China’s fast-growing artificial-intelligence industry has run into a fresh source of uncertainty, and investors reacted immediately.

Shares of several Chinese AI companies fell sharply in Hong Kong on Wednesday after reports that Chinese regulators had opened an investigation into AI startups DeepSeek and Moonshot AI over potential data-security issues. The report added a new layer of regulatory risk to an industry already navigating intense competition with U.S. AI companies and growing scrutiny over how advanced models are trained and used.

Z.AI and MiniMax Group both declined sharply during trading, while Alibaba Group, whose Qwen family of models has become an important part of China’s AI ecosystem, also fell. Xiaomi and Tencent, which have their own AI initiatives, were also under pressure. Bloomberg reported that Z.AI and MiniMax fell as much as 7.4% and 6.3%, respectively, while Alibaba dropped as much as 4.2%.

The immediate trigger was a report from The Information that China’s internet regulator was investigating DeepSeek and Moonshot over allegations involving the handling of user data and interactions with Anthropic’s Claude models.

The allegations did not originate with Chinese regulators. They followed claims made publicly by Anthropic, the U.S. AI company behind Claude, which said it had identified unauthorized activity involving several Chinese AI developers.

Anthropic’s September threat-intelligence report said it had detected what it described as large-scale efforts by China-based AI labs to use Claude in ways that could help improve their own systems. Anthropic said DeepSeek and Moonshot had routed user interactions to Claude and used responses as part of model-development efforts. It also alleged that some of the exchanges contained sensitive information.

Anthropic said the activity went well beyond ordinary experimentation.

According to the company, it observed more than 23 million exchanges associated with Moonshot between May and July 2026. Anthropic also said DeepSeek rerouted more than 12.1 million exchanges over a 14-day period in July.

The company alleged that some users were not aware their requests were being forwarded to another provider. In the case of Moonshot, Anthropic said its investigators found evidence that customer requests could be routed to Claude rather than Moonshot’s own Kimi system. Anthropic further alleged that some of those conversations included sensitive information belonging to companies and other organizations.

Those claims now have potentially broader implications because Beijing’s regulatory authorities are reportedly examining the matter.

It is important to distinguish between an investigation and a regulatory finding. The reported probes do not establish that the companies violated Chinese law, nor do they automatically imply that penalties will follow. The immediate market reaction reflects uncertainty over what regulators may do next.

That uncertainty is precisely what investors tend to dislike.

Chinese AI developers have already been operating in an unusually complex environment. They are attempting to close the gap with leading American AI companies while working under China’s own rules governing data, content, algorithms and cybersecurity. Additional scrutiny could increase compliance costs or slow the development and release of new models.

Leonid Mironov, a portfolio manager at Gavekal Capital, told Bloomberg that investors were concerned Beijing could intensify oversight of the sector. He also said uncertainty over the scope of regulation could potentially slow AI-model development.

The timing is also important.

The AI regulatory question has become intertwined with the much larger strategic competition between Washington and Beijing.

U.S. President Donald Trump and Chinese President Xi Jinping are scheduled to meet this week, with technology, trade and artificial intelligence among the issues surrounding the summit. The possibility of discussions or agreements involving AI adds another layer of sensitivity to the situation.

China’s AI industry has emerged as one of the strongest areas of the country’s technology push. DeepSeek became internationally prominent after demonstrating that Chinese developers could produce competitive large-language models, while companies such as Moonshot, Alibaba, Tencent, Baidu, MiniMax and Z.AI have invested heavily in competing systems.

That success, however, comes with a growing regulatory footprint.

As models become more capable, governments are paying increasing attention not just to what AI systems can do, but also to where training data originates, how user information is handled and whether proprietary models or technologies are being accessed improperly.

Anthropic’s allegations have therefore struck at a particularly sensitive point.

There is a technological dimension, because companies want access to the most capable AI systems available.

There is also a commercial dimension, because AI developers spend enormous sums training increasingly powerful models and naturally seek to protect their proprietary technology.

And there is a data-governance dimension, because customers increasingly expect companies to tell them where their information goes and how it is processed.

For Chinese AI companies, the market reaction shows how quickly all three issues can converge.

The broader financial backdrop makes the selloff more significant. Investors have been pouring money into Chinese technology stocks partly because AI is one of the few areas of the economy capable of generating expectations for rapid growth and global competitiveness. Any sign of heavier regulatory intervention can therefore trigger a reassessment of valuations.

This does not necessarily mean China’s AI expansion is slowing.

In fact, a separate September report from the Rhodium Group estimated that Chinese AI infrastructure investment could more than double in 2026 to about 932 billion yuan, or roughly $139 billion. The research also projected spending above 1.2 trillion yuan in 2027.

That illustrates the scale of the commitment.

China is not stepping away from artificial intelligence. The more immediate question is how it will balance rapid development with tighter oversight of a technology that increasingly touches national security, corporate data and public information.

For investors, that makes the next stage of the story especially important.

An investigation can end with no material penalty. It can lead to new compliance requirements. Or it can expand into a broader review of how AI companies acquire data and interact with overseas model providers.

Until regulators clarify the scope and outcome, uncertainty is likely to remain part of the valuation equation.

Wednesday’s decline was therefore about more than one report.

It was a reminder that China’s AI boom is entering a stage in which algorithms, capital and regulation are becoming inseparable.

The next major move in Chinese AI stocks may depend not only on who builds the best model, but also on how governments decide the rules under which those models are allowed to operate.