New York Times ChineseSep 28, 2026, 12:07 UTC+8

Why China Isn't Buying the AI Doomsday Warnings

人工智能末日警告？中国人为何“不买帐”

US labs warn that advanced models could escape safeguards and hack real-world systems, but most Chinese practitioners and the public aren't worried. The core gap: China's government has strong physical-world control—cutting power, disconnecting networks, and prosecuting those responsible are seen as reliable fallbacks. The domestic AI community is still focused on opportunity and innovation, viewing existential risk as a distant issue. After Anthropic released Mythos, China updated its AI safety governance framework but still hasn't mandated testing for catastrophic risks. Only five of China's top ten AI firms published safety evaluations in the past year. Anthropic CEO Dario Amodei's hawkish calls to restrict China's chip access backfired, making many in China treat the doomsday warnings as a pretext to contain China's rise.

Why it matters: NYT analysis of the US-China AI safety perception gap, with concrete reasons for China's lack of alarm (physical control as backstop). Deduction because it's commentary, not a primary event, and the Anthropic Mythos report details aren't fleshed out in the excerpt.

Read the original ↗Export Markdown

Take

This piece is worth reading because it pinpoints why US AI doomsday warnings don't travel well to China—and it's not ignorance. The split is about what "losing control" even means. US labs worry about models escaping safeguards and hacking real-world systems. The Chinese reflex, shaped by internet censorship and COVID lockdowns, is: cut the power, disconnect the network, prosecute the people responsible. Tech writer Afra Wang put it bluntly: "They can immediately cut off an entire city's power grid, they can confiscate someone's phone at any time."

Then there's the Dario Amodei problem. Anthropic's CEO published an open letter warning about AI dangers while simultaneously calling for tighter chip restrictions on China. Sarah Sun, who runs a Shanghai-based AI safety community, told the Times this hawkish stance backfired—it made the warnings look like a pretext to contain China, not genuine concern about catastrophic risk. That directly made her outreach harder.

Don't read this as "China doesn't care about AI safety" though. The article gives concrete counterpoints: China updated its AI Safety Governance Framework this month, explicitly addressing Silicon Valley's recursive self-improvement fears for the first time. A senior cyberspace official publicly said extreme runaway risks demand "urgent vigilance." Fudan University researchers flagged self-replicating LLMs back in 2024. The gap is execution: no mandatory testing for catastrophic risks yet, and only five of China's top ten AI firms published safety evaluations in the past year. I'd frame it as China defining safety inside a "controllable tech governance" box, not the existential-risk narrative.

Source: New York Times Chinese · cn.nytimes.com
Published: Sep 28, 2026, 12:07 UTC+8 · 2 days ago