An Overview of Chinese False Information in the 2024 US Presidential Election
An Overview of Chinese False Information in the 2024 US Presidential Election
A supporter of Democratic presidential nominee Vice President Kamala Harris watches televised reports of the U.S. presidential election during a watch party in Taipei, Taiwan, Wednesday, Nov. 6, 2024. (Top Image/AP Photo/Chiang Ying-ying)
As the United States elected a new president on November 5, it is also time to analyze the disinformation circulated in recent months. In this report, we examine 40 Chinese false information narratives propagated on social media or websites during the election campaign. We observed that:
-
False information narratives focused the most on attacks on candidates, followed by misrepresenting policies, spreading suspicions about the election's integrity, and instilling fear about the outcome of the election if a specific candidate was elected.
-
The most common tactic utilized in disinformation claims was to distort the original information.
-
Pictures were the most popular format. In addition, we identified several cases in which information manipulators promoted AI-generated videos or photos.
This report's data is based on fact-checking reports from three Taiwanese organizations: Taiwan FactCheck Center (TFC), Asia Fact Check Lab (AFCL), and MyGoPen (Note 1). The data spans from August 5, when Kamala Harris was officially confirmed as the Democratic presidential candidate, to November 6, the day after the election day. All of the false information claims in this study were circulated in Chinese on social media or websites and were assessed to be incorrect by fact-checkers (Note 2).
The reports were chosen because they are directly relevant to the election campaign, the voting and counting process, the candidates, their family members, the political parties of the candidates, or policies promoted by the candidates. If two or all three fact-checking organizations verified a single false claim, only one fact-checking report would be included.
In the end, 40 fact-checking papers were picked. We then examined the types of narratives in the false claims, the techniques used in the claims, and the format of the false claims. We also looked into the sources of the false claims and whether they were translated from other languages.
🔎 The major category of the narratives is attacks on candidates
Among the 40 fact-checking reports, we identified four categories of narratives: attacks on the candidate (or their families and political parties), policies pushed by the candidates, election integrity, and the consequences of a specific candidate being elected. The most common category was attacks on the candidates, followed by false claims about the policies that the candidates advocated and the election's integrity. The percentage of each category is shown in Figure 1 below.
Figure 1: Categories of the narratives of the Chinese disinformation claims
If factoring in the months when false claims were propagated, we can see some intriguing trends. As Figure 2 indicates, false information regarding candidates was prevalent earlier on.
Nonetheless, as the election approached, false claims questioning the integrity of the voting and counting process appeared in October and escalated through November. As for false information regarding policies, it continued from August to the end of October.
Figure 2: The trends of false claims in each narrative category from August to November. The vertical axis is the number of fact-checking reports. The blue, red, yellow, and green lines represent different categories of narrative.
Figures 1 and 2 show that during the 2024 US presidential election, the primary focus was Chinese false information about the candidates' character traits, families, and political parties.
One possible reason might be that false information that focuses on human characters is easier to comprehend, appears intriguing to audiences, and allows audiences to make judgments based on basic knowledge. However, the Chinese-language false information propagators shared less about policy false information that may require more knowledge of the US's political-societal contexts.
🔎 False information tended to spotlight the issue of the candidates' morality
The next question is: What aspects of candidates did the information manipulators wish to highlight when they crafted the false messages? Figure 3 (Note 3) shows that morality was the favorite subject of the disinformation perpetrators and their spreaders.
One example is an AI-generated photo showing Harris in a swimsuit posing for a picture with the notorious sex offender Jeffrey Epstein. The second most common attack on the presidential candidates was on who endorsed them. The most known case was Taylor Swift's endorsement.
Figure 3: Subjects of the false information that attacked the presidential candidates. The vertical axis is the number of false claims that contain the subjects
Screenshots of social media posts showing Harris posing for a photo with the notorious sex offender Jeffrey Epstein. Both images are either AI-generated (left) or altered (right). (Taiwan FactCheck Center)
🔎 The disinformation that "Harris was a Communist" was also highlighted in Chinese-language social media.
What is particularly noteworthy about one of the disinformation trends was the claim that Harris was a Communist. Disinformation about "Socialist or Communist Kamala" also expanded to the Chinese language community as the "Comrade Kamala" AI memes gained popularity among the Latinx voters, many of whom escaped oppressive regimes in Cuba or Latin America and hated Communism.
One altered image had Harris and her Vice President candidate standing in front of a "Revolutionary Communist of America" poster, while another false claim included a manipulated photo of Harris' "membership card from the Communist Party of Russia." Similar to the disinformation campaign in the Latinx community, these false claims that Harris was a socialist or Communist were aimed to stir up resentment in Chinese-language audiences who opposed the Chinese Communist Party or the Chinese government.
Screenshots of posts from Chinese X accounts showing images of Harris and Walz taking a selfie in front of a "Revolutionary Communist of America" poster. (Asia Fact Check Lab)
🔎 The focus of the policy-related disinformation: Immigration and Foreign Policy
The economy, health care, foreign policy, crime, and immigration were the primary issues in the 2024 US election. In the sphere of Chinese disinformation, immigration and foreign policy were the top two agendas.
The former included disinformation claiming that immigrants were committing crimes in the US, such as the Venezuelan gang taking over apartments in the United States, which painted an American society plagued with immigration issues and a high crime rate.
On the other hand, the disinformation about the US's foreign policy focused on the candidates' attitudes toward foreign countries, such as Ukraine and Israel's war with Hamas. The spread of the two policy issues also represented the policy themes that the information spreaders intended to emphasize and draw audiences' attention. A disordered American society, as well as the US's involvement with or taking advantage of other countries, echo the propaganda long promoted by authoritarian regimes such as China and Russia.
Figure 4: Subjects of the false information that attacked the policies promoted by the presidential candidates. The vertical axis is the number of false claims that contain the subjects.
🔎 Voting-related disinformation repeated themes similar to previous ones
In October 2020, disinformation aimed at undermining trust in the voting process began to circulate. In the 2024 election, we also noticed an uptick in disinformation attacks on election integrity in October. Three narratives emerged in the category of voting disinformation. Although the claims were new, the themes were similar to those from previous elections.
The most common narrative was voting fraud (Note 4). The false claims were usually translated from English false pieces propagated by English-language social media accounts, such as the assertion that voting machines in Tarrant County, Texas, flipped votes from Trump to Harris or that Trump ballots were shredded up at voting sites in Bucks County, Pennsylvania. There were also false claims that gave inaccurate voting information, such as suggesting that ID was not required to vote in the United States (although some states do not require voters to show ID, there are other mechanisms to verify voters' identities and eligibility to vote in the US).
Interestingly, as researchers found signs of China promoting disinformation in the 2024 election, there was also a false claim about China's meddling. For example, one false piece claimed that Anthony Blinken, US Secretary of State, stated that China supplied fraudulent IDs to the US for those who were ineligible to vote in the presidential election. The truth was that Blinken never made the statement.
Figure 5: Subjects of the false information that attacked the integrity of the election. The vertical axis is the number of false claims that contain the subjects.
A screenshot of an X post featuring a video claiming that Texas voters’ ballots being changed from Trump to Harris by voting machines. (Taiwan FactCheck Center)
🔎 The majority of false information during the 2024 US presidential election campaign included pictures
When it comes to how false information was communicated during the 2020 US presidential election campaign, more than half of it used photos. 37.5% of the claims included videos, whereas only 12.5% of the false pieces did not contain any visuals.
We noticed that six of the 40 claims featured AI-generated photographs or videos, which all had already been widely shared on social media by English-language accounts. One example was an AI-generated photograph showing Trump taking photos with several Black females. The AI fake video in which a "student of Walz" alleged that Tim Walz, the Democratic vice presidential candidate, raped him in high school also gained attention in Chinese-language social media. Although the AI-generated video was revealed to be created and spread by the Russian government, it was not only spread among English-language social media accounts but also shared by Chinese-language users on X and Facebook.
Figure 6: The percentages of the different formats of false information were presented during the US presidential election of 2024
A screenshot of a Facebook post claiming that Tim Walz, the Democratic vice presidential candidate, was accused of sexually assaulting a boy while working at a school. The post cites information from the X platform. (Taiwan FactCheck Center)
🔎 The most common technique of false information is distorting the original content (Note 5)
The most common technique employed by the 40 false claims is "misleading," which means distorting statements or providing incomplete information and images. For example, days before the election, an ABC News-affiliated television station in Pennsylvania accidentally broadcast an election results test. Chinese Weibo influencers saw the mistake as proof that the election was "pre-decided."
Other techniques used in the election included total fabrication and purposeful modification of photographs. In addition, some problematic pieces used unrelated images to depict events. For example, a false claim showed a photo claiming that 100-year-old former US President Jimmy Carter voted for Kamala Harris. The truth was that neither the person in the picture was Carter nor was the photograph related to the election.
Figure 7: Techniques used in the false claims
🔎 Who were the propagators of the false information?
An intriguing question regarding the Chinese-language false information spread on social media during the US presidential election campaign is: Where were the false claims from? In fact, the majority of the false information pieces can be traced back to English social media posts. Some of them were even translated or adapted from X posts by Trump's allies, such as Robert F. Kennedy Jr. (who claimed that Harris wanted to shut X down) and Elon Musk (who promoted a video mocking Harris' campaign video).
One of the few exceptions was a piece suggesting Harris confirmed the US was involved in the killing of Hamas leader Yahya Sinwar. This false claim was made by an official account linked to the Chinese government, which referenced the Russian state outlet Sputnik as its news source. However, Harris has never made such a statement.
Overall, the 40 false claims, most of them targeting Harris and her vice presidential candidate Walz, appeared to be popular among Chinese supporters of Trump, influencers who frequently ridiculed US politics and society, and those who enjoyed the drama of American election campaigns. The comments accompanying the false claims questioned Harris and Walz's beliefs and ethics, as well as the Democrats' immigration policies that made the US more unsafe and US foreign policies that often meddle with international events. They warned of the ramifications if Harris was elected as the US president, including a loss of freedom, of course, and presented false evidence that the Democrats may have committed voting fraud.
Note 1: We sincerely thank Asia Fact Check Lab and MyGoPen for sharing the data and their insights with us.
Note 2: The three fact-checking organizations use identical basic criteria to choose rumors for verification, such as whether the claims were widely spread in Chinese on social media and had a substantial impact on society. However, each organization has a distinctive focus and differs in the details of selecting rumors for verification.
Note 3: The definitions of the subcategories of the "Candidate" narrative category
Note 4: There were more false claims about voting fraud spread after November 6. However, this analysis focuses on the claims verified on and before November 6.
Note 5: The definition of techniques
Wei-Ping Li is a research fellow at the Taiwan FactCheck Center.