Eigen RadarAI
Analysis

EU challenges TikTok’s harmful recommendation loops

The European Commission has issued preliminary findings against TikTok’s For You feed, saying repeated recommendations can draw users into harmful content loops. It wants changes to the signals and controls shaping its recommendations. TikTok disputes the allegations and can present its defense before any final decision or financial penalty.

Artificial Intelligence··Midday
Glass entrance doors at the European Commission’s Berlaymont headquarters in Brussels, the institution challenging TikTok’s recommendation system.

Viewing time steers repeated recommendations

The European Commission says TikTok’s personalized For You feed can repeatedly recommend harmful material, creating loops that users struggle to leave. Its November 6 preliminary findings concern compliance with the Digital Services Act. The Commission points to weight-loss and body-image videos: time spent watching can be treated as interest and bring further recommendations on the same subject.[1], [2]

Explicit preferences would carry more weight

The Commission wants explicit user preferences to have more influence, controls to be easier to reach and the algorithm to undergo regular testing. It also wants viewing time removed as a signal. According to its assessment, content-diversification measures have not sufficiently reduced similar recommendations, and the Not Interested control can be outweighed by viewing behavior. Executive Vice-President Henna Virkkunen highlighted the well-being of minors.[1]

TikTok can contest the preliminary findings

TikTok disputes the allegations, saying they misrepresent its architecture and safety commitments. It says users already have direct, intuitive controls. The company can inspect the case file and submit a written defense. The investigation forms part of proceedings opened in February 2024. A confirmed breach could bring a fine of up to 6 percent of global annual turnover; the preliminary findings have not imposed one.[1]

References

  1. News sourceEFEEU issues new preliminary findings on TikTok’s recommendation feed↩1↩2↩3
  2. News sourceEl EspañolEuropean Commission challenges TikTok’s recommendation algorithm↩