Christopher Olah, the co-founder of Anthropic and its head of interpretability research, delivered a striking address at the Vatican on Monday. Seated alongside Pope Leo XIV during the formal presentation of the pontiff's first encyclical, Magnifica humanitas, Olah made a case that no other leader of a major artificial intelligence firm has made at such a prominent platform: the future of frontier AI cannot be determined exclusively by the companies that build it.
Speaking in the Vatican Synod Hall, Olah told the assembled audience—which included cardinals, diplomats, and tech executives—that every frontier AI lab operates within a set of incentives and constraints. These forces can sometimes push researchers away from what is morally right. Even those with the best intentions, he said, remain caught in that system. The conclusion he drew was unmistakable: outside scrutiny from religious institutions, governments, and civil society is not optional but essential.
The address marked a pivotal moment for Anthropic. The company has long positioned itself as a safety-first alternative to other AI leaders, emphasizing interpretability—the effort to understand and reverse-engineer how large language models arrive at their outputs. Olah, who founded Anthropic's interpretability team, is the public face of that work. By speaking from the Vatican stage, he lent moral weight to his argument and sought to anchor his company's message within a broader ethical conversation.
The encyclical itself, Magnifica humanitas, represents the Catholic Church's most consequential statement on technology since Pope Leo XIII's Rerum novarum addressed the rise of industrial capitalism in 1891. That earlier document shaped labor rights and social justice movements for generations. The new encyclical frames AI as a transformative force that demands careful ethical integration into human society. While it does not prescribe specific policies, it establishes a framework of human dignity, solidarity, and the common good that the Church argues should guide AI development.
Olah's speech complemented that framing, but he went further. He warned of a real possibility that AI could displace human work at very large scale. If that happens, he said, supporting those displaced will be a moral imperative of historic proportions. This statement is perhaps the most explicit public acknowledgment by a frontier-lab founder that the technology his company is building may, on its own internal projections, dislodge employment faster than the labor market can reabsorb it.
Anthropic's presence at the Vatican is not a one-off event. Over the previous two weeks, the company had already begun repositioning itself in the public eye. It announced a new office in Milan, signaling a deeper commitment to Europe and to engaging with the Catholic Church. The Vatican event was the culmination of that effort, placing Anthropic at the center of a conversation that blends technology, ethics, and global governance.
The political backdrop to Olah's Vatican appearance stands in stark contrast to its moral tone. During the spring, Anthropic found itself at the center of two separate confrontations with the US government. The Pentagon excluded the company from its top classified AI work in April, citing Anthropic's own usage restrictions on its technology. The Defense Department then signed deals with Nvidia, Microsoft, and Amazon Web Services to fill the gap. Shortly after, the Trump administration blocked an expansion of Mythos, an autonomous vulnerability-discovery model developed by Anthropic that had already shaken cybersecurity practices in the banking sector globally. The model's ability to find software vulnerabilities autonomously had raised concerns about misuse and national security.
Olah's call for outside oversight, delivered from the same stage as the pope, can be read as a direct response to these clashes. He did not shy away from the tensions within his own company. In a candid moment, he told the room that companies like his operate under strong commercial, geopolitical, and personal pressures that can conflict with the broader interests of society. The argument was not that Anthropic stands outside those pressures but that the answer to them lies beyond the walls of any single lab.
The commercial stakes for Anthropic are enormous. The company is reportedly in talks to raise $30 billion at an eye-watering $900 billion valuation. That figure, if realized, would make Anthropic one of the most valuable private companies in the world. The dissonance between its ethical messaging and its massive financial ambitions is sharp. Olah acknowledged this openly, which lent sincerity to his words but also raised questions about how the company's growth trajectory might affect its independence.
Interpretability research, the field Olah leads, is central to Anthropic's safety narrative. The goal is to understand what goes on inside large neural networks—often described as black boxes. By decoding the internal representations and decision-making processes, researchers hope to detect and correct harmful behaviors before they cause real-world damage. Olah's team has made significant progress in visualizing neurons and circuits that correspond to concepts like deception, truthfulness, and bias. However, critics argue that interpretability remains far from delivering the kind of guarantees needed for safe deployment at scale.
The encyclical and Olah's speech both declined to outsource the regulatory architecture of the next decade to the companies that have built the technology over the past few years. Instead, they called for a broader coalition of actors to assume responsibility. Pope Leo XIV has previously signaled interest in technology ethics, meeting with industry leaders and scholars. The encyclical is his attempt to set a moral compass for the age of AI, but it remains to be seen whether it will translate into binding regulation.
Olah's appearance at the Vatican also highlights a growing trend among tech leaders to seek legitimacy from traditional institutions. As public trust in Big Tech erodes, partnerships with organizations like the Catholic Church offer a veneer of credibility and a connection to deeply held values. For the Church, engaging with AI companies provides an opportunity to shape the ethical direction of technologies that will affect billions of people, including many of its 1.3 billion members worldwide.
The broader implications of Olah's speech extend beyond Anthropic. It signals that at least one frontier-lab leader recognizes the insufficiency of self-regulation. The AI industry has faced mounting criticism for moving faster than society can absorb the changes, from ethical lapses to job displacement. Governments around the world are scrambling to design regulations, but progress has been uneven. The EU's AI Act, for example, sets some of the most comprehensive rules, but enforcement remains a challenge. In the US, a patchwork of state and federal initiatives has yet to produce a coherent national framework.
Olah's warning about job displacement is especially timely. Recent studies estimate that generative AI could automate tasks affecting hundreds of millions of jobs globally within the next decade. White-collar professions, including legal, accounting, and creative roles, are particularly vulnerable. While new jobs will likely emerge, the transition period could be painful. Olah's call for a moral imperative to support displaced workers echoes proposals for universal basic income, retraining programs, and expanded social safety nets. But he did not specify which policies Anthropic endorses, leaving the concrete steps to governments and civil society.
The Vatican event itself was meticulously orchestrated. The launch of Magnifica humanitas was timed to coincide with the feast of a significant saint, and the location in the Synod Hall underscored the Church's desire to treat AI as a matter of global deliberation. Olah's participation was announced only days before, adding an element of surprise. The invitation reportedly came directly from the Pope's secretariat after months of quiet dialogue between Anthropic representatives and Vatican officials.
Anthropic's pivot toward European engagement makes strategic sense. The company faces increasing regulatory scrutiny in the United States, while Europe's more proactive stance on AI regulation offers a stable environment for shaping policy. The Milan office will serve as a hub for research and public engagement, and it is likely that Anthropic will hire local experts in ethics and governance. This move also allows the company to distance itself from the more aggressive tactics of some Silicon Valley rivals.
In his address, Olah did not shy away from acknowledging the elephant in the room: his own company's immense valuation. He joked that it might seem hypocritical to warn about the dangers of AI while simultaneously raising billions to build better models. But he insisted that the only way to ensure responsible development is to involve voices that are not financially or politically vested in the technology's success. His audience appeared receptive, though it remains uncertain whether the Church's moral authority will translate into practical leverage over AI's trajectory.
The choice of Olah as messenger was intentional. He is not the CEO of Anthropic, Dario Amodei, but rather the lead of interpretability—the team most associated with safety. By sending Olah, Anthropic signaled that its message about outside oversight is not just a PR exercise but a core part of its research philosophy. Olah's reputation in the academic community is strong; he has published influential papers and maintains ties with interpretability researchers across institutions. His presence lent intellectual credibility to the Vatican event.
For the Catholic Church, the encyclical is a major step into the digital age. Pope Leo XIV has made technology a central theme of his papacy, following in the footsteps of his predecessors who addressed industrialization and the information era. Magnifica humanitas draws on Catholic social teaching to argue that AI must serve the common good, protect human dignity, and never reduce people to mere data points. It calls for international cooperation and binding ethical standards, though it stops short of demanding specific legislation.
Olah's speech inside the launch echoed that language. He spoke about the need for a new social contract between technology creators and the public. He warned that without strong external oversight, the incentives of profit and national competition could lead to outcomes that harm humanity. His most striking line was that every frontier AI lab operates under forces that can conflict with doing the right thing—a rare admission of fallibility from within the industry.
The immediate reaction from other tech leaders was muted. Several companies declined to comment on the Vatican event. Industry analysts noted that while Anthropic's messaging is distinctive, its actions—such as the massive fund-raising and the development of cutting-edge models—align it closely with its competitors. The question that lingers is whether the company will use its growing resources to actually slow down and prioritize safety, or whether the ethical rhetoric will remain separate from business operations.
From a policy perspective, Olah's appearance could influence the ongoing debate about AI regulation. The Vatican carries moral weight, especially in regions with large Catholic populations. Governments may use the encyclical and Olah's speech as reference points when drafting laws. Already, the European Commission's AI office has expressed interest in the Church's perspective. In South America and Africa, where Catholic institutions are influential, the encyclical could shape public opinion and pressure governments to act.
Back in the United States, the reaction from regulators has been cautious. The White House Office of Science and Technology Policy has not issued a statement. However, the AI Safety Institute, a federal body, has recently begun engaging with Anthropic on interpretability research. It is possible that Olah's Vatican speech will strengthen the case for increased federal funding for interpretability studies.
The commercial weight of Anthropic cannot be ignored. A $900 billion valuation implies a belief that Anthropic will not only catch up to OpenAI but surpass it. That kind of speculation puts enormous pressure on the company to deliver market-competitive products. Balancing that pressure with the moral stance Olah took at the Vatican will be a defining challenge for Anthropic's leadership in the years ahead.
For now, the world's attention is on the unprecedented sight of a frontier AI founder sharing a platform with the pope. The message was clear: the technology that will reshape society cannot be guided by its creators alone. Whether that message moves practical policy remains to be seen, but the fact that it was delivered at all is itself a landmark moment in the short history of artificial intelligence.
Source: TNW | Anthropic News