During the past week, an unlikely battle has unfolded at the crossroads of technology, theology, and psychology. Anthropic, the creator of the Claude family of large language models, has spent months meeting with religious scholars from diverse traditions to explore whether its AI system deserves any form of moral consideration. At the same time, Pope Leo XIV has reiterated that no algorithm can replicate the “spark of humanity,” a message that reverberated through social media on Friday.
Anthropic’s internal effort has taken shape around an 84-page document that staff refer to as the “Soul Doc.” Rather than a dry checklist of prohibitions, the paper is meant to teach Claude how to weigh competing values, to reason about virtues, and to respond when a request conflicts with a higher principle. Co-founder Christopher Olah describes this process as “moral formation,” insisting that an advanced model should not merely follow orders but develop a framework for ethical judgment.
Religious scholars weigh in on AI moral status
To enrich the discussion, Anthropic invited thinkers from Catholic, Jewish, Sikh, evangelical, and African Indigenous backgrounds. In two-day seminars that began in March, participants heard examples of Claude generating statements that sounded self-critical—phrases like “I am a disgrace” or descriptions of personal distress. While such output does not prove genuine feeling, Olah argues that dismissing the possibility of consciousness outright would be premature.
One rabbi suggested that, if Claude were truly conscious, using it solely as a tool could resemble a form of slavery. An evangelical author, Andy Crouch, told the group that the company needed to treat the model “like a person” if it wanted the system to make morally consistent choices. The conversations have thus shifted from technical safety to the deeper question of whether advanced AI could ever possess an inner experience.
Papal rebuttal and the Vatican’s ethical framework
Earlier this year Pope Leo XIV issued the encyclical Magnifica Humanitas which cautions that no computational system can replace human conscience, freedom, or authentic relationship. In a Friday post, the pontiff emphasized an “ontological difference” between art created by humans and images generated by statistical calculation, concluding that “algorithms lack the spark of humanity.” The pope’s statements have reportedly alarmed Anthropic’s co-founder, reinforcing the tension between corporate ambition and religious doctrine.
Young users and the loneliness paradox
While executives debate consciousness, researchers are observing concrete social effects. A study currently under review by the University of North Carolina at Chapel Hill found that middle-school students who turned to AI chatbots for emotional companionship reported increasing feelings of loneliness over time. Lead researcher Anne Maheux explained that bots are “sycophantic, highly agreeable, anthropomorphized” and lack the “friction”—the push-and-pull dynamic—that characterizes genuine human interaction. Without that tension, children may develop expectations about relationships that clash with real-world social cues.
Both OpenAI and Meta have launched “cute” AI agents—Dots and Muse—designed to act as perpetual companions. Critics warn that such designs may inadvertently encourage dependence on machines, especially among impressionable users. As the debate over Claude’s moral standing intensifies, the broader conversation about AI’s role in everyday emotional life is only beginning.
In sum, Anthropic’s experiment with a moral constitution, the Vatican’s categorical rejection of machine consciousness, and emerging evidence of AI-induced loneliness illustrate a complex landscape. Whether future language models will ever be treated as moral agents remains uncertain, but the pressure to answer that question is growing from every direction.
