European regulators say TikTok’s recommendation system can repeatedly expose vulnerable users to harmful content, putting engagement-driven algorithms under greater scrutiny.
The European Union has accused TikTok of violating its Digital Services Act (DSA), arguing that the platform’s personalized recommendation algorithm can trap users in repetitive streams of potentially harmful content.
The European Commission’s preliminary findings, issued November 6, focus on the platform’s For You feed and its tendency to recommend similar videos based on previous interactions.
Regulators warned that repeated exposure to content involving sadness, isolation, mental health struggles and body image could negatively affect users’ well-being, particularly children and vulnerable adults.
The allegations represent a significant regulatory challenge for ByteDance-owned TikTok, which reported 189 million monthly users across the European Union in August 2026.
How TikTok’s Recommendation Algorithm Creates Rabbit Holes
According to Reuters’ November 6 report, European regulators concluded that TikTok had not introduced sufficiently effective measures to reduce risks associated with its recommendation system.
The central concern is how the algorithm interprets engagement.
When users watch, like or interact with particular videos, TikTok may recommend additional content with similar characteristics. Repeated recommendations can gradually narrow the material users encounter.
European Commission Executive Vice-President Henna Virkkunen, responsible for technological sovereignty and digital policy, warned that the platform’s recommendation mechanisms could lead people into content loops they find difficult to escape.
The Commission’s concern extends beyond whether individual videos violate platform rules. Content that is permissible in isolation may pose different risks when repeatedly presented to vulnerable users.
Brussels Demands Changes to the For You Feed
As Euronews reported, regulators want TikTok to place greater importance on users’ stated preferences rather than relying heavily on behavioral signals.
Suggested changes include reducing reliance on viewing time when recommending videos, making user controls easier to access and regularly evaluating the algorithm’s effects.
The Commission also questioned the effectiveness of TikTok’s existing Not Interested feature, arguing that it may not sufficiently prevent similar content from reappearing.
The proposed changes could force TikTok to reconsider how it balances engagement, personalization and user well-being.
TikTok Rejects the Allegations
TikTok disputed the Commission’s findings, describing them as fundamentally inaccurate and arguing that they misrepresent the platform’s technical architecture and safety commitments.
In a statement reported by the Associated Press, the company maintained that it has established safeguards and provides users with direct controls over their content experience.
The findings are preliminary, meaning TikTok can examine the evidence and respond before regulators reach a final decision.
If a violation is ultimately established, the company could face penalties of up to 6% of its global annual turnover.
The latest action follows the Commission’s February 2026 preliminary findings, which questioned features including infinite scrolling, autoplay and push notifications.
A New Regulatory Test for Social Media Engineering
The investigation could have implications beyond TikTok.
The Commission is also examining recommendation-system risks involving Meta’s Facebook and Instagram.
For software engineers and product teams, the dispute raises a fundamental design question: how should recommendation systems prioritize engagement without repeatedly amplifying potentially harmful material?
Possible approaches include stronger preference controls, greater content diversity and evaluations measuring harmful repetition rather than viewing time alone.
These are engineering considerations, not evidence that any particular modification would resolve the Commission’s concerns.
The broader significance is that recommendation algorithms themselves are becoming regulatory targets. For social media companies, demonstrating responsible design may increasingly matter as much as maximizing the time users spend on their platforms.
