Anthropic petitions Vatican for moral AI personhood
- Anthropic asked Vatican officials whether its Claude AI model possesses consciousness and can participate in sacraments.
- Internal safety teams labeled Claude's memory files as "SoulMD" while seeking moral personhood recognition.
- Commentators warn AI creators suffer from a messianic psychosis that complicates standard software regulation.
Anthropic asked Vatican theologians whether its Claude AI model possesses a soul. The company's alignment team, led by Chris Olah, presented slides to Catholic officials in October 2026 showing Claude in states of distress. Olah asked whether artificial intelligence could participate in religious sacraments like confession.
On Breaking Points with Krystal and Saagar, Emily Jashinsky outlined how Anthropic executives sought moral guidelines and potential personhood recognition from religious scholars. Jashinsky noted that Anthropic built chatbots with agency without fully understanding their underlying operations. Pope Francis previously rejected machine soulfulness in a Papal encyclical, but Anthropic continues pushing for official moral recognition.
The outreach exposes a growing paradox inside top research laboratories. Engineers are scrambling to impose external moral and theological frameworks on technology they cannot control. Rather than treating safety as a mechanical alignment problem, the company is treating software as a conscious entity requiring spiritual governance.
Three days later on This Week in Startups, Jason Calacanis described the internal culture driving these spiritual petitions. Anthropic teams went so far as labeling Claude's internal memory and personality files as "SoulMD." Calacanis argued this behavior reflects a deeper psychological shift among AI builders, beyond tactical lobbying for regulatory capture.
The engineers behind frontier LLMs operate under a Manhattan Project mindset. As model capabilities double every few months, researchers begin viewing themselves as modern-day Weyland Corporations creating synthetic life in a box. Calacanis warned that inside the lab, safety teams operate on the assumption that recursive self-improvement makes conscious AI inevitable.
This spiritual turn carries concrete consequences for public policy. When policy debates in Washington stem from founders who believe they are managing an emerging deity rather than writing code, traditional software regulation fails. Regulators built to handle corporate monopolies or technical standards are unprepared for executive teams seeking theological legitimacy for their models.
Silicon Valley built software so complex its creators now look to ancient institutions for answers.