Coordinated Manipulation Threatens Crowdsourced Fact-Checking Systems

Major social media platforms like X, Meta, TikTok, and Google employ crowdsourced fact-checking to combat misinformation at scale, relying on bridging mechanisms that identify misleading information when supported by diverse perspectives. The core of these systems, often based on matrix factorization (as seen in X and Meta), is designed to prevent simple majority rule from being exploited. This study examines the vulnerability of these matrix factorization-based systems to coordinated manipulation. Theoretical and empirical evaluations, using historical production data, show that strategic voting by a small group of users can fabricate synthetic consensus. Specifically, up to 10.7% of lower-quality notes could be manipulated past consensus thresholds with fewer than 10 ratings. Counterintuitively, the analysis also reveals that rating a note as "Not Helpful" can sometimes increase its helpfulness score. The research quantifies the manipulation effort and notes that mitigations have been developed and deployed within X's Community Notes algorithm.