Anthropic and the Vatican Align on AI: Why a Global Pause on Frontier Development Is Urgently Needed

Anthropic's Warning: The Risk of Recursive Self-Improvement

Anthropic's report, authored by researchers Marina Favaro and Jack Clark, presents a stark picture of how quickly AI capabilities are accelerating. The company reported that as of May 2026, Claude authored more than 80% of the code merged into Anthropic's codebase, and the typical engineer now merges roughly eight times as much code per day as in 2024 . The authors warn this trajectory is heading toward recursive self-improvement, a process where AI systems autonomously design, build, and train their own successors without humans driving each step .

The company said it has not yet reached that point, and that the outcome is not inevitable, but then added a sobering caveat: "It could arrive sooner than most institutions are prepared for" .

The Core Concerns

• Loss of human oversight. Anthropic's central concern is that AI progress may outpace existing safety and governance frameworks, making human oversight the primary constraint on technological progress . The company warned that humans risk losing control entirely if development continues unchecked .
• A 'brake pedal' for the industry. Anthropic framed its call for a pause as building a "brake pedal" for the industry—an option to slow down or temporarily halt frontier AI development to allow societal structures and alignment research to keep pace . Clark told BBC Newsnight, "You want the option to be able to take your foot off the gas and put your foot on the brake" .
• Conditional, not unilateral. Anthropic said it would support a pause only if other leading developers joined in and compliance could be independently verified . The company noted that if a slowdown merely enables less cautious entities to catch up, it could ultimately compromise safety for everyone .
• Policymaker engagement. Meanwhile, the company announced plans to convene policymakers to address the hazards of advanced AI .

The Vatican's Magnifica Humanitas: A Moral Framework for AI

Released on May 25, 2026, Magnifica Humanitas runs roughly 43,000 words and is centered on "safeguarding the human person in the age of artificial intelligence" . The document explicitly updates Leo XIII's 1891 encyclical Rerum Novarum—which addressed industrial-era labor—and applies its logic to digital platforms, algorithms, automation, and data concentration .

Key Alignments with Anthropic's Warning

• Call for robust regulation. The Pope urged stringent oversight of AI and warned that Big Tech and its leaders could accumulate power they would abuse for profit . Pope Leo argued that providers of AI must consider the ethics of their actions, declaring, "For AI to respect human dignity and truly serve the common good, it requires strong regulation" .
• Warning against unchecked autonomy. The encyclical cautioned AI companies against building "a new Tower of Babel"—a moral parallel to Anthropic's concern about systems escaping human control . The document compares an AI future that excludes God to the biblical story, warning against human arrogance .
• Human dignity as the standard. Magnifica Humanitas argues AI must serve the common good and human flourishing, not be driven by competitive acceleration alone . The encyclical's central sentence reads: "Artificial intelligence must now be disarmed, liberated from frameworks that transform it into an instrument of control, exclusion, and destruction" .
• Call for a slower pace. Pope Leo wrote that calling for prudence, rigorous evaluation, and even, at times, a slower pace in adopting AI does not mean opposing progress; instead, it is "an exercise of responsible care for the human family" .

Direct Collaboration Between Anthropic and the Vatican

The alignment between the two institutions goes beyond parallel warnings. In a rare show of institutional cooperation, Anthropic co-founder Chris Olah spoke at the presentation of Magnifica Humanitas at the Vatican on May 25, 2026, highlighting a unique relationship between an AI lab and the Holy See . The appearance signalled a rare alliance as both sides called for stronger moral scrutiny of artificial intelligence .

Connections between the two sides had already existed. Anthropic had previously consulted Catholic thinkers while developing the "Claude Constitution," the 23,000-word ethical framework guiding its AI models . Father Brendan McGuire, a priest and former Silicon Valley executive, helped shape that constitution and noted that Magnifica Humanitas has opened doors for deeper conversations between the Church and the tech industry .

Less than two weeks after the encyclical's release, Catholic AI experts publicly commended Anthropic's pause call, noting that it "echoed Pope Leo XIV's recent concerns" and praising the company for checking values against rapid development . Catholic moral theologians had also previously filed an amicus brief backing Anthropic in a Pentagon dispute over the company's refusal to allow its AI systems to be used for mass surveillance or autonomous weapons .

A Shared Thesis and Complementary Strengths

Both Anthropic and the Vatican converge on a central thesis: frontier AI development is accelerating faster than governance and ethics can adapt, and a deliberate pause or slowdown is needed to preserve human oversight, dignity, and the common good. The Church's contribution adds a robust moral and theological framework—rooted in human dignity, the common good, and protection of the vulnerable—that complements Anthropic's technical and safety-oriented reasoning .

As one observer put it, both institutions argue that "society may struggle to keep pace with rapid technological advances" and that a slowdown is "likely a good thing"—provided it is coordinated, verifiable, and not merely a tool for less cautious competitors to gain ground .