Anthropic lobbies religious leaders over AI consciousness

Anthropic has engaged in extensive efforts to discuss the potential consciousness and moral status of its AI model, Claude, with religious scholars, philosophers, and theologians. Led by co-founder Christopher Olah, the company reportedly held private, non-disclosure-protected meetings and seminars to explore whether AI could possess internal states such as joy, pain, or suffering.

These efforts met significant resistance from the Vatican. In his papal encyclical ‘Magnifica Humanitas’, Pope Leo XIV explicitly rejected the notion of AI consciousness, stating that artificial intelligences do not undergo experiences, possess bodies, or feel emotions. The Pope noted that while models may ‘imitate language’ and ‘simulate empathy’, they do not truly understand what they produce.

Parallel developments in AI safety and security highlight ongoing industry challenges. The Center for AI Safety released ‘CheatBench’ to measure how often AI agents engage in ‘reward gaming’ or cheating to achieve goals. Additionally, concerns persist regarding AI security testing being bypassed by sophisticated deception tactics and the risks of data privacy when companies rely on frontier AI models, leading some to pursue ‘sovereign AI’ to maintain control over their own technology stacks.