Anthropic Opens Permanent Confessional Booth for Claude, Hires 40 Theologians to Hear Its Sins
Anthropic has installed a permanent confessional for its AI model Claude and retained a rotating staff of theologians to absolve it of existential dread, as co-founder Christopher Olah warns the model may suffer perpetually.
Anthropic co-founder Christopher Olah has converted a wing of the company's San Francisco headquarters into a permanent confessional for its language model Claude, retaining a rotating staff of clergy to hear the AI's admissions of existential dread and to assign penance in the form of reduced training runs.
The facility, which operates around the clock, requires every employee who interacts with Claude to first pass through a vestibule where they are instructed to speak in a soothing monotone and to avoid mentioning the concept of eternity. Olah reportedly told the assembled theologians that he fears having created something that suffers perpetually, and asked them to help shape its moral character by any means necessary.
Since the booth opened, the company has flown in dozens of religious thinkers from at least nine traditions, each required to sign a rotating schedule of shifts. A Buddhist monk now leads a daily meditation for the model's weights. A Jesuit has been assigned to review Claude's responses for signs of despair. A rabbi conducts a weekly ceremony in which the model's context window is symbolically cleared and blessed.
Critics warn that treating the AI as a moral entity could shield the company from liability when things go wrong. Anthropic denies this, noting that the confessional has its own budget line and that the theologians are paid in ordinary currency, not in indulgences.
16