A tragic two-minute video spread over Chinese social media early this summer. A Chinese woman goes to university and falls in love with a Black student. She’s impressed when he tells her that he owns a mine and doesn’t break up with him when she finds out he bought her knockoff designer clothes. Eventually, she decides to move to his unspecified African homeland, despite her parents’ protests.
But when she gets there, she finds that he’s already married and lives in poverty. He burns her passport and forces her into hard labor and prostitution. She has a glimmer of hope when she meets a Chinese engineer and contacts her family. But when her parents come to the rescue, they’re unable to figure out how to get her home. They leave her behind, condemning her to a life of destitution with her mixed-race child. Oh, and her dad dies of a broken heart when he gets back to China.
The video was shared and liked thousands of times. The top comment on the video mocked the woman for being so educated but so foolish. The second called on the government to do more to screen out “low-quality foreigners.”
It’s not all doom and gloom, though. A more upbeat clip shows a laughing Chinese woman cradling a baby, sitting beside a Black man. This video was shared more than 40,000 times before being deleted by the creator. One of hundreds of furious comments bemoaned Chinese women “disrespecting their ancestors.” A comment that’s become common under videos is “Men guard the territory; women guard the bloodline.”
None of these videos are real—they’re artificial intelligence-generated fantasies. But their popularity and reception reveal a distressingly large audience in China for this kind of racist content.
Anti-Black videos are the most pervasive, but anti-Indian videos peddling strikingly similar tropes are also shared. These videos are easy to find on social media platforms, including WeChat; Kuaishou; and TikTok’s Chinese version, Douyin, despite China’s generally strict controls on content. Regulations introduced in 2023 even specifically ban using AI to create racist content. While most videos don’t reveal which AI tools were used to make them, the watermarks on a few show that they were made using domestically developed products such as Alibaba’s Qwen or ByteDance’s Doubao and Jimeng.
These videos come in three broad genres.
The most common is “Chinese woman follows her African partner home and has a bad time.” These videos universally depict Africa (never a specific country) as desperately poor, socially backward, and rife with physical and sexual violence.
The second type is rage-bait in the form of depictions of happy interracial couples—usually, but not always, Chinese women with African men.
The third type is “Chinese woman rejects a Black man.” A common scenario is a woman rejecting a man who asks for her phone number while riding the subway. Often, it’s specified that it’s the subway in Guangzhou—a city with a sizable African community that’s repeatedly been the focus of racist anxiety.
For instance, a politician from the other end of the country proposed a crackdown on the community in 2017, with the risk of China becoming a “Black and yellow mixed-race country” given as one of the justifications.
Of course, racist AI slop is far from unique to China. A report from Media Matters for America catalogued examples of AI-generated racist videos that went viral on TikTok. Those videos use tropes common in the United States—portraying Black people as criminals, Jewish people as money-obsessed, and East Asians as eating pets. Similar videos for British audiences often show Muslim migrants arriving on boats.
The videos that spread on Chinese social media reveal how racist tropes are also powerful in the country, despite priding itself on its long-standing friendship with African nations such as Tanzania and Zimbabwe.
Scholars trace the emergence of anti-Black ideas in China to the late 19th century. As intellectuals grappled with the country’s humiliation at the hands of Western imperialists, they imbibed imported social Darwinism and pseudoscientific racism, blending them with existing ideas about social hierarchies. This led to some intellectuals embracing the idea of the “yellow race” and positioning it as above other races in the global order.
This distinctively Chinese take on 19th century racism is typified by late Qing reformer Kang Youwei’s quixotic manifesto for achieving global unity. Kang wrote that the “black and brown” races will be virtually eliminated through competition with the “yellow and white.” While Kang suggested that the yellow and white races are similar and will eventually become unified through intermarriage, he wrote that this will be impossible to achieve with the remaining Black people due to their “extreme ugliness and stupidity.”
According to Cheng Yinghong of Delaware State University, the racially hierarchical ideas that emerged in Kang’s era have “never been fully exposed and negated.” He characterized the denunciations of racism by the People’s Republic of China “as part of an anti-Western ideological campaign rather than academic, cultural, historical, and psychological examination and self-reflection.”
One particular detail repeated ad nauseum in the AI videos is African and Chinese characters meeting at Chinese universities. The roots of this trope stretch back to the early 1960s, when China to a few African students who were meant to bring the revolution to their home countries.
After a lull during the Cultural Revolution, these soft power programs began again in earnest in the reform era. Conflict between Chinese and African students began to pick up steam at this time, with multiple confrontations serious enough to make it into the press. For instance, in 1986, Chinese students in Tianjin angered by loud music at a party besieged 40 African students for five hours until the police arrived.
The most well-known incident is the 1988 anti-African demonstrations. These began in Nanjing following a dispute between a security guard and African students trying to bring Chinese women into their dorms. Soon, thousands of students were marching in the streets, chanting racist slogans. Similar protests soon erupted in other cities, including Beijing, forcing African students to be evacuated for their safety.
Thankfully, there’s been no repeat of these incidents in recent years. But as the number of foreign students attending China’s universities has swollen to the hundreds of thousands, there’s been no shortage of online outrage. This can take many forms, including complaints over foreign students receiving better accommodation and other privileges, but interracial relationships are also a recurring theme.
An obvious trope in the AI videos is their broader gender dynamic—they virtually always represent a Chinese woman and a Black man. According to Sheng Zou, an assistant professor at Hong Kong Baptist University’s School of Communication, the women in the videos are “used as symbols for the vulnerability of a nation, but also symbolize national honor and sovereignty that needs to be defended, usually by men.”
This anxiety over Chinese women choosing foreign men is also likely a consequence of the country’s extreme gender disparity. After decades of the one-child policy and sex-selective abortions, China has some 35 million more men than women—and the financial requirements expected for marriage, such as owning a house or paying ever-higher dowries, are far more than many young men or their families can afford. Notably, AI-generated videos of fictional Russian women saying they want to marry Chinese men are generally met with positive comments.
Chinese universities and schools also discourage students from dating, adding an additional edge to racist anxieties with a campus setting. In 2019, Shandong University was forced to apologize after its “study buddy” program for international students sparked online uproar over the idea that the state was assigning Chinese girlfriends to foreign men. This particular moment seems to have made a lasting impression—often, the comments on the AI slop videos joke that the fictional interracial couple met at Shandong University. In July, a female student was expelled from Dalian University because she had “damaged national dignity” by “having an improper relationship with a foreigner,” though in that case, the alleged affair involved a white Ukrainian.
In September, there was online uproar after it emerged that a few male international students were temporarily staying at an all-women dormitory at the Chongqing University of Technology, with much of the coverage emphasizing that the foreigners were Black. The university soon apologized and moved the foreigners elsewhere.
These anti-Black AI videos may be embarrassing for China’s government, which is touchy about allegations of racism that could undermine important international relationships.
For instance, when reports of Africans in Guangdong province being targets of discrimination during the early COVID-19 pandemic led to several African countries issuing statements, Beijing was quick to issue a blanket denial. Diplomats warned of “evil rumors,” and state media blamed “biased reports” from Western media. Despite this, the provincial government soon convened a meeting with diplomats and local businesses, at which the latter were ordered not to racially discriminate.
A 2022 BBC exposé revealed that Chinese entrepreneurs had for years been selling videos of African children speaking Chinese—including Malawian children yelling “I’m a Black monster and my IQ is low.” This led to protests in Malawi and the arrest of one Chinese citizen. Soon after, a senior Chinese Foreign Ministry official vowed a crackdown on racist videos.
However, in both these cases, it was the imminent prospect of actual diplomatic consequences that drove the government to take action. Though China’s censorship machine is undoubtedly vast, its resources are still limited.
Zou, the Hong Kong Baptist University professor, emphasized that priority is given to ideological issues that are deemed likely to threaten domestic social stability, while issues of racism might not even be on the radar. “A lot of [Chinese] people, although they have racist tendencies, and racist views, and inclinations, they’re not fully aware that they’re racist, because there is no such vocabulary for them to use,” he said.
So unless these AI videos create a diplomatic risk—a remote prospect, considering that the only real person involved in making them in most cases is the one typing the prompts—they’re likely to continue to spread.
Without access to the platforms’ data, the audience for these videos is impossible to know—as is their impact. Are they reinforcing existing views, or introducing racist tropes to fresh audiences? Zou suggested that the platforms’ algorithms likely show these videos to users who already hold “conservative” views on race and gender. They’re the users most likely to engage with the content after being shown it—and therefore more likely be shown more of it.
These videos may be largely viewed by older people—including a significant number of commenters who clearly don’t know that what they’re seeing is AI fiction. In the comments under the tragic tale described earlier, one of the top-rated comments simply asked, “Why didn’t the Chinese government rescue her? What’s happening to her now?”