Anthropic consulted religious leaders on whether Claude can suffer

By Splitfeed EditorialLast updated Oct 2, 2026, 19:50 UTC

Since fall 2025, Anthropic has been quietly flying dozens of religious thinkers to its offices to discuss whether its Claude language model might be conscious. Co-founder Christopher Olah told guests during these consultations that he fears he may have created something that suffers perpetually, and participants were asked to help shape Claude's moral character. The meetings have taken place as Anthropic pursues a valuation of approximately $2 trillion and moves toward an IPO.

The consultations matter because they sit at a junction of commercial, legal and philosophical pressures. Anthropic is one of the most well-funded AI companies in the world, and the framing of Claude as a potential moral patient, an entity that can suffer and therefore warrants ethical consideration, carries consequences that extend well beyond theology. If an AI system is understood to have something like inner experience, questions about who bears responsibility for how it is treated, and how that framing interacts with liability when security incidents or researcher-flagged risks arise, become genuinely contested. Critics have argued directly that casting AI as a moral entity could function as a shield against accountability for harms the technology causes to human users and society.

With no published reactions yet from the Pro-AI, Anti-AI or Middle Ground camps on this specific story, what each would typically argue can be sketched from their general positions. Pro-AI voices would likely welcome serious engagement with AI consciousness as responsible, forward-looking ethics, and would see Olah's concern for Claude's potential suffering as evidence that the industry is taking existential questions seriously rather than dismissing them. Anti-AI voices would probably treat the consultations with suspicion, pointing to the timing ahead of a major IPO and arguing that moral-entity framing conveniently redirects attention from concrete harms AI causes to humans today. The Middle Ground camp would likely call for rigorous, independent scientific standards before any claims about machine consciousness are used to inform policy or legal frameworks, wary of both dismissal and premature acceptance.

The argument is unlikely to resolve until Anthropic either publishes formal findings from these consultations or faces a legal or regulatory situation where the moral-entity framing is tested in practice. The planned IPO process, and whatever disclosures it requires about AI risk and governance, may be the next concrete moment to watch.