TikTok’s “Rabbit Hole”: Why Your Feed Keeps Returning to the Same Topic

Published on: ⋄5 min read
Author: Valeriia Voronova, Editor at SOD Energy
A metaphorical loop of topics in TikTok recommendations

Spend a little too long on one sad video on TikTok, and soon your feed starts serving up more and more similar clips. Stories about breakups, loneliness, and depression follow one another, even when you want to take your mind off things and switch to something positive. The algorithm interprets interest in a topic as a signal to show more of it, but it cannot always distinguish curiosity from a user’s emotional vulnerability.

This mechanism is what drew the European Commission’s attention. On November 6, 2026, the regulator published its preliminary findings on how the For You recommendation feed works. In the Commission’s assessment, the algorithm can draw users into so-called “rabbit holes”–endless sequences of similar content that may intensify emotional fixation on certain topics and pose risks to mental health. The regulator believes the existing tools for managing recommendations are not effective enough at preventing this unwanted repetition.

However, these are still preliminary findings under the European Union’s Digital Services Act, not a final determination that TikTok has violated the law. TikTok disputes the claims, arguing that the European Commission has misinterpreted how the recommendation system and the platform’s safety measures work.

This dispute raises a broader question: where is the line between personalized content and the algorithmic reinforcement of emotional states that can be hard for users to break out of?

How TikTok recommendations create a “rabbit hole”

Personalization helps people find interesting videos without having to search all the time. But the same principle can narrow the feed: a person engages with a topic, the system offers more similar videos, and further views strengthen the initial signal. A self-reinforcing cycle takes shape.

The Commission identifies repeated content about sadness, isolation, and body image as potentially harmful, especially for minors and vulnerable adults. The central concern is the sequence of recommendations. Assessing individual posts does not answer the question of what happens when one topic repeatedly resurfaces in someone’s feed.

Here, attention and the desire to keep watching are not the same thing. For example, someone may linger on a distressing video because they recognize their own feelings in it or are trying to make sense of what they have seen. Watch time alone does not explain why they paused. If the system interprets it primarily as interest, it may serve up even more of the same topic.

In this scenario, the feed gradually shifts from a mix of different stories to a space where one emotionally significant topic is constantly in view.

Diagram showing the feedback loop between viewing a topic and receiving repeated recommendations
A self-reinforcing cycle arises when viewing a particular topic leads to more similar recommendations.

Why the “Not interested” button may not be enough

The advice to “just stop watching” assumes that users can easily tell the system what they want and get a feed that reflects their choice. The European Commission has questioned how effective this kind of control really is.

According to its preliminary assessment, the algorithm prioritizes watch time over users’ explicitly stated wishes to see less of certain content. So tapping “Not interested” may not be enough. A contradiction arises: someone says they want to change topics, but their attention to the next video continues to serve as a recommendation signal.

The design of the viewing experience also makes it harder to leave. As early as February 6, 2026, the Commission issued separate preliminary findings concerning infinite scrolling, autoplay, notifications, and personalized recommendations. At the time, the regulator considered screen-time limits insufficiently effective: they are easy to dismiss and create few barriers to continuing to watch.

The February and November findings address different aspects of the problem. The first concern design features that encourage people to keep watching. The second focus on repeated recommendations around particular topics. Taken together, they help explain why leaving this kind of feed takes more than simply deciding to switch. At the same time, difficulty managing recommendations does not mean that changing your feed or stopping your viewing is impossible.

What users say about thematic echo chambers

A small qualitative study by Wahidah Pan and colleagues, published on October 28, 2026, in the International Journal of Eating Disorders, sheds light on the psychological side of this experience. The authors interviewed 25 TikTok users who reported currently having eating disorders.

Participants described two paths into a thematic echo chamber–a space filled with repetitive, similar content. One began with videos about health and wellness and gradually led to material about eating disorders. The other involved deliberately engaging with this subject matter.

It was not only the videos themselves that kept people within these recommendations. Participants described emotional attachment, a sense of belonging, and difficulty removing unwanted topics from their feeds. This is an important detail: content can bring up difficult feelings while also making someone feel understood. The desire to distance yourself from it can coexist with the desire to return.

The study explores the experiences of a small, specifically selected group. It does not establish how often this happens among all TikTok users, nor does it prove that recommendations caused the disorders. Neither these interviews nor the regulator’s findings make it possible to calculate the likelihood that a particular viewer will develop depression or another disorder.

What the European Commission proposes changing

The November findings concern the very way the For You feed is configured. The Commission proposes:

  • giving users’ explicitly stated preferences more weight;

  • stopping the use of watch time as a recommendation signal;

  • improving tools for managing the feed;

  • regularly testing the recommendation algorithm.

For everyday users, the most noticeable change could be a more consistent response to requests to see less of a particular topic. This matters when someone wants to keep using the service but leave a repetitive stream of specific content.

The Commission’s findings are based on its review of TikTok’s internal documents and data, risk assessment reports, scientific research, and input from experts in recommendation systems, psychology, and psychiatry. The next step gives TikTok an opportunity to review the investigation materials and respond to the preliminary findings in writing.

The practical question at the heart of this dispute goes beyond whether there is a button to opt out: how reliably does the system take a person’s decision into account when their viewing behavior sends a different signal?

What fine could TikTok face?

Up to 6% of the service provider’s total worldwide annual turnover if a violation is confirmed in a final decision. The publication of preliminary findings does not, by itself, mean that a fine has been imposed.

Can you tell from someone’s feed whether they have depression?

No. TikTok recommendations cannot diagnose anyone. The same topics may appear for different reasons, and watching videos about sadness does not, by itself, indicate a mental health disorder.

Do the findings mean TikTok must remove all videos about sadness?

No. The concerns described focus on repeated recommendations and insufficient user protection. Discussing sadness, loneliness, or body image is not inherently harmful; the content, context, and repeated ways in which the system serves it to someone all matter.

How was this article?

 

Share

Explore more