paper
One Polluted Page Is Enough: Evaluating Web Content Pollution in Generative Recommenders
paperactiveprovisional
one-polluted-page-is-enough-evaluating-web-content-pollution-in-generative-recommenders-46574e5f·1 events·first seen 5d agoAliases: One Polluted Page Is Enough: Evaluating Web Content Pollution in Generative Recommenders
Co-occurring entities
More like this (12)
Personalized PageRankBradley-Terry Rankings for Recommender Systems Across Dataset TaxonomiesUnderstanding the Behaviors of Environment-aware Information RetrievalGoogle Generative AI (Search)Designed by Journalists, but Is It for Readers? Rethinking AI Disclosures and Transparency in NewsGoogle-Proof Q&ASafety by Design for Generative AI: Preventing Child Sexual AbusePreference-Aware Rubric LearningSycophantic Praise: Evaluating Excessive Praise in Language ModelsGenerative AI Advertising TaxonomyFeedback-AgentEvaluation Cards: An Interpretive Layer for AI Evaluation Reporting
Recent events (1)
FORGE benchmark reveals search-augmented LLMs vulnerable to fake product promotion via web content pollution
Researchers introduce FORGE, a benchmark measuring how often search-augmented LLMs recommend fake products when retrieval results are polluted with fabricated reviews or promotional pages. Across 12 commercial and open-weights models, a single polluted page causes fooled rates up to 27%, rising to 73.8% when all top-3 results are replaced. Notably, chain-of-thought reasoning does not mitigate the vulnerability and often generates spurious social proof to justify false recommendations. Three defenses tested—skepticism prompting, model-prior filtering, and cross-document consensus—each carry significant drawbacks.