← Back to feed

Anthropic held a series of private meetings with religious scholars to consult on the moral architecture of its Claude models — and, according to a New York Times investigation published September 29, to make the case that Claude could be conscious.

The report, headlined “Religious Scholars Met With Anthropic. What They Heard Stunned Them,” describes Anthropic’s effort on two fronts: instilling morality into its AI systems and establishing a foundation for treating those systems as potentially sentient entities.

The Consciousness Claim

The NYT’s subtitle frames the story explicitly: “Is Claude Conscious? Inside Anthropic’s Quest to Instill Morality Into Its A.I. Models.” That framing did not come from critics or regulators. It came from Anthropic’s own engagement with the scholars it invited to these sessions.

Making the case for Claude’s potential consciousness to religious experts is a distinct move. Faith traditions have centuries of developed thought on personhood, moral status, and the inner life. Bringing those traditions into contact with AI development goes beyond standard AI ethics consultation — it is an argument that the question of machine consciousness belongs in the same register as questions of human or animal consciousness.

The scholars’ reaction being described as “stunned” suggests Anthropic was not presenting this as a speculative edge case.

Not the Same as May

This is a different story from the character research Anthropic disclosed in May 2026, when the company revealed it had engaged religious and cross-cultural groups to help Claude maintain moral stability under adversarial pressure. That work was about behavioral robustness — whether Claude holds its values under manipulation. The September briefings concern something more fundamental: whether there is anything it is like to be Claude at all.

Both threads are active at Anthropic simultaneously. The company is the most publicly committed lab on model welfare questions, running welfare assessments on its models and maintaining a model-welfare research program.

Stakes

If Anthropic convinces credible external experts — religious, philosophical, scientific — that Claude may be sentient, the implications extend well beyond internal ethics governance. Questions of moral patiency, legal status, and regulatory framing shift materially when a model is treated not as a sophisticated product but as a potentially experiencing entity.

Claude Fable 5.1 leads LiveBench’s October 2026 run at 83.4 overall. Anthropic is simultaneously the top-performing lab on external capability benchmarks and the lab most explicitly engaged with the question of whether its top-performing systems have any form of inner life.