AI BriefingChina fines companies for AI data poisoning, with 30% of cited links showing fake placement
·2026.10.22 00:42
Key point
Regulators in Beijing, Guangzhou, and Shenzhen have issued fines ranging from 50,000 to 120,000 yuan to companies planting fake content to manipulate AI chatbot citations.
Details
- Chinese regulators have begun fining companies for "AI data poisoning," issuing three penalties between 50,000 and 120,000 yuan in Beijing, Guangzhou, and Shenzhen.
- A test by 21st Century Business Herald found that over 30% of more than 1,000 links cited by Chinese AI models in April showed signs of paid placement, often using fabricated sources or data.
- The practice involves paying middlemen to publish client articles on websites already cited by AI tools, with contracts promising top-three or top-five spots in AI answers.
- A CCTV investigation demonstrated that an AI model praised a fake product within two hours of publishing fake articles, and two models recommended it after three days of posting fake reviews.
- The campaign also targets AI apps for failing to verify sources or display citation links; the author noted they could not find a comparable fine in the US or EU.
This summary was generated automatically by AI. Check the original for the author's claims and context. Copyright belongs to the original author.
Our guide explains how the AI works. Report summary errors, attribution issues, or removal requests via Contact.
Related
China expands AI travel curbs to executives' spouses and children
Alter launches YC S25 platform for privileged access management of AI agents