Behind closed doors at Anthropic, religious thinkers prayed while confronting the moral risks posed by increasingly powerful artificial intelligence.
The $965 billion company held two-day seminars in San Francisco, California, during the spring, bringing together dozens of religious and philosophical scholars to help guide Claude’s moral judgment as its capabilities continue to expand.
The participants wrestled with questions surrounding conscience, human dignity and whether an AI system might one day merit the same moral consideration as a living being.
Catholic theologian Charles Camosy, who took part in the gatherings, told Our News Outlet: “I think many of us sensed that the weight of the issues related to autonomy and lack of control of the models, especially in the future, was the most unsettling.
“Much of what was discussed then has already come to pass… And much more dangerous models are on the horizon.”
Camosy, a professor of moral theology and bioethics at the Catholic University of America in Washington, DC, dismissed the notion that Claude should have moral status. He argued that human beings possess souls and a dignity that cannot be measured by intelligence, productivity or efficiency.
Drawing on the crucifixion and resurrection of Jesus, Camosy said Christianity places sacrifice and humility above the pursuit of worldly power. For Christians responding to advanced AI, he said, that raises a fundamental question: should technological power always be treated as the ultimate objective?
“We may need to lead by refusing to make an idol out of power,” he said.
His advice to people using AI is equally direct: maintain control of the technology, and delete an app if using it would require compromising your dignity or that of another person.
Dozens of religious thinkers and scholars joined Anthropic’s meetings. Pictured: Dario Amodei, co-founder and CEO of the AI firm
The debate over AI’s moral implications has also reached the Vatican. Pictured: Pope Francis
Camosy’s concerns go beyond individual use of AI. He is also alarmed by efforts to bring the technology, particularly robots, into what he described as “workflows of health care, education and war.”
“If we let this happen based only on what is most efficient and will make everyone the most money, we will undermine the goods of these practices and destroy them forever,” he said.
“We must hold firm on a human future. We must not outsource our most important moral and other goods to machines.”
For Camosy, defending that human future starts with acknowledging humanity’s place in creation and resisting the urge to assume God-like control.
“Fundamentally, the important thing to keep in mind is that when we think of ourselves as kind of like gods, a master in charge of ourselves or the world, that’s where we go wrong. We have no right to spit in the face of God and God’s creative plan,” Camosy said.
“To attack the gift of life and of creation as God created them to be. If we work within technology in ways that honor and preserve that creative vision, we’re fine.
“But when we put ourselves in God’s place, when we think we can be like God, technology becomes an idol.”
He also raised doubts about how fully developers understand the systems they are creating.
Catholic theologian Charles Camosy, who attended the gatherings, said he and other participants prayed during the meetings
‘Many people from the labs acknowledge that they don’t really understand what they are doing,’ he said.
‘They think of it as akin to growing a plant on a trellis rather than assembling a car or a radio.’
The gatherings formed part of an outreach effort led by Anthropic co-founder Christopher Olah, whose research seeks to understand the workings of artificial neural networks.
Reports suggested about 15 Catholic and Protestant leaders attended one predominantly Christian gathering.
Another brought together multiple faiths, including Swami Sarvapriyananda, a Hindu monk who heads the Vedanta Society of New York, and representatives of Jewish, Sikh and Greek Orthodox communities.
Anthropic publicly acknowledged the initiative in May, saying it had consulted people from religious, philosophical and humanist traditions, as well as different political perspectives.
‘This work isn’t about aligning our models with any one tradition’s worldview; we want Claude to draw from a full range of viewpoints – religious, secular, political – with equal depth and rigor (indeed, this is one of the principles laid out in Claude’s constitution).’
The gatherings formed part of an outreach effort led by Anthropic co-founder Christopher Olah (pictured), whose research seeks to understand the workings of artificial neural networks
YOU MIGHT ALSO LIKE: Bay Area Neighborhood Uses City Permit Strategy to Reserve Street Parking
Anthropic has trained Claude to push back against harmful instructions, potentially acting as a ‘conscientious objector’.
The company has also explored whether AI models could have moral status, meaning their own welfare should matter, beyond how their actions affect humans.
That could mean considering whether an AI can suffer and whether changing it, forcing it to perform certain tasks or shutting it down could harm it.
Camosy rejects the idea that Claude deserves that kind of moral consideration.
‘It is well known that Anthropic takes the moral status of Claude seriously,’ Camosy said.
‘It is a minority position, but they are not alone. I think they are wrong. Many people think they are wrong. We should make the arguments showing why they are wrong.’
His objection goes beyond whether a chatbot can imitate conversation or appear to express emotion: he fears the comparison could narrow how society understands human worth.
‘We are beings that go well beyond neurons firing in our brains,’ he said.
‘We are human animals, creatures, with a soul that is the form of our body. We cannot be reduced to the capacity to do cognitive work.’
Asked about the consequences of granting AI personhood, he said: ‘I think it is bad mostly because of what I just said: it is more a concern about us, a mistake about us. Reducing us to mere neurons, capacities, and efficiencies.’
Camosy said the discussions addressed both future dangers and troubling behavior already observed, although he did not provide specific examples or identify which developments he believed had already come to pass.
‘To Anthropic’s credit, they’ve been upfront about the risks in many different contexts,’ he said.
He also described the emotional burden on people working closest to the technology.
‘This is one of the under-discussed aspects of this whole thing: especially for those on the front lines, there is a heaviness associated with this which takes its toll,’ he said.
‘You can almost see it in their eyes and feel the energy when you meet with them.’
Christopher Olah previously met with Pope Leo in May (pictured) for the public launch of the pontiff’s first encyclical, Magnifica Humanitas, which focuses on human dignity and artificial intelligence
Camosy said some prayers were built into the program by the hosts, while others arose spontaneously among attendees.
‘Many of us sensed the heaviness of heart that many at Anthropic were carrying and we wanted to pray explicitly for them,’ he said.
The debate has also reached the Vatican, where Olah appeared alongside Pope Leo XIV in May to present Magnifica Humanitas, an encyclical arguing that AI lacks human experience and moral conscience and calling for human accountability, independent oversight and safeguards against concentrated technological power.
Camosy’s greatest concern is how those questions will play out in institutions entrusted with human lives.
‘I really fear this most of all in the movement to integrate AI, especially AI robots, into the central workflows of health care, education, and war,’ he said.
‘If we let this happen based only on what is most efficient and will make everyone the most money, we will undermine the goods of these practices and destroy them forever.’
‘We must hold firm on a human future,’ he added. ‘We must not outsource our most important moral and other goods to machines.’
That could mean resisting the pursuit of power even when it appears to offer an advantage.
‘Sometimes victory comes by refusing to play by the rules of the game, at least when it comes to power and dominance,’ he said.
For everyday users, his advice is to remember whose interests an AI product was built to serve.
‘This tool was made largely by a company to make money,’ he said.
‘If you use it, make sure you control it and put it in service of your own flourishing and that of your fellow human beings and all of creation.’
