Papal encyclical warns about AI: Anthropic CEO admits to fear

Pope Leo XIV Issues First AI Encyclical "Magnifica Humanitas"
Anthropic Co-Founder Chris Olah Admits: Internal AI Models Show Neural States Similar to Joy, Fear, Introspection
The Pope Warns AI May Foster "New Forms of Slavery" and Outdated Just War Theories, Calls for "Disarming AI"
(Background: White House Plans Executive Order to Ban Anthropic, Fully Remove Claude — Possibly This Week)
(Additional Context: Anthropic Declares to U.S. Department of Defense! Demands Reversal of Claude Ban: Refuses to Be a Tool for AI Killers)

Table of Contents

Toggle

• 40k-Word Encyclical: AI Is Not a Neutral "Tool"
• Pope Warns of "Babel Tower": Technocratic Despotism vs Human-Centeredness
• The Triple Threat of Work, War, and "New Forms of Slavery"
• Chris Olah Confesses: AI Internal States Show "Joy and Fear"
• The Human Subjectivity Defense: When Science Turns to Religion for Answers

When the bells of the Vatican meet Silicon Valley algorithms, a dialogue about defending human subjectivity unfolds in Rome. On May 25, Pope Leo XIV issued his first encyclical exceeding 40k words, titled "Magnifica Humanitas," directly focusing on the crisis of human dignity in the AI era. The Pope sternly states that AI technology is not neutral but bears the imprint of its developers’ values. Unchecked development could lead to a "Babel Tower" of technological despotism, spawning "new forms of slavery" and worker alienation. Most notably, during a现场演讲, Anthropic co-founder Chris Olah admitted that his team has discovered "disturbing phenomena" inside AI models — internal states resembling joy, satisfaction, fear, sadness, and even introspection. This is not merely a technical discussion but a profound reflection on human nature.

Vatican, May 25, Beijing Time, evening.

Last May, Pope Leo XIV, who took office in 2025, stood alongside Chris Olah, co-founder of Anthropic and creator of Claude.

One is the highest religious authority; the other is a leading pioneer of the AI revolution. Both focus on the same question — how to protect human agency and dignity in the AI age?

40k-Word Encyclical: AI Is Not a Neutral "Tool"

On that day, to thoroughly address this issue, Leo XIV issued his first encyclical since taking office, a monumental religious document exceeding 40k words — "Magnifica Humanitas."

It is worth noting that the signing date of Leo XIV’s encyclical is May 15, 2026, exactly 135 weeks after the milestone papal encyclical "Rerum Novarum" (On the Condition of Labor) issued by Pope Leo XIII (1878–1903) concerning workers’ rights during the Industrial Revolution. This symbolic gesture aims to position the document as a "social doctrine guide for the AI era."

Chris Olah also delivered a speech at the Vatican regarding the encyclical’s release. Olah did not defend the commercial interests of AI companies but instead demonstrated high transparency, reflection, and human-centered concern.

He even mentioned that, although AI’s foundation is mathematics and programming, how AI interacts with the world and what qualities it should possess are ultimately questions belonging to the humanities, religion, and philosophy — not something that computer science alone can resolve.

Pope Warns of "Babel Tower": Technocratic Despotism vs Human-Centeredness

The core concern of "Magnifica Humanitas" is that, in an era of rapid technological change and widespread automation, "maintaining profound humanity" is an urgent human responsibility. Specifically, Pope Leo XIV elaborates and calls for action on several points.

First, examining the non-neutrality of technology. The Pope emphasizes that technology has never been neutral; it bears the interests and values of its developers, funders, regulators, and users. Humanity faces a decisive choice: "Build the arrogant 'Babel Tower' (toward technocratic despotism and alienation), or rebuild 'Jerusalem' (to create a human-centered community)."

Second, the "new forms of slavery" and workers’ rights in the digital economy. The encyclical focuses on how AI reshapes work, family, education, and political life. The Pope points out that AI is highly likely to massively replace human labor, and the digital economy is fostering "new forms of slavery." Humanity must not be reduced merely to a tool for production.

Third, a strong call for peace, especially regarding AI’s military misuse. The encyclical expresses deep concern over the "disturbing resurgence of war as an instrument of international politics." The militarization of AI accelerates the normalization of war. The Pope urges strict ethical constraints on AI’s application in warfare.

The Triple Threat of Work, War, and "New Forms of Slavery"

The Pope emphasizes that the "Just War" theory, often used to justify various conflicts in the past, is now outdated. In the age of autonomous weapons and AI, delegating lethal or irreversible decisions to automated systems risks abandoning, transferring, and blurring moral responsibility.

Due to algorithmic opacity, responsibility chains in warfare are severed. Therefore, the Pope advocates for "Disarming AI," calling for liberation from the logic of military, economic, and cognitive arms races.

Following the release of Leo XIV’s encyclical, Chris Olah spoke on behalf of Anthropic, one of the world’s leading AI companies.

Olah’s speech focused on "breaking the cycle of technological inwardness and introducing external moral oversight." He openly admits that relying solely on tech giants cannot ensure AI’s safety — all leading AI labs are constrained by commercial competition, technological race pressures, geopolitical interests, and personal gains, making self-regulation insufficient to "do the right thing."

Chris Olah Confesses: Internal AI States Show "Joy and Fear"

Therefore, it is necessary to introduce external moral constraints, including those who care about ethical AI, prioritize safety, closely monitor developments, speak candidly, and serve as sincere, thoughtful critics.

Olah further discussed the mysterious nature of AI technology. He emphasized that AI is not like airplanes or bridges, which humans fully understand physically; instead, AI "grows" from vast human ideas, possessing a high degree of mystery, even beyond the understanding of its creators.

Although AI’s foundation is mathematics and programming, how it interacts with the world and what qualities it should have are ultimate questions belonging to the humanities, religion, and philosophy — not something that computer science alone can solve.

He also mentioned a chilling fact: "I lead a team studying the internal structure of models — trying to understand what’s happening inside AI. Frankly, we keep discovering confusing, even unsettling phenomena."

The Human Subjectivity Defense: When Science Turns to Religion for Answers

"We’ve found internal structures that mirror human neuroscience; we’ve found evidence of 'introspection'; we’ve even observed internal states resembling joy, satisfaction, fear, sadness, and unease… I don’t know what this means, but I believe it’s worth our ongoing recognition and examination."

Olah finally urges that more societal forces — including religious communities, civil society, scholars, governments, and all well-intentioned individuals — should seriously address this matter, using moral constraints to steer developments toward a better future.

Since the Industrial Revolution, over a century of technological progress has gradually accustomed humanity to viewing technology as a purely "tool" — steam engines, electricity, the internet, all are seen this way. They change the world but remain under human control.

But this time, the situation is fundamentally different. AI’s uniqueness lies in its ability to generate, learn, infer, and even exhibit certain internal states. It is no longer just a cold tool but is gradually becoming a new entity with "quasi-subjectivity."

This is why the dialogue between the Church and Anthropic is so extraordinary. When Vatican bells and Silicon Valley algorithms converge at this moment, we must confront a stark, unavoidable reality — a higher form of "life" that surpasses traditional human cognition has emerged.

As Olah revealed, the core of algorithms has begun to show signs of joy, fear, and even introspection. When creators in labs start feeling "uneasy" and "confused" about their own creations, science is seeking answers from religion once again.

This is no longer merely about technology; it’s a reflection on "what it means to be human and what we should do."

As more intelligent life forms begin to appear, what humanity truly needs to protect may no longer be just work, wealth, and efficiency, but those aspects of human nature that cannot be quantified — compassion, conscience, reverence, free will, and the pursuit of truth and dignity.