Chinese Seeking To Use AI Disinformation Database for Cognitive Defense

Rumor Crusher.

Rumor Crusher.

“The models built by AI still need further language training. For example, some users’ share [rumors] with irony, sarcasm, or as a metaphor, and machines find it difficult to judge whether they are true expressions of emotion or not. Therefore, we should focus on improving the public’s media literacy so that they are more serious about interacting and sharing content ….”

According to the excerpted article from PLA-owned strategic communications journal Military Correspondent, the PLA is exploring an early warning mechanism for monitoring and combating digital disinformation utilizing the Cyberspace Administration of China (CAC)’s[i] “AI Rumor Crusher.”[ii] While the article is not a definitive accounting of how the CAC will use artificial intelligence (AI) to monitor and counter digital disinformation, it does provide insight into a methodology employed across China’s information environment to control flows of information. The author, Li Beibei, a researcher at China’s National Defense University’s College of Politics, suggests that an ideal approach for countering disinformation would include a multipronged strategy that includes expansion of China’s legal framework, improvement of data-sharing among social media platforms and the government, and improved media literacy.

Li explains AI Rumor Crusher’s operational methodology as an eight-step iterative process composed of recognition, word separation, comparison with a rumor database, determination of credibility, analysis of rumor and non-rumor features, and finally supervised and reinforcement learning. In the first two steps, AI Rumor Crusher identifies the source of a piece of information and analyzes the disseminator. Next, it analyzes the information against preexisting rumor samples and tracks sources for their credibility (website, publisher, professionals, etc.). Finally, the key arguments are labeled and cross-referenced with authoritative knowledge databases[iii] to verify the veracity of the information. Li argues that AI, when compared to human counterparts, possesses a superior ability to work around the clock, identify and eliminate rumors in a timely way, track disseminators of disinformation, judge the veracity of information, and even determine the disseminator’s motivation for sharing rumors. Li further advocates for a multipronged approach to digital disinformation governance that would include the expansion and strengthening of China’s relevant legal regulations and public media literacy to reinforce social values and ensure both originators and disseminators (witting or unwitting) do not exacerbate the issue.[iv]


“利用人工智能技术治理网络谣言探析 (An Exploration of the Uses of Artificial Intelligence in Governing Digital Rumors),” Military Correspondent (PLA-owned strategic communications journal ), 23 February 2023.

Management of digital rumors has become an urgent issue. We can establish an early warning mechanism for digital rumors based in artificial intelligence technologies. With AI “Rumor Crusher” we can build a data-sharing platform for disinformation to monitor the trajectory of digital rumors and prevent them from spreading in order to effectively combat digital disinformation.

With the promotion of new media interactive platforms, the difficulty of Internet rumor management is gradually increasing, and users’ behavior and speech on the Internet are difficult to be effectively and timely restrained, making rumor suppression a more complicated and difficult task.

The most central factor in the formation and dissemination of digital rumors is the openness and ambiguity of incoming information. People are most likely to generate and spread rumors in the absence of reliable information. … people use their virtual identities to send and receive information over their various accounts without temporal, spatial, or moral constraints … resulting in many online rumors leading to serious consequences. Since August 2020, the Cyberspace Administration of China (CAC)’s Reporting Center has organized more than 10 website platforms such as Weibo, Douyin and Baidu to carry out digital rumor tagging and labeling work. This ensures that digital rumor samples are promptly exposed, thereby minimizing the space for digital rumors to survive. However, despite the active measures relevant government departments have taken to control digital rumors, there is still a long way to go. In particular, the legal punishment for digital rumors should be strengthened so the disseminators of rumors can be punished according to the degree to which the rumor is malign and causes a negative impact.


[i] Cyberspace Administration of China’s Reporting Center (网信办举报中心, literally, “Illegal and Harmful Information Reporting Center, 违法和不良信息举报中心). As early as 2004, under the Internet Society of China, the China Internet Illegal Information Reporting Center was established with the stated goal of standing up for “virtue and right thinking” while opposing pornography, violence, and fraud. See: Sumner Lemon, “Chinese website lets users report illegal content,” Computer World, 15 June 2004.

[ii] “Rumor (谣言)” is both a colloquial and legal term in the PRC. Colloquially, it connotes a similar meaning to the term “rumor” in English. According to the Cyberspace Administration of China (国家互联网信息办公室,简称:网信办), internet rumors网络谣言 refers to “false information that intentionally fabricates facts to cause harm to society and others.” This definition partly aligns with Western conceptions of disinformation (often translated as 虚假信息, lit. fake information). For further exploration of the PRC’s unclear use of similar terms in legal matters, see: 谣言型犯罪中“谣言”该如何理解 (How Should “Rumor” be Understood in Rumor Crimes), The Supreme People’s Procuratorate of the People’s Republic of China, 12 February 2022.

[iii] While not explicitly stated, these would likely include both open source, commercial, and classified government databases accessible by government, military, and commercial personnel. Much of the labeling and tagging until recently has been conducted by human regulators working on behalf of social media companies like Alibaba, Baidu, ByteDance, and Tencent.

[iv] Many of these same issues are addressed in China’s preexisting legal infrastructure managing cyberspace including the Cybersecurity Law, Data Security Law, Personal Information Protection Law, and Cybersecurity Review Measures. For more on China’s vision for global cyberspace governance, see: Thomas Shrimpton, “Beijing’s Vision for Global Cyberspace Governance,” OE Watch, 1-2023.

Image Information:

Image: Rumor Crusher.
Attribution: Public Domain