On a crisp morning in late May 2026, inside the Apostolic Palace at the Vatican, a scene unfolded that would have seemed impossible just a few years ago. Seated beside Pope Leo XIV, in front of an audience of cardinals, diplomats, and religious leaders, was a figure more accustomed to the fluorescent glow of a San Francisco engineering lab: Chris Olah, co‑founder of Anthropic, one of the world’s most important AI companies.

Olah was not there to announce a new model. He was not unveiling a faster chip or a cheaper API. He was there to help launch the first papal encyclical on artificial intelligence – a formal teaching document that carries immense moral weight for the world’s 1.4 billion Catholics.

The setting could not have been more different from a typical tech keynote. There was no stage with flashing LED screens. No product demos. No “one more thing” applause lines. Instead, Olah sat in a gilded hall, beneath Renaissance frescoes, speaking in measured tones about responsibility, fear, and the need for external guidance.

This was not an accident. Olah, and the Vatican, chose this stage deliberately.

Every other major AI announcement in 2026 has come from familiar venues: Google I/O in Mountain View, Microsoft Build in Seattle, OpenAI’s virtual briefings broadcast to developers worldwide. Those events are about features, pricing tiers, and competitive positioning. They are aimed at engineers, investors, and early adopters.

The Vatican event was aimed at everyone else.

By speaking beside the Pope, Olah was sending a clear signal: the conversation about AI cannot be left to Silicon Valley alone. It must involve moral authorities, religious institutions, and the broader public. The stage was not a backdrop – it was the message.

Why the Vatican – The Search for Moral Legitimacy

Chris Olah had many stages to choose from. He could have delivered the same warning at the World Economic Forum in Davos, surrounded by CEOs and hedge fund managers. He could have testified before a US Senate committee, speaking into C‑SPAN cameras. He could have given a TED Talk, a commencement address at MIT, or a keynote at an AI safety conference.

He chose the Vatican instead. The question is: why?

The answer lies in what each stage represents. Davos speaks to power and capital. A Senate hearing speaks to law and regulation. A tech conference speaks to engineers and products. But the Vatican speaks to moral authority – a currency that no government agency or corporate boardroom can mint on its own.

The Catholic Church is not a technical institution. It does not set export controls, enforce antitrust laws, or approve new drugs. But it commands the attention of over a billion people worldwide. Its teachings shape the moral frameworks within which millions of families, businesses, and even policymakers operate. When the Pope speaks, the world listens – not because he can compel obedience, but because he represents a tradition of ethical reasoning that spans two millennia.

Olah understood that framing AI as a purely technical challenge has failed. The public does not trust tech companies to regulate themselves. Governments are moving too slowly, or in conflicting directions. And the voices most often heard in the AI debate – engineers, investors, and futurists – are not necessarily the ones best equipped to answer moral questions.

By stepping into the Vatican, Olah was acknowledging a simple truth: the most urgent questions about AI are not “can we build it?” but “should we build it?” and “who decides?” Those are not engineering questions. They are moral and political questions. And for that conversation, a stage in Rome carries more weight than a stage in San Jose.

The Vatican, for its part, was eager to host him. The Church has been actively building its competence in AI ethics for years, convening experts from industry, academia, and civil society. By inviting Olah – a respected researcher, not a flashy CEO – the Pope signaled that the Church seeks serious engagement, not photo opportunities.

The stage was chosen with care. And its choice changed the nature of the message.

Why the Pope Invited Anthropic – The “Values‑First” AI Lab

The Vatican did not invite just any AI company to sit beside the Pope. It did not invite OpenAI, the creator of ChatGPT and the most recognized name in consumer AI. It did not invite Google DeepMind, whose Gemini models power hundreds of millions of devices. It did not invite Microsoft, whose partnership with OpenAI has made it the largest investor in frontier AI.

It invited Anthropic.

The choice was deliberate. Among the major AI labs, Anthropic has consistently positioned itself as the “values‑first” alternative. Its founding team, including Olah, left OpenAI in 2021 over disagreements about safety and commercialization. The company built its reputation around Constitutional AI – a technique for training models to follow a explicit set of principles rather than simply mimicking human feedback from the internet.

Constitutional AI is not just a technical curiosity. It is a philosophical statement. It says that AI systems can and should be aligned with human values by design, not just by trial and error. For an institution like the Catholic Church, which has spent centuries articulating a moral framework for human behavior, this approach resonates deeply.

Anthropic has also been more willing than its rivals to engage with religious and civil society organizations. Olah has personally participated in dialogues with leaders of more than 15 faith traditions, exploring how different religious frameworks might inform AI safety. The Vatican’s own AI ethics advisory board has drawn on Anthropic’s research.

The contrast with OpenAI and Google is stark. OpenAI’s leadership has been embroiled in governance turmoil and a high‑profile lawsuit with Elon Musk. Google’s AI efforts have faced criticism for rushing products to market before safety concerns are fully addressed. Anthropic, by comparison, has cultivated an image of deliberate, cautious progress – even at the cost of slower growth or lower revenue.

By inviting Olah, the Pope was not endorsing every decision Anthropic has ever made. He was signaling that the Church sees Anthropic as the lab most willing to listen, most committed to transparency, and most aligned with the idea that AI must serve humanity, not the other way around.

The invitation was also a message to the rest of the tech industry. If you want a seat at the table where moral authority is discussed, you need to earn it – not through market cap, but through demonstrated commitment to values beyond profit.

The Message Olah Delivered – Self‑Regulation Is Not Enough

When Chris Olah finally spoke, his words were measured, but their meaning was unmistakable. He was not there to praise the tech industry. He was there to warn that the industry cannot be trusted to govern itself.

Olah began by acknowledging the obvious: the people building frontier AI systems operate inside powerful incentive structures. “Every frontier AI lab,” he said, “is inside a set of incentives and constraints that can sometimes conflict with doing the right thing.” Those pressures include commercial competition (the race to release the next model), geopolitical tensions (the fear that China will pull ahead), and personal ambition (the lure of fame, funding, and influence).

He did not exclude his own company from this assessment. Anthropic, he conceded, is subject to the same forces. Even the most well‑intentioned researchers, he argued, can be unconsciously shaped by the environment in which they work. When your competitors are shipping features, when your investors are demanding growth, when your nation’s rivals are accelerating – the pressure to cut corners is real.

This is why, Olah said, voluntary self‑regulation is not enough. Anthropic has a Responsible Scaling Policy (RSP), a framework that commits the company to take specific safety measures as its models become more powerful. But Olah admitted that even well‑designed internal policies can be weakened over time. In February 2026, Anthropic restructured its RSP, a move that some observers interpreted as a concession to competitive pressure. Whether or not that interpretation is fair, Olah’s point stands: internal commitments are fragile.

What he called for instead was “collective governance” – a system in which external voices, including civil society, religious institutions, and democratic governments, have meaningful oversight over the development of the most powerful AI systems. He did not specify exactly what that would look like, but the implication was clear: the era of tech companies writing their own rules, in private, must end.

The setting amplified the message. Olah was not speaking to a room full of engineers who could be swayed by technical arguments. He was speaking from a moral pulpit, beside the leader of the world’s largest Christian denomination. His words carried weight not because of the data he presented, but because of who was listening and where he stood.

“I think this is a scary moment,” Olah said, referring to the speed of AI development and the uncertainty about where it will lead. He validated the public’s anxiety. He did not dismiss it as irrational fear. And he warned that the displacement of human labor by AI could happen “at very large scale,” calling support for displaced workers “a moral imperative of historic proportions.”

This was not a technical lecture. It was a moral appeal. And by delivering it from the Vatican, Olah ensured that the appeal would be heard far beyond the usual tech audience.

What This Means for AI Governance

Chris Olah’s appearance at the Vatican was not merely a symbolic gesture. It was a practical intervention in a growing global debate: who gets to set the rules for artificial intelligence?

For the past several years, AI governance has been shaped largely by three forces: industry self‑regulation (voluntary safety commitments from companies like Anthropic, OpenAI, and Google), national legislation (the EU AI Act, China’s AI regulations, and piecemeal US efforts), and technical standard‑setting (organizations like NIST and ISO developing benchmarks and testing protocols).

Each of these approaches has proven incomplete. Industry self‑regulation is vulnerable to competitive pressure, as Olah himself acknowledged. National legislation is fragmented, with the EU moving faster than the US and China charting its own course. Technical standards, while useful, do not address the deeper moral questions: What values should AI serve? Who decides when those values conflict? How do we ensure accountability when systems cause harm?

By stepping into the Vatican, Olah was pointing toward a fourth pillar of AI governance: moral and ethical authority from outside the tech industry. The Catholic Church does not write laws, enforce contracts, or certify models. But it does something that no government or corporation can easily replicate: it commands moral attention.

The Church’s involvement legitimizes the idea that AI is not merely a technical or economic matter – it is a human matter, with implications for dignity, justice, and the common good. This framing has practical consequences. When the Pope speaks, bishops listen; when bishops speak, parishes listen; when parishes are informed, congregations become constituents who can pressure elected officials. The Vatican’s voice in AI ethics could indirectly shape public opinion, and through public opinion, legislation.

The event also signals a shift in who tech companies choose to listen to. For years, the only external voices that mattered were investors, customers, and regulators. Olah’s appearance suggests that Anthropic – and perhaps others – are now seeking legitimacy from moral authorities as well. This is not altruism; it is strategic. In an era of declining public trust in tech, an endorsement from the Vatican carries weight.

Other institutions may follow. The World Council of Churches, the Islamic World League, the Hindu Dharma Acharya Sabha – all have begun developing AI ethics frameworks. If tech companies engage seriously with these bodies, the landscape of AI governance could become far more diverse, and far more contested, than the current model of “industry + government” envisions.

The Vatican event was not the beginning of this trend, but it was its highest‑profile expression to date. By seating an AI co‑founder beside the Pope, the Church signaled that it intends to be a permanent player in the AI governance conversation. And by accepting the invitation, Olah signaled that Anthropic is willing to be held accountable not just by markets and laws, but by moral communities as well.

The stage was chosen deliberately. Now the question is: will other tech leaders follow?

Frequently Asked Questions (FAQ)

Q1: Who is Chris Olah?
A: Chris Olah is a co‑founder of Anthropic and one of the world’s leading researchers in “mechanistic interpretability” – the field that tries to understand how large language models actually work inside their black box. He previously worked at Google Brain and OpenAI, and has been focused on AI safety for over a decade.

Q2: What is the encyclical the Vatican released?
A: An encyclical is a formal teaching letter from the Pope, addressed to the global Catholic Church. The document launched at this event was the first papal encyclical on artificial intelligence. It outlines ethical principles for the development and use of AI, drawing on Catholic social teaching.

Q3: Why didn’t the Vatican invite OpenAI or Google?
A: The Vatican chose Anthropic because of its long‑standing commitment to “Constitutional AI” and its willingness to engage with religious and civil society organizations. Anthropic’s co‑founder Chris Olah has participated in dialogues with leaders of more than 15 faith traditions. The invitation was a signal that the Church values safety and values alignment over market size or brand recognition.

Q4: Does this mean the Catholic Church will regulate AI?
A: No. The Church has no legal authority to regulate technology. But it does have moral authority. By issuing teachings and hosting figures like Olah, the Church seeks to influence public opinion and ethical norms, which can in turn shape legislation and corporate behavior.

Q5: Was this just a publicity stunt for Anthropic?
A: The event certainly raised Anthropic’s profile, but it also aligned with the company’s long‑stated values. Olah has warned about the limits of self‑regulation for years, and Anthropic has clashed with the Pentagon over military use of its models. The Vatican appearance was consistent with that record.

Q6: What did Olah mean by “collective governance”?
A: He argued that AI oversight should not be left to tech companies alone. Instead, it should involve external voices – including civil society, religious institutions, democratic governments, and affected communities – in setting rules, auditing systems, and ensuring accountability.

Q7: Will other tech leaders follow Olah’s example?
A: Possibly. As public trust in tech declines, companies may seek legitimacy from moral authorities outside Silicon Valley. Other AI labs may begin engaging with religious bodies, human rights organizations, or other value‑based institutions to build credibility.

Q8: How does this connect to your earlier coverage of AI regulation?
A: Directly. Our previous articles have covered the collapse of Trump’s AI executive order, the EU AI Act, and the limits of self‑regulation. Olah’s Vatican speech is a concrete example of an industry leader acknowledging that those limits are real – and that external voices must have a seat at the table.

Conclusion – The Stage Is the Message

Every speaker chooses a stage. That choice shapes how the message is received. A product launch at a tech conference says: “This is about features and competition.” A Senate hearing says: “This is about law and policy.” A TED Talk says: “This is about ideas and inspiration.”

Chris Olah chose the Vatican. That choice said: “This is about morality and humanity.”

By sitting beside the Pope, Olah acknowledged a truth that the tech industry has been reluctant to accept: the most important decisions about AI are not technical, but moral. Who decides what AI should and should not do? What values guide those decisions? How do we ensure that the most powerful tools ever built serve human flourishing, not just corporate profit or national power?

Those questions cannot be answered by engineers alone. They cannot be answered by investors. They cannot be answered by any single nation or industry. They require a broader conversation – one that includes moral authorities, religious communities, civil society, and the public.

Olah did not provide a blueprint for collective governance. He did not announce a new policy or a new product. But he did something perhaps more important: he legitimized the idea that the conversation about AI must expand beyond its current borders. He invited the world into a discussion that has, for too long, been held behind closed doors in Silicon Valley.

The stage was not incidental. It was the message. And the message is this: the future of AI is too important to be left to the people building it.

The post Why Anthropic’s Co-Founder Spoke at the Vatican, Not a Tech Conference appeared first on Explain This Tech.