According to a February 2019 publication by the
Center for a New American Security,
CCP general secretary Xi Jinping – believes that being at the forefront of AI technology will be critical to the future of global military and economic power competition. By 2025, the State Council aims for China to make fundamental contributions to basic AI theory and to solidify its place as a global leader in AI research. Further, the State Council aims for AI to become "the main driving force for China's industrial upgrading and economic transformation" by this time. The CCP's
fourteenth five-year plan reaffirmed AI as a top research priority and ranks AI first among "frontier industries" that the Chinese government aims to focus on through 2035. Due to security concerns around strategically sensitive economic sectors, the government dissuades executives at Chinese AI companies to travel to the U.S. and, if required to travel, to brief authorities before and after travel.
Research and development Chinese public AI funding mainly focused on advanced and applied research. The government funding also supported multiple AI R&D in the private sector through venture capitals that are backed by the state. According to national guidance on developing China's
high-tech industrial development zones by the Ministry of Science and Technology, there are fourteen cities and one county selected as an experimental development zone.
Zhejiang and
Guangdong provinces have the most AI innovation in experimental areas. However, the focus of AI R&D varied depending on cities and local industrial development and ecosystem. For instance,
Suzhou, a city with a longstanding strong manufacturing industry, heavily focuses on automation and AI infrastructure while
Wuhan focuses more on AI implementations and the education sector. Many of these systems are now being integrated into China's
domestic surveillance network. Interdisciplinary collaborations play an essential role in China's AI R&D, including academic-corporate collaboration, public-private collaborations, and international collaborations and projects with corporate-government partnerships are the most common. Besides, according to an AI index report, China surpassed the U.S. in 2020 in the total number of global AI-related
journal citations. As of 2023, 47% of the world's top AI researchers had completed their undergraduate studies in China. In 2025,
The Washington Post reported that China has surpassed the United States in the release of publicly available "open" artificial intelligence model. In April 2026, Chinese regulators blocked Meta Platforms’ $2 billion acquisition of AI startup Manus, citing national security and export control concerns. Although Manus was registered in Singapore, its core technology originated in China. The move reflects China's tighter oversight of AI and strategic technologies, particularly regarding foreign involvement.
Population China's large population generates a massive amount of accessible data for companies and researchers, which offers a crucial advantage in the race of big data. , China has the world's largest number of internet users, generating huge amounts of data for machine learning and AI applications.
Facial recognition Facial recognition is one of the most widely employed AI applications in China. Collecting these large amounts of data from its residents helps further train and expand AI capabilities. China's market is not only conducive and valuable for corporations to further AI R&D but also offers tremendous economic potential attracting both international and domestic firms to join the AI market. The drastic development of the
information and communication technology (ICT) industry and AI chipsets in recent years are two examples of this. China has become the world's largest exporter of facial recognition technology, according to a January 2023
Wired report.
Censorship and content controls In April 2023, the Cyberspace Administration of China (CAC) issued draft measures stating that tech companies will be obligated to ensure AI-generated content upholds the
ideology of the CCP including
Core Socialist Values, avoids discrimination, respects
intellectual property rights, and safeguards user data. Before releasing a large language model to the public, companies must seek approval from the CAC to certify that the model refuses to answer certain questions relating to political ideology and criticism of the CCP. Prompts related to politically sensitive topics such as the
1989 Tiananmen Square protests and massacre or
comparisons between Xi Jinping and Winnie the Pooh must be declined.
Microsoft has warned that the Chinese government uses generative artificial intelligence to interfere in foreign elections by spreading
disinformation and provoking discussions on divisive political issues. The Chinese artificial intelligence model
DeepSeek has been reported to refuse to answer questions relating to things about the 1989 Tiananmen Square protests and massacre,
persecution of Uyghurs, comparisons between Xi Jinping and Winnie the Pooh or
human rights in China. According to research outside China, AI regulations have had the effect of Chinese AI models performing better on AI safety and suicide prevention metrics such as producing content with less violence and pornography, as well as less information that leads users toward self-harm. However, they also had less constraints against "
jailbreaking". In November 2025, the Chinese government finalized a set of rules regarding training data. It states that human testers from companies should randomly evaluate 4,000 pieces of training data for each format of content their AI can handle, including text, video and images, and not use a given source unless at least 96% of the material is deemed safe. The document defines unsafe in 31 categories, including data that includes "incitement to subvert state power and overthrow the socialist system", as well as sources that promote violence, false information or discrimination, and content that uses someone's likeness without permission. Prompts designed to trigger responses leading to subversion of state power or discrimination must be declined at a rate of 95%. The regulations state that chatbots should be tested with 2,000 questions and regular updates of the questions at least once a month. Regulations also require chatbots to implement a
Internet real-name system, requiring users to register with a phone number or national ID. == Leading companies ==