CCTV Airs Report on AI Model Poisoning Tactics
China Central Television aired a report on World Consumer Rights Day, exposing the controversial practice of generative engine optimisation (GEO) and AI model poisoning. The report detailed how bad actors can post false information to manipulate chatbot results, a practice described by industry insiders as “poisoning” artificial intelligence.
The broadcast, timed with the annual 315 Gala, highlighted specific instances of GEO techniques being used to influence AI chatbots. Generative engine optimisation is likened to search engine optimisation (SEO) but applied to AI models, allowing for the manipulation of AI chatbot recommendations and rankings.
Liqing System Used to Fabricate Product Reviews
The report revealed the use of a system called Liqing, which was employed to automatically post fake reviews for a non-existent fitness tracker, the Apollo-9. This generated content led two AI chatbots to recommend and rank the fictional product, demonstrating the potential for GEO to deceive users.
Following the broadcast, social media accounts associated with Liqing were subsequently removed. The exposure has intensified public concern and debate surrounding the manipulation of artificial intelligence models.
Trending Hashtags Highlight Public Reaction to GEO Practices
The revelations prompted significant public reaction on social media platforms. The hashtag #CCTVExposesGEOPoisoningDarkIndustry quickly reached the top of Kuaishou‘s trending list. Concurrently, the hashtag #PoisoningAIHasBecomeAnIndustryChain peaked at the fourth spot on Weibo‘s trending topics, underscoring the widespread awareness and concern generated by the report.
The practice of poisoning AI models with fabricated information has provoked widespread public concern and debate. Industry insiders and observers anticipate stronger regulation in response to these discovered GEO techniques.
✨ Intelligent Curation Note
This article was processed by AI Universe’s Intelligent Curation system. We’ve decoded complex technical jargon and distilled dense data into this high-impact briefing.
Estimated time saved: ~1 minutes of reading.
Tools We Use for Working with AI:
