AI-enabled WMDs
Biological weapons created with the help of AI are not mentioned despite this being a primary concern and seemingly the kind of thing the Vatican has spoken about in the past. This, too, may be explained by an unwillingness to speculate, but AI-enabled bioweapons are a much more mundane threat than AI takeover or superintelligence making human labour obsolete so I am somewhat surprised bioweapons aren’t discussed. This seems like a missed opportunity to draw red lines and urge their adoption. Re-iterating the stance of the Catholic church on this issue could have sent a strong moral signal to the international community.
Religious use of AI
The encyclical talks a lot about the moral questions associated with AI use, but is interestingly silent about explicit religious use of AI. This is a real and very likely growing phenomenon. People in EA circles might be familiar with spiralism and the related phenomenon of AI parasitism, but AI is also used in more traditional religious settings, though not necessarily in any less disturbing ways. You can, for example, have a video call with an AI Jesus for $1.99/minute. There is also more neutral religious use of AI, like various Christian chatbots (including a Catholic one trained on 30,000+ sacred works), the use of AI to help in translating spiritual content, etc. The pope told priests not to use AI for sermon preparation, though.
The lack of discussion on direct religious use of AI is probably a result of the Catholic social doctrine framing of the encyclical. This area of doctrine mainly deals with the social, economic, political, and cultural aspects of human life.
My concluding thoughts
The encyclical is a dense, deeply theological document that’s hard to do justice in a blog review. As a non-Catholic Christian in the EA community who’s deeply interested in the risks of AI, it reads as both insightful and disappointing. I’m happy that the pope is speaking about this subject. The Catholic church has generally been the most proactive Christian denomination in engaging with AI. Leo XIV appears quite well informed on the topic.
The document is deeply structured by Catholic social thought. This framing directs attention to questions about justice, participation, work, etc. The encyclical has significant overlap with typical “AI ethics” concerns, but it’s fundamentally a Catholic church document which heavily affects the framing; even if I’m in the broader intended audience of “all Christians” and (hopefully) “all men … of goodwill”, this document isn’t written for me or the AI safety movement.
Still, I find the lack of discussion on AGI/ASI a significant omission. The Tower of Babel image would have been a great opportunity to reflect on the risks of speeding ahead. The dismissal of AI as not being “true intelligence” is a related point that bothers me. I can appreciate it as a spiritual-anthropological point, as AI is indeed very different from the human mind and I’m fine with the theological claim that it doesn’t bear the image of God. But the encyclical comes so close to equating this with a lack of capability that it’s easy to come away from it with an impression that AI can never reach human-level cognitive capacity (I’m still unsure whether this is the intended reading). This dismisses a large and very important class of AI-related risks.
To recap, the encyclical overlaps with AI safety priorities in the call for AI regulation, concerns about power concentration and lethal autonomous weapons, but diverges dramatically in the lack of discussion on ASI/AGI and catastrophic AI risk, particularly bioweapons.
