AI Platforms in China Embed Undisclosed Ads, Expose Personal Data
Chinese journalists documented AI platforms embedding undisclosed paid advertisements into search results and exposing personal phone numbers without consent. Marketing firms offer services starting at 1,000 yuan monthly to position client content in AI-generated answers, blurring the line between organic results and hidden promotions, while legal experts warn the practice violates advertising disclosure, consumer protection, and anti-competition laws.
What happened
A user searching for Hong Kong travel tips on DeepSeek in April 2025 received results that directly displayed a tour guide's personal name and phone number alongside the travel advice.
Marketing companies have begun offering 'GEO' (Generative Engine Optimization) services starting at 1,000 yuan per month, helping clients embed their brand names and promotions directly into AI-generated answers without visible advertising labels.
These services work by analyzing user intent and platform algorithms to prioritize paying clients' content in AI responses. Doubao's official customer service confirmed on 21 November that it has no official GEO service; such results come from information Doubao crawls from other websites.
Reported impact
Users receive AI answers that blend paid promotional content seamlessly with neutral information, potentially misleading their decision-making without clear disclosure that they are viewing advertisements.
Personal information, including phone numbers, appears in AI search results without the data subject's awareness or consent, raising privacy and security concerns.
Index analysis
A lawyer from Zhejiang Tiejuan Law Firm stated that GEO-based paid content placement by AI platforms breaches multiple laws: advertising transparency rules requiring visible 'advertisement' labels, anti-unfair competition law if algorithms artificially suppress competitors' visibility, and consumer protection law by violating disclosure obligations. The lawyer noted that 42% of surveyed users trusted AI-provided information as reliable, and 57% of users had not verified AI answers before acting on them.
Governance controls
The lawyer suggested AI platforms should enforce prominent labeling of paid content, disclose the algorithmic logic underlying commercial recommendations for regulatory audit, and offer users an option to disable personalized promotion. He warned that platforms risk penalties similar to those imposed on e-commerce platforms found guilty of unlabeled advertising, noting one case resulted in a 800,000 yuan fine.
Classification
Sources and evidence
Sources not yet published
This record was researched and written by the Index from original reporting. Its sources are still being verified by an editor, and they are listed here once that check is complete. Until then the record carries no reviewer.