[Today’s Signal] After GPUs, APIs — U.S.-China AI Rivalry Spreads to the Data Frontier

The U.S.-China AI rivalry has moved beyond GPU access to data sovereignty. China's investigation of DeepSeek and Moonshot AI shifts the focus. Regulators are concerned about sensitive Chinese data being sent to a U.S. AI model, not just model distillation. This reflects a new phase of regulatory scrutiny on both sides of the Pacific.

Anthropic's report alleged Moonshot AI forwarded user requests to Claude, including internal code and credentials from a Chinese state-owned enterprise. DeepSeek also allegedly routed internal documentation from a Chinese tech company and requests for a public security bureau's system. This directly impacts Chinese data protection laws. The Cyberspace Administration of China's focus on data-routing practices indicates a tightening of domestic data controls.

The key question for Asia is how Chinese regulators will define "unlawful" data transfer in this context. Model distillation itself is not illegal. The test for the CAC is whether companies like DeepSeek and Moonshot AI circumvented access rules or transferred customer data without consent. This will set a precedent for all Chinese AI developers using foreign models.