Why a Reported China Probe Hit AI Stocks—and Complicated US-China AI Talks

A report that Chinese regulators were investigating DeepSeek and Moonshot AI over data-security concerns triggered a sharp sell-off in Hong Kong-listed AI shares. The reported inquiry followed Anthropic’s allegations that the two companies routed sensitive data through its Claude models; those allegations were not, by themselves, proof of a regulatory violation. 12

How Hong Kong AI shares reacted

Z.AI shares fell as much as 12% and MiniMax Group dropped as much as 6.8% after the report. Alibaba, which develops the Qwen model, also slid nearly 5%. These were reported intraday declines, not necessarily closing losses. 1

The inquiry reportedly focused on DeepSeek and Moonshot: Bloomberg, citing The Information, said China’s Cyberspace Administration sent officials to interview executives and employees to assess whether breaches occurred and how serious they might be. 24 But the market reaction reached other AI-linked shares as investors weighed the prospect of broader regulatory scrutiny. 17

That reaction should be distinguished from a formal finding. The cited reporting described an investigation and allegations; it did not establish that the companies had breached data-security rules. 215

Why the issue reached beyond data security

The allegations raised two connected but distinct concerns: whether sensitive user information could be sent to a foreign AI service, and whether Chinese developers used outputs from US models to improve their own systems. Anthropic described the latter as unauthorized “distillation”; US agencies also accused Chinese AI companies of copying capabilities from leading US models. These are claims by the companies and agencies, not findings established by the reported Chinese inquiry. 10

Together, the disputes made AI governance a point of US-China friction. They also gave investors reason to consider whether scrutiny of model development and data handling could affect companies beyond the two named in the reported probe. 7

What the report meant for Trump-Xi AI talks

The timing placed the probe against a backdrop of discussions over AI competition and safety. The reporting available here does not establish that the probe produced an AI-industry agreement between Trump and Xi. Subsequent coverage said the leaders’ summit ended without a major agreement, while critics argued it missed an opportunity to seek a pause in the AI arms race. 39

The episode therefore highlighted the difficulty of separating cooperation on AI risks from wider disputes over competition, model development and data security. The market sell-off showed that even a reported inquiry—before any public determination of wrongdoing—could add uncertainty for companies and investors across the sector. 12