In the last eight months, more videos of foreigners generated by artificial intelligence, or deepfakes, have been popping up on Chinese social media.

In one video on TikTok’s Chinese sister site Douyin, a woman by the name of Lilian with brown hair and glasses speaks near-perfect Mandarin.

“Do you know why men nowadays don’t want to spend money on women anymore?” she asks.

Marketplace found her videos in at least two dozen accounts across Douyin and video platform Kuaishou under different names such as Elena, Aurora and Rosalie. A Shanghai digital media outlet, the Paper, went undercover at a company creating deepfake accounts and revealed there is a real woman behind the videos.

Her real name is Andrea Gabor, Bloomberg professor of business journalism at Baruch College at the City University of New York. She was first informed of her deepfakes when Marketplace reached out.

“I was stunned,” she said. “First of all, I’m a relative nobody.”

It is not just politicians, celebrities and influencers who should be worried about their likeness being stolen. Gabor is active on X, formerly Twitter, but said she barely uses other social media platforms and thinks twice before putting things in email. That did not protect her.

“They basically stole an image of mine from a relatively wonky obscure economic institute where I had done an interview,” Gabor said.

Instead of talking about economics, her deepfakes give love advice. That is typical, according to Chen Yan in Beijing. He keeps tabs on the latest AI technologies.

“A foreigner in China giving what we call ‘Chicken Soup for the Soul’ type advice is a big contrast. It attracts people’s attention,” he said.

Once there is enough attention or followers, the account content shifts. “Then, [those accounts] could sell products or online courses,” he said.

These days, most of Gabor’s deepfakes no longer talk about love and are pedaling Russian honey, Russian biscuits and Russian sea salt.

The Paper also identified other deepfakes of Peter Navarro, former top White House advisor to Donald Trump, and Houston Christian University professor Nancy Pearcey. Marketplace reached out to both for comment but did not receive any responses.

University of Pennsylvania student Olga Loiek, who has a few videos on YouTube, discovered her face was cloned on Chinese social media. They are similar to Gabor’s, where the videos start off talking about good Russia-China relations and how much Russia needs China’s economic support, and sometimes end on selling Russian goods.

“It seems like the main narrative of these clones is to strengthen the relationship between these two countries, presenting me or my clone as a Russian [point of view],” Loiek said in a YouTube video earlier this year. “As a Ukrainian, this has obviously been infuriating for me. After all, my family has to hide from air raid sirens and hundreds of thousands of my fellow Ukrainians are getting displaced, injured or killed because of the Russian attacks.”

Loiek did not respond to Marketplace’s request for comment.

Marketplace also discovered actor Chris Evans’ deepfake under an account called Immersive Thinking on Douyin, which appears to be in the stages of trying to grow viewership and does not sell anything at the moment. In one video, Evan’s deepfake talks about a lack of money leading to other problems in life. In another, his deepfake said when a man is down on his luck, he is unlikely to listen to advice. Marketplace reached out to Evans’ agents but did not receive any reply.

Chen said deepfake technology has been around since 2017 but has only recently become more affordable. He generated one of his own recently.

“It took me 20 seconds to choose a topic, a digital person I’ve trained before, a voice model I’ve trained, then I click download,” Chen said.

A couple minutes later, a video of him standing inside an apartment selling medical insurance appears. His mouth is not in sync with the audio. That rings true for Gabor’s deepfakes too, but there are a lot of details the videos do get right.

“Obviously, it’s weird to have Mandarin or whatever was coming out of my mouth, but it sounded like my voice,” Gabor said.

Chen, who posts regularly on China’s photo-rich platform Xiaohongshu or Little Red Book, and does livestreams for his company, said it’s hard to protect his image. His biggest worry is that people could use his deepfakes to scam his family.

“I have a verification word with my family when we have video calls,” he said. He hopes that extra layer of protection helps his family know they are speaking to the real person.

Chen adds that there are legitimate business uses for this technology. AI clones could be used to deliver online courses, give virtual tours of museums and galleries, or help e-commerce vendors conduct livestreams, which typically last anywhere from six to eight hours. Chinese government media use AI clones of their reporters to produce video content faster or to deliver the news in perfect English.

Douyin and Kuaishou have started to label accounts as AI-generated. Under China’s AI regulation, it is prohibited to use someone else’s likeness without their consent. China’s police have cracked some AI deepfake cases and even made some arrests in recent years.

However, it is hard for foreigners like Gabor to even find out if there are deepfakes of herself in China. Marketplace had to guess hashtags on Douyin and Kuaishou before Gabor’s deepfakes showed up.

“The whole idea of the Bloomberg professor giving relationship advice is comical on some level,” Gabor said. “To me, [this technology is] the tip of the iceberg.”

She said the dangers are much more profound to U.S. democracy. “My preoccupation right now is how can this material and this technology be used to manipulate [voters in] our election.”

Additional Reporting by: