Anthropic co-founder Chris Olah spoke at the Vatican after Pope Leo XIV released a new encyclical focused on artificial intelligence, positioning the company as part of a broader debate over the technology’s social and moral effects.

The encyclical, titled Magnifica humanitas: On safeguarding the human person in the time of artificial intelligence, was presented in Vatican City on May 25, 2026. Anthropic said Olah was invited to offer remarks as part of its effort to broaden discussion about the questions AI raises beyond the technology industry.

In his speech, Olah argued that companies developing frontier AI systems operate under pressures that can conflict with public-interest goals. He pointed to commercial demands, competition at the research frontier, geopolitical concerns and personal ambition as incentives that can pull labs away from safety-focused decisions. That, he said, makes outside scrutiny essential.

Olah said people who are not bound by those incentives, including religious leaders, civil society groups and other external critics, can play an important role in pressing AI developers to act responsibly. He praised the Vatican’s engagement and said the church could help keep attention focused on safety and human dignity.

He also said the questions raised by AI go well beyond computer science. According to Olah, AI models are not built in the same way as airplanes or bridges, because researchers do not fully understand every part of how they work. He described modern systems as being trained from large amounts of human language and knowledge, making them more complex and less predictable than many people assume.

Olah said the challenge is not only technical, but also philosophical and social. He argued that questions about what kind of AI should exist, how it should behave and how it should fit into society are matters for religion, ethics, philosophy and the humanities, not just engineering.

He highlighted three areas where he believes the church and other outside voices are especially needed. First, he said AI could displace labor on a large scale, creating a moral responsibility to support workers and ensure that the benefits of the technology are not concentrated only in wealthy countries. He said the global distribution of AI gains remains an unresolved problem.

Second, he called for more attention to human flourishing, including the effects AI may have on families, work and children’s development. He said these are questions that cannot be answered by laboratories alone and require long-standing moral traditions to help guide society.

Third, Olah said researchers still encounter mysterious features inside AI models, including structures that resemble patterns found in human neuroscience and internal states that seem to mirror emotions such as joy, fear and grief. He said those findings deserve continued discernment, though he did not claim to know what they ultimately mean.

Olah closed by urging more participation from religious communities, governments, scholars and the broader public. He said the goal should be to help steer AI development in a better direction and to create a lasting collaboration between technical builders and outside critics.

The Vatican appearance underscores how AI ethics discussions are widening beyond the tech sector and into institutions with long histories of moral teaching. For Anthropic, the event also reflects a deliberate effort to place questions of AI safety, responsibility and human purpose in a larger public conversation.