A new report by the Australian Strategic Policy Institute (ASPI) warns that China is rapidly converting artificial intelligence into a precision instrument of political control reshaping censorship, policing, surveillance and even economic influence overseas.
ASPI finds that Chinese large language models now censor politically sensitive images, not just text. Using a dataset of 200 images, researchers discovered multiple layers of embedded visual censorship across China’s LLM ecosystem, an area largely overlooked until now. This sits alongside AI systems that already scan and delete online content within seconds, forming the backbone of China’s industrialised censorship regime.
AI Across the Justice and Surveillance Systems
The report shows AI spreading across the entire criminal-justice pipeline, from predictive policing and vast biometric surveillance networks to AI-assisted courts and prisons. Judges are being encouraged to use AI not only for paperwork but for recommending verdicts and sentences trends ASPI says risk deepening discrimination and eroding accountability. Companies such as iFlyTek feature prominently as major suppliers of these systems.
Monitoring Minority and Diaspora Communities
Another revealing strand of ASPI’s research is China’s development of minority-language LLMs in Uyghur, Tibetan, Mongolian and Korean. Unlike commercial models, these are state-funded tools designed to monitor public sentiment and communications in minority languages. ASPI notes that Beijing is also exploring their use to track diaspora communities across Belt and Road countries.
The report highlights how censorship laws have created a thriving domestic market for AI-enabled content-control tools. Tech giants including Tencent, Baidu and ByteDance now sell turnkey censorship platforms to smaller firms, making compliance cheaper and more automated further embedding state priorities into China’s digital economy.
ASPI also warns that AI is amplifying China’s economic reach abroad, particularly in sectors involving vulnerable communities. Chinese fishing fleets have begun using AI-powered “intelligent fishing” systems that give them a significant technological advantage. ASPI identifies vessels operating in sensitive waters, including Mauritania and Vanuatu, where Chinese fleets have been linked to illegal activity.
With China already the world’s largest exporter of AI surveillance technology, ASPI concludes that these developments will not remain confined within its borders. Instead, they risk globalising a model of AI-driven repression that shapes behaviour, information flows and economic outcomes far beyond China.
Research Associate at StratNewsGlobal, A keen observer of #China and Foreign Affairs. Writer, Weibo Trends, Analyst.
Twitter: @resham_sng
