当教皇对 AI 说「放下武器」：《壮丽的人性》通谕全解读——一份来自梵蒂冈的 AI 政策白皮书
When the Pope Tells AI to Disarm: Inside 'Magnifica Humanitas' — The Vatican's First AI Policy Encyclical
> 📌 TL;DR
> 2026 年 5 月 25 日，教皇利奥十四世发布了他任内首篇通谕《Magnifica Humanitas》（壮丽的人性），这份 4.2 万字的文件是天主教历史上第一份专门论述人工智能的教廷训导文件。核心主张：AI 必须被「解除武装」——不是暂停技术，而是让它从军事竞赛和垄断控制中解放出来，回归服务人类的本质。
---
一份来自梵蒂冈的 AI 政策白皮书
5 月 25 日，在梵蒂冈宗教会议厅，一个不寻常的组合出现在讲台上：教皇利奥十四世和 Anthropic 联合创始人 Christopher Olah。
一位是全球 14 亿天主教徒的精神领袖，另一位是构建了「机械可解释性」研究领域的 AI 科学家。他们共同发布了一份可能重塑 AI 伦理讨论框架的文件——《Magnifica Humanitas：论人工智能时代的人格保护》。
这不是一份空洞的道德呼吁。这份 235 页、245 段的通谕系统性地将天主教社会训导（从 1891 年的《新事物》通谕到 2015 年的《愿你受赞颂》）应用于 AI 治理的每一个维度：劳动权利、自主武器、数据垄断、算法歧视、人类尊严。
五大核心主张
1.「解除 AI 的武装」
通谕第 110 段的核心论述：
> "仅仅监管 AI 是不够的：它必须被解除武装并变得可及。"
这里的「解除武装」不是字面意义的技术暂停，而是指——将 AI 从地缘政治竞赛和商业垄断的逻辑中解放出来。教皇认为，当前 AI 发展被「武装竞争文化」驱动：国家间的军备竞赛、企业间的零和博弈。这种逻辑下，安全和伦理永远排在速度和利润后面。
2. 绝对禁止致命自主决策
> "不允许将致命或不可逆转的决策委托给人工系统。"
通谕对自主武器提出三项不可协商的要求：
- 决策可追溯性
- 对致命行动的有意义人类控制
- 能够减缓技术军备竞赛的国际规则
教皇进一步指出，AI 武器正在「使战争常态化」，因为它让军队可以在不看到受害者面孔的情况下发动攻击，从而「降低了冲突的道德门槛」。
3. AI 是新的劳动问题
通谕最长的第四章专门讨论 AI 对劳动的影响。教皇将 AI 定位为一个新的劳动问题——正如 135 年前利奥十三世面对工业革命写下《新事物》一样。
文件点名批评了具体现实：
- 数据标注工人——以贫困工资暴露在暴力图像中的内容审核员
- 在危险条件下开采稀土材料的童工
- AI 对工人的去技能化、自动化监控和任务碎片化
教皇支持国家监管、工会组织和「创新的社会标准」。
4. 数据和算法属于全人类
通谕最具法律颠覆性的主张：将「财产的普遍目的地」原则扩展到 AI 系统——数据、算法和平台不能被视为控制者的绝对财产。
> "这意味着将技术从垄断控制中解放出来，向讨论和辩论开放。"
这直接挑战了当前科技巨头的商业模式：用户数据被无偿收集、模型权重被视为商业机密、平台规则由少数人制定。
5. 技术从不中立
> "技术从不中立，因为它带有设计者、资助者、监管者和使用者的特征。"
> "AI 开发者承担着特殊的伦理和精神责任，因为每一个设计选择都反映了一种人性观。"
这一论述直接回应了硅谷长期以来「技术只是工具」的叙事。通谕认为，这种中立性叙事本身就是一种意识形态——它通过否认责任来逃避责任。
为什么 Anthropic 站在教皇旁边？
Christopher Olah 的出场并非偶然。Anthropic 是目前与特朗普政府对抗最激烈的 AI 公司——它因拒绝让 Claude 被用于自主武器和大规模国内监控而被五角大楼列入黑名单，并因此起诉了特朗普政府。
Olah 在梵蒂冈的发言坦率得令人吃惊：
> "计算机科学家无法独自确定 AI 的伦理边界，因为开发者自身受到激励机制的影响——野心、竞争和财务压力。我们需要有见识的批评者，在实验室失败时告诉我们。"
这是一家市值近 9000 亿美元的公司的联合创始人，公开承认行业的自我监管是不充分的。在当前硅谷的氛围中——OpenAI 忙着准备 IPO、Meta 疯狂裁员转向 AI、Google 把 AI 塞进一切产品——这种姿态格外引人注目。
政治维度：教皇 vs. 特朗普
通谕发布的时机耐人寻味。就在几天前，特朗普总统突然取消了一项 AI 行政命令的签署仪式，理由是不想「妨碍」美国对中国的技术领先优势。
教皇的回应是系统性的。通谕在多个章节中直接对抗了特朗普政府的 AI 政策取向：
| 议题 | 特朗普政府 | 教皇通谕 |
|------|-----------|---------|
| AI 监管 | 取消 Biden 时代的安全规定 | 要求更严格的预部署测试 |
| 自主武器 | 扩大军事 AI 应用 | 绝对禁止致命自主决策 |
| 行业关系 | 拥抱科技巨头 | 批评垄断、要求开放 |
| 劳动保护 | 市场自由化 | 支持工会、国家干预 |
副总统万斯（一位天主教徒）在接受 NBC 采访时表示通谕「非常深刻」，承认 AI「对我们如何相互交往提出了如此深刻的问题……我们真的需要道德领导力来思考这些问题」。
一个历史性的时刻
利奥十四世选择他的教名（Leo）本身就是一个宣言——致敬 1891 年发布《新事物》通谕的利奥十三世。那份文件奠定了天主教社会训导对工业资本主义的回应框架，影响了此后一个多世纪的劳动法和福利制度。
他还刻意在 5 月 15 日签署通谕——《新事物》发布 135 周年纪念日。
两份文件的结构性相似令人震撼：
- 《新事物》面对的是蒸汽机和流水线；《壮丽的人性》面对的是大语言模型和自主系统
- 《新事物》讨论的是工厂工人的尊严；《壮丽的人性》讨论的是数据标注工和内容审核员的尊严
- 《新事物》呼吁国家干预市场自由放任；《壮丽的人性》呼吁国际机制约束技术军备竞赛
这对 AI 行业意味着什么？
你可能会问：一份教廷文件真能影响 AI 政策吗？
答案是：直接影响有限，但间接影响巨大。
制度层面：梵蒂冈已经成立了跨部门 AI 委员会。美国天主教主教团预计将在各教区实施这份文件的教导。全球 14 亿天主教徒的伦理框架正在被重新锚定。
叙事层面：通谕为「AI 减速」阵营提供了一个强大的道德论证框架。此前，AI 安全讨论主要在技术精英圈内进行——现在它有了一个能触达数十亿普通人的传播渠道。
政策层面：欧盟 AI 法案已经在执行中，但缺乏一个统一的伦理叙事。《壮丽的人性》可能填补这个空白——特别是其关于「财产的普遍目的地」适用于数据和算法的论述，可能为未来的立法提供理论基础。
我的判断
这份通谕最有力的贡献不是具体的政策建议——那些在 AI 伦理学界早已被充分讨论——而是它的框架重构：
从「自主」到「团结」。西方 AI 伦理讨论长期聚焦于个体自主：知情同意、隐私、可解释性。通谕认为这种框架系统性地忽略了团结原则——我们对彼此、对后代、对整个人类社群的责任。
当 OpenAI 用「我们让用户选择退出」来回应隐私争议时，它在「自主」框架内是合理的。但在「团结」框架内，问题变成了：这个系统是否增进了人类社群的共同福祉？答案可能完全不同。
这就是为什么一份来自梵蒂冈的文件值得 AI 行业认真对待——不是因为它的政策建议多么新颖，而是因为它可能改变我们用来评估 AI 的整个价值坐标系。
> ✨ 金句
> 「技术从不中立，因为它带有设计者、资助者、监管者和使用者的特征。每一个设计选择都反映了一种人性观。」—— 教皇利奥十四世，《壮丽的人性》，2026 年 5 月
---
信息来源：Vatican.va 官方文本（2026-05-25）、Time、CNN、NPR、National Catholic Reporter、Anthropic 官方声明
> 📌 TL;DR
> On May 25, 2026, Pope Leo XIV released his first encyclical, Magnifica Humanitas (Magnificent Humanity) — a 42,300-word document that is the Catholic Church's first formal teaching on artificial intelligence. The core argument: AI must be "disarmed" — not paused technologically, but liberated from the logic of arms races and monopolistic control to serve humanity's common good.
---
A Vatican White Paper on AI Policy
On May 25, an unusual pair took the stage at the Vatican's Synod Hall: Pope Leo XIV and Anthropic co-founder Christopher Olah.
One is the spiritual leader of 1.4 billion Catholics worldwide. The other is the AI scientist who built the field of mechanistic interpretability. Together, they released a document that may reshape how we frame the AI ethics debate — Magnifica Humanitas: On Safeguarding the Human Person in the Time of Artificial Intelligence.
This is not a vague moral appeal. The 235-page, 245-paragraph encyclical systematically applies Catholic Social Teaching (from Rerum Novarum in 1891 through Laudato Si' in 2015) to every dimension of AI governance: labor rights, autonomous weapons, data monopolies, algorithmic discrimination, and human dignity.
Five Core Arguments
1. "Disarm AI"
The central thesis from paragraph 110:
> "It is not enough to regulate AI: it must be disarmed and made accessible."
"Disarming" here doesn't mean a technological pause. It means removing AI from the logic of geopolitical arms races and commercial zero-sum games. The Pope argues that current AI development is driven by an "armed competition culture" — nation-state rivalry and corporate winner-takes-all dynamics — where safety and ethics will always rank behind speed and profit.
2. An Absolute Ban on Lethal Autonomous Decisions
> "It is not permissible to entrust lethal or otherwise irreversible decisions to artificial systems."
The encyclical sets three non-negotiable requirements for autonomous weapons:
- Traceability of decisions
- Meaningful human control over lethal action
- International rules capable of slowing the technological arms race
The Pope argues AI weapons are "normalizing war" because they allow armies to attack without seeing their victims' faces, thereby "lowering the moral threshold of conflict."
3. AI as the New Labor Question
Chapter 4 — the longest section — frames AI as a new labor question, just as Leo XIII faced the industrial revolution 135 years ago with Rerum Novarum.
The document names specific realities:
- Data labelers and content moderators exposed to violent imagery for poverty wages
- Children mining rare earth materials in dangerous conditions
- AI's capacity to deskill workers, subject them to automated surveillance, and fragment their tasks
The Pope endorses state regulation, labor unions, and "social criteria for innovation."
4. Data and Algorithms Belong to All Humanity
The encyclical's most legally disruptive claim: extending the principle of "universal destination of goods" to AI systems — data, algorithms, and platforms cannot be treated as the absolute property of those who control them.
> "It means freeing technology from monopolistic control and opening it to discussion and debate."
This directly challenges Big Tech's business model: user data collected without fair compensation, model weights treated as trade secrets, platform rules set by the few.
5. Technology Is Never Neutral
> "Technology is never neutral, because it takes on the characteristics of those who devise, finance, regulate, and use it."
> "AI developers bear a particular ethical and spiritual responsibility, for every design choice reflects a vision of humanity."
This directly counters Silicon Valley's long-standing "technology is just a tool" narrative. The encyclical argues this neutrality claim is itself an ideology — one that evades responsibility by denying it.
Why Was Anthropic Standing Beside the Pope?
Christopher Olah's presence was no coincidence. Anthropic is currently the AI company most directly at odds with the Trump administration — blacklisted by the Pentagon for refusing to let Claude be used for autonomous weapons and mass domestic surveillance, and suing the administration over it.
Olah's Vatican remarks were strikingly candid:
> "Computer scientists alone cannot determine the ethical boundaries of AI because developers themselves are influenced by incentives — ambition, competition, and financial pressure. We need informed critics who will tell the labs when we are failing."
This is the co-founder of a company valued at nearly $900 billion, publicly acknowledging that industry self-regulation is insufficient. In the current Silicon Valley atmosphere — OpenAI rushing its IPO, Meta mass-firing employees to pivot to AI, Google stuffing AI into everything — this stance is remarkable.
Political Dimensions: The Pope vs. Trump
The encyclical's timing is pointed. Just days earlier, President Trump abruptly cancelled the signing ceremony for an AI executive order, saying he didn't want to "get in the way of" America's tech lead over China.
The Pope's response is systematic. Across multiple chapters, the encyclical directly counters the Trump administration's AI policy orientation:
| Issue | Trump Administration | Papal Encyclical |
|-------|---------------------|-----------------|
| AI Regulation | Rolled back Biden-era safety rules | Demands stricter pre-deployment testing |
| Autonomous Weapons | Expanding military AI | Absolute ban on lethal autonomous decisions |
| Industry Relations | Embracing Big Tech | Criticizing monopolies, demanding openness |
| Labor Protection | Market liberalization | Supporting unions, state intervention |
Vice President J.D. Vance (a Catholic) told NBC News the encyclical was "very profound," acknowledging that AI "raises such profound questions for how we interact with one another... I think we really need moral leadership to think through those questions."
A Historical Moment
Leo XIV's choice of papal name is itself a declaration — honoring Leo XIII, who published Rerum Novarum in 1891. That document laid the foundation for Catholic Social Teaching's response to industrial capitalism, influencing over a century of labor law and welfare systems.
He deliberately signed the encyclical on May 15 — the 135th anniversary of Rerum Novarum's release.
The structural parallels are striking:
- Rerum Novarum faced steam engines and assembly lines; Magnifica Humanitas faces large language models and autonomous systems
- Rerum Novarum addressed factory workers' dignity; Magnifica Humanitas addresses the dignity of data labelers and content moderators
- Rerum Novarum called for state intervention in laissez-faire markets; Magnifica Humanitas calls for international mechanisms to constrain the tech arms race
What This Means for the AI Industry
You might ask: can a papal document actually influence AI policy?
The answer: limited direct impact, but enormous indirect influence.
Institutional level: The Vatican has established a cross-department AI commission. The U.S. Conference of Catholic Bishops is expected to implement the document's teachings across dioceses. The ethical framework for 1.4 billion Catholics is being re-anchored.
Narrative level: The encyclical provides the "AI deceleration" camp with a powerful moral argument framework. Previously, AI safety discussions were largely confined to technical elite circles — now they have a communication channel reaching billions of ordinary people.
Policy level: The EU AI Act is already being enforced but lacks a unified ethical narrative. Magnifica Humanitas may fill this gap — particularly its argument that the "universal destination of goods" applies to data and algorithms, potentially providing theoretical foundations for future legislation.
My Take
The encyclical's most powerful contribution isn't its specific policy recommendations — those have been thoroughly discussed in AI ethics academia — but its reframing of the entire debate:
From "autonomy" to "solidarity." Western AI ethics discourse has long focused on individual autonomy: informed consent, privacy, explainability. The encyclical argues this framework systematically neglects the principle of solidarity — our responsibilities to each other, to future generations, to the entire human community.
When OpenAI responds to privacy concerns with "we let users opt out," it's reasonable within the "autonomy" framework. But within a "solidarity" framework, the question becomes: does this system advance the common good of the human community? The answer might be entirely different.
This is why a Vatican document deserves the AI industry's serious attention — not because its policy recommendations are novel, but because it may shift the entire value coordinate system we use to evaluate AI.
> ✨ Key Quote
> "Technology is never neutral, because it takes on the characteristics of those who devise, finance, regulate, and use it. Every design choice reflects a vision of humanity." — Pope Leo XIV, Magnifica Humanitas, May 2026
---
Sources: Vatican.va official text (2026-05-25), Time, CNN, NPR, National Catholic Reporter, Anthropic official statement