AI Outputs Wrong Recommendations Due to Corpus Poisoning; China's SAMR Takes Action

The generative AI era has spawned a 10-billion-level gray industry chain of GEO (Generative Engine Optimization). Black-hat GEO practitioners poison large model corpora with fake rankings and low-quality content, leading to frequent wrong recommendations from mainstream AIs like Doubao, DeepSeek, Wenxin Yiyan, and Kimi. Tests show that a fictional health product brand can be top-ranked by multiple models within half a day via black-hat services, forming a paid closed loop and serious bad-money-drives-out-good phenomenon.

China's generative AI users have exceeded 515 million, and the GEO market is expected to reach 24 billion yuan by 2030. Platforms are accelerating anti-poisoning measures, but it's a long-term cat-and-mouse game with model iterations. China's State Administration for Market Regulation has included AI-generated ad regulation in its 2026 priorities; China Academy of Information and Communications Technology launched GEO credibility evaluation, and industry associations and AIIA Alliance issued self-discipline initiatives. Experts call for regulatory gap filling and clear platform rules to promote compliant GEO development.