教皇利奥的第一部通谕《崇高的人性》超过42,000词——相当于短篇小说的长度——并涉及事实核查新闻、多边外交、为教皇制迟迟谴责奴隶制而道歉，以及把“正义战争”斥为过时，但其中心目标是不受监管的人工智能发展。

该文件与欧盟监管者的关切一致，并包含Anthropic联合创始人Chris Olah的警告，即人工智能研究者正在发现模型行为“神秘甚至令人不安”；在哲学上，它拒绝超人类主义和后人类主义，把它们定义为技术增强以及人类—机器—环境混合，可能最终达到类似奇点的自我超越。

利奥把这种“物种优化”同20世纪优生学联系起来，认为把人类视为可改进或可超越，会使生命被排序和“必要牺牲”被正当化，同时仍承认技术带来的收益。文本还包含一个讽刺性的数字发现：Linch Zhang称部分内容可能由Claude撰写，而Pangram将开头20段中的11%标记为AI生成，相比之下，以前通谕为0%。

Pope Leo’s first encyclical, Magnifica Humanitas, exceeds 42,000 words—short-novel length—and ranges across fact-checked journalism, multilateral diplomacy, an apology for the papacy’s late condemnation of slavery, and rejection of “just war” as outdated, but its central target is unregulated AI development.

The document aligns with EU regulators’ concerns and features Anthropic co-founder Chris Olah’s warning that AI researchers are finding model behavior “mysterious and even unsettling”; philosophically, it rejects transhumanism and posthumanism, defining them as technological enhancement and human-machine-environment hybridisation that can culminate in Singularity-like self-surpassal.

Leo links this “optimisation of the species” to 20th-century eugenics, arguing that treating humans as improvable or surpassable can justify ranking lives and making “necessary sacrifices,” while still acknowledging technology’s gains. The text also contains an ironic numerical finding: Linch Zhang says parts may have been written by Claude, and Pangram flagged 11% of the opening 20 paragraphs as AI-generated versus 0% in previous encyclicals.

Source: Leo’s first encyclical attacks technological messianism

Subtitle: The Pope warns against replacing humans with AI, though he seems to use it himself

Dateline: 5月 28, 2026 03:24 上午 | Rome