The European Commission has preliminarily found TikTok in breach of the Digital Services Act over the way its For You feed can pull users into harmful "rabbit holes," while calling for greater weight to be given to the preferences users state rather than engagement signals such as viewing time. For Cybersmile, this is the clearest victory yet for Stop the Forced Feed and a major validation of the principle at the heart of the campaign that engagement is not consent.
The Commission says TikTok's recommendation system can treat behaviors such as viewing or liking content as signals to serve more of the same, including material related to mental health struggles, sadness, isolation, weight loss, and body image. Over time, that can pull users deeper into feeds that may negatively affect their wellbeing, particularly minors and vulnerable adults.
What makes the finding especially important is that the Commission is not only challenging harmful content but also the way the recommendation system behaves. The Commission says TikTok should rely more on the preferences users state, stop using viewing time as a signal for recommendations, make its controls easier to use, and regularly test its algorithm. It has also preliminarily found that controls such as "Not Interested" may not be enough when the algorithm continues to prioritize viewing time over a user's choice to see less of certain content.
For Cybersmile, this distinction is hugely significant because Stop the Forced Feed has always been built around the principle that engagement is not consent. Watching something for a few seconds, pausing on it or interacting with it should not be treated as permission for an algorithm to keep serving more of the same. People need meaningful control over what recommendation systems put in front of them, with their choices carrying more weight than assumptions made from their behavior.
The issue we took directly to the European Commission
We began developing Stop the Forced Feed in June 2025 because we believed the online safety debate was focusing too heavily on harmful content after exposure and not enough on the systems deciding what gets promoted, amplified, and repeated. Our research during 2025 gave us concerning evidence of how quickly that can happen.
In February 2026, we published our Forced Feed 2026 report and officially launched Stop the Forced Feed, alongside our PROTECT framework calling for greater transparency, accountability, and user agency over recommendation systems.
In our behavioral study, harmful content appeared on adult test accounts in as little as 16 seconds and on accounts representing children in as little as eight minutes. Within three days, 90% of the accounts had encountered racist content, while 60% had been exposed to misogyny and 60% to violence.
We also found that people wanted much more control over the systems shaping their feeds. Sixty-five percent wanted more control over their algorithms and the content they saw, 69% wanted greater insight into why algorithms were showing them particular content, and only 36% felt in control of their algorithmic experience.
Those findings became the basis of Stop the Forced Feed and our PROTECT framework, which called for a transfer of power from platforms to users and for meaningful control over recommendation systems to become a basic user right. We then took that evidence directly to the European Commission.
In February 2026, Cybersmile submitted its research on algorithmic harm to the Commission, highlighting the need for greater transparency, accountability, and user agency. In March, we met Werner Stengg to discuss the research and its wider policy implications. In May, we met Prabhat Agarwal and members of the Commission's enforcement team specifically to discuss algorithmic harm, recommendation systems, and user control.
Those conversations focused on the same underlying problem now at the center of the Commission's findings against TikTok.
User agency over algorithms
The Commission says TikTok should rely more on the preferences users state rather than relying so heavily on signals inferred from their behavior. It has also said viewing time should not be used as a signal for recommendations and that tools designed to help people shape their feeds need to work properly in practice.
That closely reflects the principle behind our PROTECT framework. Cybersmile has consistently argued that platforms already have the technical ability to categorize and personalize content, but users have too little agency over how that power is used. Our position has been that people should be able to shape the subjects and categories recommendation systems serve them, with the choices they make taking priority over an algorithm's interpretation of their behavior.
A "Not Interested" button is not meaningful control if the underlying system continues to reward viewing time and repeatedly returns users to similar content. User agency has to mean more than being offered a button while the recommendation system continues to make the most important decisions.
What an algorithm thinks will keep someone watching should not carry more weight than what that person has actually said they want to see. We researched that problem and submitted the evidence directly to the European Commission. To now see the Commission challenge TikTok's For You feed on those grounds is a major victory for the campaign. People should have meaningful control over the algorithms shaping their realities.
Scott Freeman, CEO, The Cybersmile Foundation
A major campaign victory
We researched how recommendation systems can expose people to harmful material, documented the public demand for more control, developed a framework centered on user agency, and put that evidence directly in front of the European Commission.
The Commission has now preliminarily found that TikTok's For You feed breaches the Digital Services Act because of harmful recommendation rabbit holes and is calling for changes that include giving greater weight to users' own preferences and providing more effective control over recommendations.
That gives Stop the Forced Feed a clear and documented place in the record of how this issue developed. Cybersmile was making the case for greater user agency and meaningful control over recommendation systems before this enforcement finding was issued, and we took that case directly to the institution now acting on the same underlying problem.
The TikTok finding also sits within a wider shift toward greater user agency, following Meta's move toward non-algorithmic feed controls and the European Commission's KIDS Act proposal on safer recommender systems.
What happens next for TikTok
TikTok now has the right to examine the Commission's evidence and respond before any final decision is made. If the findings are confirmed, the Commission could issue a formal non-compliance decision and impose financial penalties under the Digital Services Act. This is a preliminary finding, not the final outcome of the case, but the significance for Stop the Forced Feed is already clear.
We launched the campaign because people should not have to accept whatever an algorithm decides to keep serving them. We argued for content with consent, meaningful control, and greater user agency over recommendation systems. The European Commission is now challenging TikTok on those same underlying principles, making this the clearest win yet for Stop the Forced Feed and a major step toward putting users, rather than algorithms, back in control of what they see.
