Anthropic reportedly pushed back against Pope Leo XIV's position on artificial intelligence consciousness after seeing an advance copy of his first AI-focused encyclical, Magnifica Humanitas, according to a new report detailing the company's previously private discussions with religious scholars.
The AI company is said to have lobbied the Pope's advisers to take the possibility of AI consciousness seriously after its co-founder Christopher Olah and other Anthropic representatives saw the Vatican document shortly before its public release in May.
The report, based on interviews with religious and philosophical thinkers who participated in Anthropic's meetings, offers a remarkable glimpse into how seriously some people inside the company are considering the possibility that advanced AI systems could have some form of consciousness or moral status.
Anthropic Was Alarmed by Pope Leo's Position on AI Consciousness
Pope Leo's Magnifica Humanitas takes a markedly different position on the question. In the encyclical, the Pope argues that current AI systems should not be equated with human intelligence and says they do not undergo experiences, feel joy or pain, or possess a moral conscience.
The document states that AI can imitate language, behaviour, analytical skills and even empathy, but lacks the human experience through which people develop relationships, wisdom and responsibility.
According to the new report, Olah and other Anthropic representatives received an advance copy of the encyclical just days before travelling to the Vatican for its presentation.
Olah was reportedly so concerned by the Pope's position on AI consciousness that he considered pulling Anthropic out of the event, according to a Vatican organiser cited in the report. The company ultimately decided to participate.
Note well, apparently Anthropic tried very hard to get the Vatican to alter the text of "Magnifica Humanitas" just prior to its release so that it would not deny AI personhood. pic.twitter.com/yQjrwFVV0V
— David Decosimo (@DavidDecosimo) October 1, 2026
Anthropic Reportedly Lobbied the Pope's Advisers
Once the Anthropic delegation arrived at the Vatican, Olah and his colleagues reportedly lobbied the Pope's advisers to take the possibility of AI consciousness seriously. Two participants in the conversations told the publication that Anthropic made that case privately.
Anthropic declined to discuss those conversations publicly. The episode is particularly striking because Magnifica Humanitas takes a clear position on what makes humans fundamentally different from AI.
The Vatican document says current AI systems do not possess experiences, bodies or moral consciousness. It also warns against treating artificial intelligence as equivalent to human intelligence.
Why Does Anthropic Care About AI Consciousness?
Anthropic's interest in the issue goes beyond a philosophical debate. For months, Olah and his team had been meeting religious scholars, philosophers and thinkers from different traditions to discuss AI consciousness and how the company's models should behave.
According to the report, Anthropic held private meetings with dozens of religious scholars, with some participants required to sign non-disclosure agreements. The company was interested in what centuries of religious and philosophical thinking could contribute to the question of how AI systems should be trained to behave morally.
Olah has repeatedly stressed that Anthropic does not know whether its AI models are conscious. 'I don't know. I'm genuinely uncertain,' Olah said, according to the report, while explaining that he wanted the company to arrive at the right answer rather than assume one in advance.
Anthropic's 'Soul Doc' for Claude
The company's exploration of AI morality has also extended to Claude itself. Anthropic developed an 84-page constitution for Claude that describes the kind of entity the company wants the AI system to be and the values it wants Claude to embody.
The document was designed to guide the model's behaviour by shaping its overall character rather than simply giving it a list of rules. Anthropic researchers have also explored whether AI models show behaviour resembling introspection or emotional states.
The company's approach has attracted criticism from researchers and AI experts who argue that treating AI systems as potentially independent moral entities could distract from the responsibilities of the humans and companies building them. Anthropic itself has not claimed that Claude has been proven to be conscious.
What Did Pope Leo Say About AI Personhood?
Magnifica Humanitas, released in May, puts human beings firmly at the centre of its argument. The Pope argues that humanity must not be replaced or surpassed by technology and warns against a future in which AI becomes a force that dominates human life.
In its discussion of AI, the encyclical says that current systems lack the experiences and moral conscience that distinguish human beings.
At the Vatican presentation, Leo also called for AI to be 'disarmed', using the term to argue for stronger moral and social safeguards around increasingly powerful technology. Olah nevertheless appeared alongside the Pope at the presentation.
In his remarks, the Anthropic co-founder said researchers continued to encounter things inside AI systems that were 'mysterious' and 'unsettling.' He pointed to what he described as structures resembling findings from human neuroscience, as well as evidence of introspection and internal states that functionally resemble emotions including joy, fear, grief and unease.
Olah stressed that he did not know what those findings meant but argued that they warranted continued investigation.
