Auditing the Impact of Cross-Site Web Tracking on YouTube Political and Misinformation Recommendations

A new study from researchers Salim Chouaki, Savaiz Nazir, and Sandra Siby provides the first experimental audit of how Google's cross-site web tracking directly influences YouTube's recommendation algorithm, particularly for political and misinformation content. Published on arXiv, the research introduces a novel 'sock-puppet-based' framework that uses automated bots to simulate user behavior. These bots first interact with news articles across the web—where Google trackers collect browsing data—and then analyze the subsequent YouTube recommendations they receive. This methodology isolates the impact of off-platform tracking, a factor previous studies, which focused solely on on-platform watch history, had overlooked.

The study's key finding is that browsing activity on external news sites, monitored by Google's pervasive tracking infrastructure, plays a significant role in shaping the political lean and reliability of videos YouTube suggests. The researchers ran parallel audits in both tracking-permissive and tracking-restrictive browser environments (like those with enhanced privacy protections). This comparative approach allowed them to assess whether common privacy-focused browsers and tools can effectively protect users from being funneled into tracking-driven 'filter bubbles' of polarized or misleading content. The results indicate that restricting cross-site tracking can alter the recommendation pathway, offering a potential technical mitigation to a major societal concern surrounding algorithmic amplification.