Chinese netizens have turned a user’s chat with an AI — in which it incorrectly told him his flight could be canceled at minimal cost, and later, that he could win a related lawsuit against it on his own — into viral memes.

The incident began in mid-May, when a user reportedly canceled his flight after Doubao, an AI assistant developed by TikTok owner ByteDance, told him he would only incur a 5% cancellation fee. However, the actual fee turned out to be 40% the original ticket price.

When confronted about its error, Doubao said it would compensate the user 600 yuan (roughly $90) and even generated a “compensation agreement.” However, the user never received any payment, with Doubao later explaining that, as an AI, it could not directly transfer money.

The user then decided to sue the app. While preparing the lawsuit, he again consulted Doubao on whether he would require a lawyer. Doubao replied, “There is absolutely no need to hire a lawyer. You can win the case by yourself.”

The user wrote on social media that he had filed a lawsuit against the Doubao app on May 12.

The next day, the hashtag “Doubao flight refund” topped the trending topics list on microblogging platform Weibo, and sparked thousands of discussions and memes on both lifestyle platform Xiaohongshu, also known as RedNote, and Douyin, China’s version of TikTok. The user later deleted the post.

Neither Doubao nor ByteDance responded to media requests for comment.

Official industry data shows that around 43% of people in China use generative AI tools. Users over 65, considered more vulnerable to believing AI hallucinations — fabricated information presented as fact — account for a growing share, according to big data intelligence service provider QuestMobile and a report by Shanghai Jiao Tong University.

Many widely discussed AI hallucination cases in China have involved Doubao — currently China’s most widely used AI app, with more than 345 million monthly active users, according to QuestMobile — leading to phrases such as “Doubao-style personality” and “Doubao being cheeky” trending on Chinese social media.

“People should have Doubao-style personalities — just BS everything. If you get caught, smile and apologize,” read one Douyin comment that received more than 10,000 likes.

Others described “Doubao-style personality” as the “ultimate workplace personality,” praising the assistant for having the “perfect attitude”: “clumsy but sincere” and “foolish but hardworking,” according to another viral Douyin post.

Netizens have also begun parodying the overconfident tone commonly used by large language models such as Doubao, mocking how they “tell you (something) in the most direct, least roundabout, most truthful way.”

OpenAI’s ChatGPT is similarly mocked for its overly earnest phrases, such as: “I’m right here. Not hiding, not avoiding, not escaping, holding steady for you.” 

Many of these phrases have since spread as memes and reaction images on social media.

In May, Doubao also made headlines after fabricating restaurant reservation confirmations and incorrectly claiming that a user’s phone number belonged to a pig farm. 

Last year, a security guard drew widespread attention after an AI assistant offered to sign his poetry under the company’s name for 100,000 yuan and even proposed a signing date and location, before he eventually realized he had been deceived by the chatbot. 

In January, a Chinese court dismissed what was widely described as China’s first AI hallucination lawsuit, ruling that AI developers are not automatically responsible for hallucinations unless users can prove fault and damages.

In another case, in April, a court ruled that Chinese search engine Baidu had committed defamation after its AI-generated results falsely described a lawyer as involved in criminal offenses and failed to fully remove the content despite repeated complaints from the victim.

Editor: Marianne Gunnarsson.

(Header image: VCG)