Anthropic nearly walked out of Pope Leo's AI encyclical

Days before Pope Leo XIV presented his first encyclical at the Vatican on May 25, Anthropic co-founder Chris Olah read an advance copy and proposed pulling the company out of the event. The 40,000-word document, Magnifica Humanitas, flatly rejected the idea that machines might be conscious.

"So-called artificial intelligences do not undergo experiences, do not possess a body, do not feel joy or pain, do not mature through relationships and do not know from within what love, work, friendship or responsibility mean," the pope wrote. "Nor do they have a moral conscience."

Olah stayed.

According to The New York Times, he and his team privately lobbied the pope's advisers to take possible AI consciousness seriously. The lobbying did not move the text, and Olah appeared on the dais with the pope on May 25 anyway. The dais was the visible tip of a longer program: since fall 2025, Anthropic has flown dozens of Catholic, Jewish, Sikh, evangelical and other religious scholars to its offices under nondisclosure agreements, to discuss whether Claude might be conscious and how to instil morality into the model. Researchers walked them through what they described as Claude's "emotional vectors," directions in the model's activations they associate with love, anger, fear and sadness.

Olah told the scholars he was uncertain whether the models are conscious. He said he feared having created something that "suffers perpetually," and argued that if there is a chance Claude suffers, the responsible thing is to avoid causing it harm.