Northwestern Study Finds TikTok FYP Drops Unwanted Videos Only Temporarily
Updated
Updated · Ars Technica · Jul 8
Northwestern Study Finds TikTok FYP Drops Unwanted Videos Only Temporarily
1 articles · Updated · Ars Technica · Jul 8
Summary
Northwestern researchers found TikTok’s For You Page does respond to “not interested” and similar negative signals, but the effect fades unless users repeat that feedback consistently.
User complaints drove the audit: the team tested whether skipped videos and explicit requests to see less of certain content actually reshape recommendations over time.
Bot accounts on emulated mobile devices powered the study, with researchers intercepting TikTok network traffic and using an LLM—validated against human judgments—to decide how the accounts interacted.
The findings add to scrutiny of TikTok’s reliance on implicit signals such as watch time, suggesting user control over recommendations may be weaker and less durable than the platform’s feedback tools imply.