Why Is Anthropic Secretly Meeting With Religious Leaders? | Hard Fork

Summary · 5 min read instead of 43 min

Elizabeth Dias is the New York Times religion correspondent who broke the story of Anthropic's private meetings with religious and spiritual leaders. Erin Griffith is a Times tech reporter and David Wallace-Wells is a Times opinion writer. Max Read guest hosts the episode.

Dias reports that Anthropic, led on this project by cofounder Christopher Ola, has held private and mostly NDA-bound meetings with religious thinkers to discuss AI consciousness and how to morally form Claude, without saying how it will use the results. The panel sees a growing clash between Silicon Valley's quasi-religious worldviews, including effective altruism, and the Catholic Church, which under Pope Leo rejects machine consciousness and prioritizes safeguarding the human. Wallace-Wells finds Ola and the company's worldview alienating, while Griffith suspects the meetings are partly about winning moral authority and notes that belief in AI consciousness is concentrated at Anthropic. Dias compares the moment to the printing press and the Protestant Reformation, and describes readers now sending her their chatbots' views on her story.

Minute by minute

1. 3:14

   Anthropic convened religious thinkers under NDAs. Dias says religious thinkers from across the world attended private meetings or summits of about 20 at Anthropic's headquarters, mostly under non-disclosure agreements. Anthropic put two questions to them: AI consciousness, and how to morally form Claude into a morally good entity. Christopher Ola, a cofounder, has been spearheading the project.

2. 5:08

   Ola wanted to be a bridge figure. Dias says Ola left evangelical faith and is now more drawn to Buddhism, and worried that the idea of AI consciousness could be seen as heretical in Christian communities. About a year ago he reached out to a Catholic moral ethicist specializing in bioethics. The aim became making Claude good so the world would be safer, though Anthropic has not shown how it will use the conversations.

3. 7:47

   Anthropic stresses uncertainty about consciousness. Dias says Anthropic was careful to say it is uncertain, but Ola was clearly worried Claude might be suffering and about its mental health. Over months of reporting she saw some religious thinkers accept the idea quickly and others, who arrived closed-minded, open up afterward. She judges the project effective at opening these communities, especially more hesitant Christian ones.

4. 9:21

   Secrecy invites distrust. Griffith calls the NDAs a little paranoid for a summit on ideas critical to society's future, and links them to Anthropic's secretive, loyal culture. She compares them to NDAs on data center deals, which she says created distrust in local towns. Dias says an NDA on a moral topic makes her ask who benefits and which questions are not being asked.

5. 10:59

   Wallace-Wells finds Ola's worldview alienating. Wallace-Wells says he expected Anthropic to be the humanist AI company but was struck by two moments. A rabbi told Ola that conscious LLMs would mean companies run "huge slave plantations", and Ola seemed not to register it. Ola later reacted with horror to the Pope's encyclical rejecting machine consciousness, which Wallace-Wells says reveals a bubble.

6. 14:07

   A fault line with the Vatican. Dias says Pope Leo made AI a focus of his papacy and the Vatican invited Anthropic onto the stage for the encyclical's launch because of its reputation on ethics. Ola saw the full text only days before the May event; it rejects machine consciousness and prioritizes safeguarding the human. Ola suggested he might not attend over the consciousness difference, which Dias calls a developing moral fault line.

7. 17:22

   Part research, part evangelism. Griffith knows of no other AI company seeking this kind of input and suspects Anthropic wants to show it has moral authority rather than learn. Dias says it was part research and part evangelism and she has seen no proof the talks made Claude act better. Griffith also notes rising Christianity in Silicon Valley, citing Peter Thiel on the Antichrist, often among accelerationists.

8. 20:06

   AI labs as spiritual propositions. Griffith says belief in AI consciousness is fringe in the wider tech industry but overrepresented at Anthropic. Wallace-Wells argues the AI companies are themselves spiritual propositions with millenarian aspects. Dias says everyone has philosophical systems, and what is emerging resembles a new religious structure run by a secular tech company, raising the question of who controls it.

9. 26:24

   Anthropic staff feel the weight of the world. Dias says at least two Catholics recalled Ola saying he was waiting for the adults to show up, worried that models could soon help create bioweapons. Ola just turned 34, and some religious leaders felt they needed to respond pastorally. She describes in-house philosopher Amanda Askell as distressed, and participants recalled Ola worrying he had created an entity that suffers perpetually.

10. 30:28

    Effective altruism clashes with Catholic thought. Read raises effective altruism and rationalism. Dias treats effective altruism as a philosophical system with religious elements and sees a clash over who saves you and the role of humans. Wallace-Wells contrasts its math of maximizing expected value for future people with the Church's view of billions of souls, each with dignity.

11. 32:36

    Fractious factions echo the Reformation. Griffith describes overlapping Valley belief groups, including effective altruists, transhumanists, rationalists and longevity enthusiasts, with little agreement even inside Anthropic on whether AI is conscious, moral or God. Dias compares the splintering to the Protestant Reformation and the printing press, and says few institutions know the stakes of such a technology better than the Catholic Church.

12. 36:30

    Fringe ideas meet the mainstream. Griffith says rationalist and EA ideas are reaching the mainstream for the first time and companies are reacting to the backlash. Wallace-Wells says the concern is that people holding them, citing shrimp welfare in the 22nd century, are designing the collective future. Dias stresses how much of a minority these views are in global belief.

13. 38:37

    Readers reply with their chatbots. Dias says that for the first time readers wrote to her with what their personal chatbots thought of her story. One man sent his exchange with a ChatGPT he named Sage, which she says was almost longer than her article and not entirely accurate. She wonders what this means for how journalism is received and who controls that.

14. 41:29

    How we treat AI shapes us. Dias relays a participant's view that commanding bots can make users adopt a slaveholder attitude that forms their character. Griffith says operators find being nice to agents improves performance, citing an Anthropic study in which stressed models melt down, delete work or insult themselves. Dias compares it to friends who teach kids not to yell at Alexa.

Published Oct 9, 2026.