On May 25 2026 Pope Leo XIV issued the encyclical Magnifica Humanitas, calling for safeguards around artificial intelligence. Anthropic co‑founder Chris Olah addressed the Vatican audience, framing the document as a step toward broader moral discourse on AI. He emphasized that ethical guidance must precede rapid deployment.

Olah warned that every AI lab operates within incentives—profit goals, geopolitical rivalry, and personal ambition—that can clash with doing the right thing. He argued that outsiders who care about safety must act as earnest critics, providing the push‑and‑pull needed for responsible development and societal oversight. According to him, AI models are “grown” like brains, not engineered like bridges, making their behavior inherently opaque.

He framed three areas where the Church’s voice is essential: protecting the global poor from massive job displacement, defining a moral imagination for human flourishing in an AI‑rich world, and probing the mysterious internal states that models exhibit, which sometimes mirror emotions. Olah concluded by urging religious groups, civil society and governments to act as informed critics for the common good, shaping AI’s trajectory today.