AI Chatbot Promised User US$14,000, Court Says It Doesn't Have to Pay

When an AI chatbot insisted it was right – and offered a 100,000 yuan (US$14,374) wager to prove it – a user surnamed Liang thought he had an open-and-shut case.

But in a landmark ruling, the Hangzhou Internet Court has decided that AI "hallucinations" and empty promises do not constitute legal intent, shielding developers from liability for machine-generated falsehoods.

The case, which marks China's first judicial response to tort claims arising from generative AI errors, dates back to June 2025. Liang had used an AI platform to query university details for his sibling's college application, but the chatbot provided the wrong location for a main campus.

When Liang challenged the error, the system doubled down.

"If the generated content is incorrect, I will compensate you 100,000 yuan," the chatbot declared, even advising Liang to "file a lawsuit at the Hangzhou Internet Court." Only after Liang presented official admissions documents did the AI admit its mistake.

Liang sued the platform's developer for 9,999 yuan, arguing he was misled by false information and entitled to the compensation the bot had promised.

The court rejected his claims in a first-instance judgment that sets significant boundaries for AI liability.

"Artificial intelligence lacks the capacity for civil rights and therefore cannot make legally recognized expressions of intent," the court stated, ruling that a machine's output cannot be interpreted as a binding offer from the provider.

The court found the developer had fulfilled its obligations by implementing user agreements and disclaimers about potential inaccuracies. Under these safeguards, the mere generation of incorrect information – a known phenomenon called "hallucination" – did not amount to a rights infringement.

Finding neither fault nor infringement, the court dismissed all of Liang's claims. Neither party appealed the decision.