AI tool used to generate fake news about Chinese car makers; two arrested
Two individuals used an AI paraphrasing tool to generate and distribute 700,000 false articles about Chinese car companies including Xiaomi, NIO and Volvo, earning 80,000 yuan from platform rewards. Shanghai police arrested them in March 2026 and charged them with illegal operations for knowingly distributing fabricated information that disrupted the companies' business operations.
What happened
Two individuals, 梁某 and 郭某, purchased approximately 4,000 social media accounts and used an AI paraphrasing tool to generate misleading articles about multiple car companies including Xiaomi, NIO and Volvo. The AI tool processed trending topics and created illustrated posts in approximately 5 minutes per 100+ articles, with no human review required.
The accounts collectively published 700,000 articles that generated over 80 million views. One rumor about 'Xiaomi Car SOS 1-second connection' spread widely through image manipulation. In March 2026, the two individuals were arrested by Shanghai police for illegal operations.
Reported impact
Multiple car companies reported long-term harassment from the false information, which fabricated company performance, battery fire risks, accidents with casualties, declining sales, expanding losses and reduced employee benefits. The false stories used sensational language such as 'slashed in half', 'collapse of trust' and '50 billion yuan hole', creating manufactured controversies about operational difficulties and illegal fees that seriously disrupted normal business operations.
Index analysis
The AI tool failed to preserve accuracy when paraphrasing, omitting critical details. For example, when generating stories about new energy vehicle fires, the tool removed context that fires resulted from collision and battery leakage during accidents, instead claiming vehicles caught fire during normal driving. The individuals deliberately published during hours of weaker platform monitoring (1 am to 6 am) despite knowing AI-generated content could not be verified as truthful.
Governance controls
The individuals obtained platform rewards for original content posted by new account registrations, then discarded accounts when rewards declined and obtained new ones through illegal channels. Platform monitoring gaps during late-night hours enabled systematic posting of unreviewed content.
Classification
Sources and evidence
This record was researched and written by the Index, based on the following original reporting.