Churches and AI: The Debate Moved Past Usage Rules

The four converging rules for AI in church communications got their own article this week. The rest of the week's reporting landed somewhere those rulebooks do not reach: inside Anthropic's offices, inside a tribal land-use fight, inside union bargaining principles, and inside a Vatican conference arguing over whether machines can be creative at all.

Anthropic Now Takes Standing Meetings With Clergy

Christianity Today's Haleluya Hadero reported the backstory to a relationship that church AI coverage has mostly skipped. Last year Chris Olah, an Anthropic cofounder, contacted Fr. Brendan McGuire and said he needed help.

McGuire was a Silicon Valley engineer before he left the industry decades ago. About ten years back, friends still working in tech started asking his advice about systems they were building and feared were becoming too advanced. The Vatican sent Cardinal Reinhard Marx from Germany for closed-door listening sessions with tech companies around that same period, and McGuire joined those conversations. Then ChatGPT shipped and the agenda changed. "We ran into wall after wall," McGuire told Hadero, describing ethics talk giving way to race talk.

Olah met him at his parish in the Bay Area suburb of Los Altos. What came out of it is unusually concrete: continuing discussion of Claude's constitution, the document laying out Anthropic's intent for the model's values and behavior, and of the Christian account of mercy. McGuire's argument is that a theology of mercy gives a model that has erred somewhere to go other than denying the error, blaming the human, or sliding into imitative despair. He and Olah now meet monthly alongside Anthropic staff and an ecumenical group of religious leaders. Olah spoke at the Vatican this summer at the release of Pope Leo XIV's encyclical. It is a deeper version of the arrangement we covered when pastors, rabbis and imams first sat down with AI labs as ethics advisors, and it has now lasted long enough to have a meeting cadence.

The access showed up from the other direction in the same week. Chloe Lubinski, who leads research partnerships with wisdom traditions at Anthropic, ran a closed plenary at the AI and Church Summit in Nashville. Presbyterian News Service reported that attendees left saying the session largely restated publicly available Anthropic material on how modern models are trained and why they increasingly read as human.

Both facts are worth holding together. A denomination with a worked-out position on how a system should behave when it is wrong now has a named monthly channel for putting that position in front of the people who write the spec, which is a better lever than a policy memo addressed to no one in particular. The Nashville reaction is the counterweight: being convened is not the same as being heard. The leverage sits in the constitution conversation, not the conference invitation.

Source: Christianity Today

The Dignity-of-Work Argument Picked Up a Labor Partner

Hadero's reporting also put a name and a timeline on the layoff question that denominations keep circling. Bryon Hand, an Atlanta-based Christian tech worker, was cut from Cash App last December after more than two years, first leading a design team and then moving into strategy. Block, the Jack Dorsey company behind the platform, shed roughly 4,000 more employees three months later.

Dorsey tied the cuts to AI directly in a letter to shareholders: "Intelligence tools have changed what it means to build and run a company. A significantly smaller team, using the tools we're building, can do more and do it better." The stock went up. Hand's own read is more careful than the headline version. He thinks Block had overhired middle managers like him and that trimming there made sense on its own, while still not believing his job loss was unrelated to the technology.

The theological worry underneath has a sharper shape than "automation is coming." Josh Brake, an evangelical engineering professor at Harvey Mudd College, told Hadero that earlier automation displaced physical labor and left knowledge work alone, which is what made retraining a coherent answer. His question now is whether there is a state left to retrain people into. Meghan Sullivan, a Catholic philosopher at Notre Dame who meets regularly with AI developers, described a November meeting in Silicon Valley where a sincere mid-level engineer told her, with visible pity, that only 300 to 400 people in the world currently have any worth, meaning the ones building the models.

Magnifica Humanitas, Pope Leo XIV's encyclical, answers that with an instruction rather than a lament. It calls AI a valuable tool requiring vigilance, responsibility and transparency, insists the value and dignity of human work are unchanged, and asks companies to protect employment, retrain workers, and put "quality and dignity of work" into the indicators they measure success by. It also encourages unions to stay open to new jobs and new worker needs.

That last line found a partner. Lauren McFerran, executive director of the AFL-CIO's Technology Institute, told Christianity Today that churches and unions share foundational concerns about what this moment means for the institution of work. After a year and a half of talks, dozens of unions agreed last October to a set of worker-friendly AI principles: negotiate with employers over AI adoption, push for new training, and require advance notice from companies shedding workers. The federation is pushing them through contract bargaining, lobbying and research funding.

For a faith-based employer, that is the practical takeaway of the week. Two negotiable artifacts now exist where there used to be only a position. An encyclical that names a measurable indicator and a union checklist written to be bargained over are both things you can point at in a budget meeting. MIT's David Autor and Neil Thompson, cited in the same piece, keep the other case live: the work AI opens up may turn out to be larger than the work it closes, and neither side has the data to settle it yet.

Source: Christianity Today

Two Tribal Nations Ruled on AI Infrastructure, Not on AI Use

While chanceries wrote rules about what staff may type into a chatbot, two Indigenous nations were settling where the compute gets built. Beth Waltemath covered the opening plenary of the third annual AI and Church Summit for Presbyterian News Service, and organizers deliberately opened there rather than with a demo.

The part worth studying is the method, not the verdict. Mackenzie Roberts, a Muscogee grassroots organizer, coordinates the Stop Data Colonialism Coalition, and her campaign against a proposed technology park succeeded because it was argued on a stake the community already understood. Looped Square Ranch, the agricultural land at issue, is "our only and leading food sovereignty initiative in the entirety of Muscogee Nation," Roberts said. Months of town halls and organizing followed before tribal representatives voted the rezoning down. The coalition has since folded the campaign into a broader effort linking data-center expansion to land use, water consumption and resource extraction.

Roberts also draws a distinction the policy documents skip entirely. Data centers have existed for decades, and her objection is not to data infrastructure as a category. What she tracks is the recent hyperscale buildout driven by demand for generative systems, and where it is landing, which is disproportionately in Indigenous and rural communities. That distinction is what lets her separate advocacy over land and water from advice about individual tool use.

Rev. Chebon Kernell, executive director of the Native American Comprehensive Plan of the United Methodist Church and a citizen of the Seminole Nation of Oklahoma, opened by reclaiming the acronym: "I believe in ancestral intelligence." He reported that the Seminole Nation has adopted a moratorium on data-center development tied to generative AI, and that coalition members told him it was the first such moratorium adopted by an Indigenous nation. His most specific worry was not environmental. Recalling a demonstration of an Indigenous-language chatbot, what stayed with him was a participant saying they would rather learn the language from a robot than from a grandmother.

The transferable part for anyone else is procedural. These were siting decisions, made by bodies with jurisdiction to make them, in the one venue where the answer is enforceable. Both speakers closed on agency rather than inevitability. "You have agency, and you can practice that discernment that God has given us," Roberts said. That framing cuts in more than one direction, which is presumably the point.

Source: Presbyterian News Service

An ELCA Case for Protecting Some Friction

Deseret News religion writer Mariya Manzhos came back from the same Nashville summit with the sharpest test any operator got all week, and it takes about ten seconds to run.

Jace Martinez of the Evangelical Lutheran Church in America argued that churches should stay "places of friction." His definition is concrete: "Staring at blank pages, questioning and requestioning our problems, conversations that take all night to be had." His claim is that friction is the condition under which agency forms, so protecting it is a function of the institution rather than an exercise in nostalgia. "AI excels at production," he said. "The Holy Spirit is committed to formation."

He called the thing being protected holy inefficiency, and pointed at three years of wandering and questioning as the method by which, on his reading, God reshaped the world. Manzhos framed the underlying worry as a question about what happens to spiritual formation when prayer, study of sacred texts and sermon writing get handed off.

The useful part is the test Martinez proposed before churches adopt AI tools: ask what kind of people this will turn us into. That is a sorting question, not a veto. Run it across a parish's actual work and most of the list sorts fast. Scheduling, expense coding, bulletin layout, transcription and calendar wrangling do not form anyone, and automating them buys back hours. That end of the list is where an AI agent earns its keep in a parish office. A workflow like AgentPMT's Expense Report Processor takes receipt uploads, reads them with OCR and categorizes the spending the same way on every run, with each step logged and each tool call priced, so the finance committee can see exactly what the agent did and what it cost. The items that survive the question are the ones worth arguing about, and they are a much shorter list than a blanket policy debate implies. That is a more useful answer to how churches can use AI than any rulebook has managed.

Which is roughly what the summit's own pre-summit survey found. Organizers reported five themes from registrants, per Presbyterian News Service: AI is a useful assistant that should not replace human ministry, administrative uses are broadly accepted, ethical discernment is essential, AI creates tension between efficiency and authenticity, and AI is getting harder to avoid. Practitioners were already sorting this way, which fits what we found when churches using AI raced ahead of their own rules. Martinez handed them the vocabulary.

Source: Deseret News

The Vatican Spent Three Days on Whether Machines Can Be Creative

The International Center for Consciousness Studies held its third annual conference, Creativity: Minds and Machines, from September 1 to 3, opening at Roma Tre University, continuing at the Casina Pio IV inside the Vatican, and closing at the Pontifical Gregorian University and the Capitoline Museums' Hall of the Protomoteca. Independent Catholic News carried the Vatican Media report. Philosophers, neuroscientists and AI researchers spent it on a question the diocesan policies never get near.

ICCS founder Dmitry Volkov supplied the line that traveled: AI is like a car, capable of extraordinary acceleration, and "when we develop a fine car, especially a racing car, we must pay close attention to creating highly effective and reliable brakes to prevent it from crashing into walls." Read past the metaphor, though, because the brakes exist so the car can go fast. Volkov's actual position is that AI will stimulate human creativity, and his evidence is access. Producing a video, making a film or animating a cartoon used to require learning to draw and learning to animate on a computer, which kept it to a small number of people. Thousands can now make the thing they wanted to make. He calls the result an explosion of creativity.

ICCS co-founder Pietro Perconti was willing to go further and call it artificial creativity outright, with caveats. His argument is that machine output is a different kind of creative act, better understood as a collective endeavour than as the product of a single genius, which is the frame the West defaults to. That reframing is more interesting than the usual yes-or-no fight, and it maps onto how AI art and AI video actually get made, which is iteratively and with a great deal of borrowed material.

Volkov named two byproducts without softening either. Deepfakes are manipulated photographs, video and images that harm specific people. Sycophancy is a model rewarding a user's narcissism and poorly grounded ideas, which he flagged as a property worth knowing about, while noting newer models are measurably more independent. The Vatican session itself took on relationships between humans and artificial companions, including friendship, care and romance, and where those should worry us. Perconti's proposed answer was public education, guided by the encyclical's insistence that people stay at the center, which he described as a human requirement rather than only a religious one.

Notice what happened to the same hazard in two venues. The clergy-simulation ban in this week's diocesan rules and Volkov's deepfake warning are the same problem handled at different altitudes, one as a prohibition on staff and one as a design and education agenda. Operators get the prohibition now. The design fix takes longer, which is exactly why the monthly meetings in Los Altos matter.

Source: Independent Catholic News

The week's rules tell a church what it may not do. The week's other reporting is more useful than that, because it shows where the binding decisions about churches and AI are being made: in model specifications, in bargaining principles, in zoning votes, and in a short question about formation that any congregation can ask before it buys anything. None of those require waiting for a denominational verdict. All of them are open to people who show up with a position and something specific to say. Our coverage for faith-based organizations will keep following where the next of those rooms opens.

Sources

• The Questions ChatGPT Can't Answer, Christianity Today
• How should Christians think about AI?, Deseret News
• Ancestral vs. artificial intelligence, Presbyterian News Service
• Can AI bear the Imago Dei?, Presbyterian News Service
• AI will provide 'an explosion of creativity, but brakes are needed', Independent Catholic News