New York Times ChineseJun 2, 2026, 11:37 UTC+8

China Is Trying to Use AI to Predict Dissent

中国尝试利用人工智能预测异见倾向

Geedge is developing an AI system to predict dissent using telecom, social media, and location data, according to 100,000 leaked documents reviewed by Vanderbilt researchers; U.S. officials say there is no evidence that the predictive technology has been finalized or deployed.

Why it matters: HKR-H/K/R all pass: the NYT report adds leaked-file evidence, data-source detail, and a clear surveillance-governance nerve. Deployment is unconfirmed, so this stays in the 78–84 band rather than P1.

Read the original ↗Export Markdown

Take

Geedge’s threat is not sci-fi prediction; it is old surveillance infrastructure wired to scalable classifiers. The evidence is concrete: 100,000 leaked documents show MESA Lab working in early 2024 on profiles built from telecom, social media, and location data, with meeting notes about “identifying intent.” U.S. officials also say there is no evidence the predictive system has been finalized or deployed.

I don’t buy the headline version that AI can already forecast dissent. The practical version is harsher: police systems use models like DeepSeek to triage huge alert streams and move human review earlier. Geedge hitting GPU limits and older models says export controls bite less on chatbots than on scaled multimodal surveillance. The missing number is false-positive rate, and that is the number a system like this would hide first.

Source: New York Times Chinese · cn.nytimes.com
Published: Jun 2, 2026, 11:37 UTC+8 · Jun 2, 2026