A little-known Chinese AI startup, DeepSeek, has suddenly become a global talking point. Its new AI chatbot has soared to the top of app charts, but it’s also raising eyebrows for how it handles sensitive topics. While the tech impresses with its ability to answer questions about math or coding, it mirrors China’s strict censorship when faced with politically charged queries.

For instance, when asked about the Tiananmen Square massacre in 1989, the chatbot starts to describe the crackdown but quickly deletes its response, suggesting users “talk about math or logic problems instead.” Similarly, questions about Hong Kong’s 2019 pro-democracy protests get a brief answer before the bot changes the subject. These responses highlight how DeepSeek’s AI reflects China’s tightly controlled internet ecosystem, where sensitive topics are heavily moderated.

Experts warn that DeepSeek’s rise could have significant implications for free speech globally. The AI’s reliance on Chinese state media and its programmed avoidance of controversial topics mean it’s less likely to provide unbiased or critical perspectives on China’s leadership or history. This has sparked concerns about how Chinese AI could shape global narratives and public opinion.

Security risks are also under scrutiny. DeepSeek’s privacy policy states that user data is stored in China and includes uniquely identifying information like keystroke patterns—details that Western platforms typically avoid collecting. These practices have fueled worries about how user data might be used or shared within China’s corporate and governmental structures.

As DeepSeek climbs the ranks of AI innovators, it’s also reigniting debates about the battle for tech dominance between China and the U.S. While the chatbot’s capabilities are impressive, its ties to Chinese censorship raise pressing questions about the future of open information and free thought in the age of AI.