China uses AI to tighten Uyghur surveillance, report says

BEIJING, China (MNTV) — China is rapidly developing artificial intelligence systems designed to monitor Uyghurs and other ethnic minority communities, raising concerns about state-driven digital repression and the potential export of these tools abroad, according to a report by the Australian Strategic Policy Institute (ASPI).

The report details Beijing’s creation of large language model-based public-opinion surveillance tools capable of analyzing text, video, audio, and even emojis in Uyghur, Tibetan, Mongolian, and Korean.

These systems could vastly expand the state’s ability to police online speech, track dissent, and tighten political control in heavily securitized regions.

Titled “The Party’s AI: How China’s new AI systems are reshaping human rights”, the report identifies Minzu University of China as a key research hub. Its National Key Laboratory of Ethnic Language Intelligent Analysis and Security Governance, funded by the Ministry of Education, states its mission is to “maintain national stability and ethnic unity.”

ASPI notes that researchers collect online data from minority regions, building real-time public-opinion monitoring systems and extensive databases in over 10 minority languages.

Minority languages, previously challenging for China’s censorship tools optimized for Mandarin, can now be monitored with AI, effectively closing gaps.

The report highlights implications for Xinjiang’s 12 million Uyghurs, already subjected to mass surveillance, detention, forced cultural assimilation, and practices described as potential crimes against humanity.

The AI systems could further criminalize everyday religious and cultural expressions. Similar surveillance extends to Tibetans, Mongolians, and ethnic Koreans, reflecting a broader architecture of digital authoritarianism.

The study warns these AI models could be exported through the Belt and Road Initiative (BRI), targeting diaspora and minority communities abroad. China’s major tech firms, including Tencent, Baidu, and ByteDance, act as “deputy sheriffs” for state surveillance, integrating AI censorship into social platforms and messaging apps.

Tencent’s WeChat, for instance, now applies risk scores to private messages, while Baidu markets AI censorship commercially.

ASPI researchers found AI models refusing politically sensitive content, often issuing censorship notices or critical state-aligned commentary. The system still relies on human moderators fluent in minority and foreign languages, forming a hybrid model combining AI and manual oversight.

Human rights groups warn that these developments establish a dangerous precedent: AI deployed for authoritarian control, not innovation, with potentially global consequences as Chinese surveillance technology spreads across emerging markets.