The European Union’s action against TikTok marks a significant escalation in the debate over how social media companies use artificial intelligence to influence user behaviour. The central issue is no longer simply whether platforms remove illegal content or protect children from inappropriate material. Regulators are increasingly examining how the underlying technology determines what people see, how long they remain online and whether repeated exposure to particular types of content can create harmful patterns of behaviour.

TikTok has become one of the most influential examples of this model. Its recommendation system continuously analyses user interactions, including viewing duration, likes, comments, shares and browsing behaviour, to identify content likely to maintain attention. This approach has helped the platform establish enormous global reach, but it has also attracted growing scrutiny over the consequences of repeatedly presenting users with similar material. The European investigation reflects a wider concern that the commercial incentives behind engagement-driven algorithms may not always align with the interests of users.

The regulatory action could have implications well beyond TikTok. If European authorities establish that a platform’s recommendation system creates unacceptable risks, other social media companies may also face pressure to explain how their algorithms operate and demonstrate that they have taken reasonable steps to prevent foreseeable harm.

The Business Model Behind Endless Content Recommendations

TikTok’s popularity is closely connected to its ability to deliver highly personalised content without requiring users to search actively for videos. Unlike traditional social media platforms, where users often begin by following friends, organisations or particular creators, TikTok places substantial emphasis on its recommendation feed.

The system learns from individual behaviour and adjusts the material presented to each user. A person watching cooking videos, for example, may gradually receive more food-related content. Someone interested in fitness, politics, celebrity news or personal finance may encounter an increasingly specialised stream of material connected to those interests.

The commercial advantage is considerable. When users repeatedly encounter videos that match their preferences, they are more likely to remain on the platform. Longer viewing sessions create more opportunities to display advertisements, collect engagement data and strengthen the recommendation system.

However, this approach can also produce an unintended consequence. A user who watches several videos dealing with a particular sensitive subject may begin receiving a much larger volume of similar material. The algorithm may interpret repeated viewing as evidence of interest without necessarily understanding whether the user is engaging out of curiosity, distress or concern.

This distinction is important because human behaviour is more complicated than the signals recorded by an automated recommendation system. A person may watch a video because it is disturbing, controversial or difficult to ignore, rather than because they genuinely want to see more of the same content.

The commercial system, however, primarily evaluates observable engagement. That creates a potential gap between what keeps users watching and what is actually beneficial for them.

The European Union Is Targeting the Design, Not Just the Content

The European Commission’s investigation represents an important development in the enforcement of the Digital Services Act. Rather than concentrating exclusively on individual videos or illegal material, regulators are examining whether the design of TikTok’s recommendation system itself creates systemic risks.

The Commission’s preliminary findings suggest that TikTok did not adequately assess the possibility that repeated recommendations could expose users to harmful patterns of content. It also questioned whether the safeguards introduced by the company were sufficiently effective.

One concern is the platform’s reliance on viewing duration as an indicator of user interest. If someone watches several videos about body image or emotional distress, the system may interpret that behaviour as a preference and continue recommending similar material. The algorithm does not necessarily distinguish between someone watching because they find the content useful and someone who is unable to disengage from it.

The Commission has also questioned whether TikTok’s existing content diversification measures work effectively. Its findings suggest that introducing unrelated videos into the recommendation feed does not sufficiently prevent users from repeatedly encountering similar material.

The proposed changes are therefore aimed at the underlying recommendation mechanism. Regulators want TikTok to give greater importance to users’ explicitly expressed preferences, make content controls easier to access and regularly assess whether its algorithm is producing harmful outcomes.

This is a significant departure from the traditional approach of asking platforms to remove harmful videos after they have been identified. The European Union is effectively asking whether the system responsible for distributing content should itself be redesigned.

The Limitations of User Controls

TikTok has introduced several tools intended to give users greater control over their experience. These include the ability to indicate that they are not interested in particular videos, refresh recommendations and manage the subjects appearing in their feeds.

However, the effectiveness of these features is now being questioned.

The Commission believes that simply providing users with controls does not necessarily mean that those controls offer meaningful protection. A feature may technically exist but still be ineffective if it is difficult to locate, requires repeated intervention or does not substantially change the content being recommended.

The problem is particularly relevant for younger users. Children and teenagers may not have the same ability as adults to recognise potentially harmful patterns of online behaviour or regulate their engagement with highly personalised content.

There is also a practical difficulty with expecting users to manage algorithmic risks themselves. Recommendation systems operate continuously and adjust according to changing behaviour. A user may indicate that they do not want a particular category of content, but the platform still has to determine how that preference should influence future recommendations.

If the underlying commercial system continues prioritising viewing duration, individual controls may not be sufficient to overcome the broader incentives driving recommendations.

This is why the European investigation places considerable importance on changing the system itself rather than relying exclusively on users to manage their exposure.

TikTok Rejects the European Commission’s Findings

TikTok has rejected the allegations, arguing that the Commission’s assessment misrepresents how its platform operates and the measures already introduced to protect users.

The company maintains that it provides users with direct control over their content experience and has invested in safety mechanisms intended to reduce harmful exposure.

This disagreement is important because the Commission’s preliminary findings do not constitute a final legal determination. TikTok has the right to examine the evidence, respond to the allegations and challenge the regulator’s assessment before a final decision is made.

The company is likely to argue that recommendation systems are considerably more complicated than the regulatory description suggests. User behaviour can have several explanations, and repeated exposure to a subject does not automatically establish that the platform is causing psychological harm.

There is also a difficult question about how regulators should distinguish between harmful recommendations and legitimate personalisation. A user interested in mental health, for example, may actively seek educational material, personal experiences and professional advice. Preventing repeated exposure indiscriminately could also restrict access to useful information.

The challenge for TikTok will be demonstrating that its safeguards can distinguish between these different situations and respond appropriately.

For the European Commission, the challenge is establishing that its proposed interventions are proportionate, technically practical and supported by sufficient evidence.

The Financial Consequences Could Be Substantial

The regulatory action carries considerable financial implications for TikTok’s parent company, ByteDance. Under the Digital Services Act, serious violations can attract penalties of up to 6% of a company’s annual worldwide turnover.

For a business operating at ByteDance’s scale, such a penalty could represent a substantial financial burden. However, the immediate significance extends beyond the possibility of a fine.

If TikTok is required to redesign important elements of its recommendation system, the company could face operational changes affecting how content is ranked, distributed and monetised.

The recommendation feed is central to its commercial model. Any intervention that reduces the emphasis on viewing duration or changes the way content is personalised could influence advertising exposure and user engagement.

There is no certainty that these changes would necessarily reduce revenue. A platform that users consider safer and more trustworthy could potentially benefit from stronger long-term engagement. Nevertheless, the financial consequences will depend on how effectively TikTok adapts its recommendation system without weakening the features that distinguish it from competitors.

The investigation could also increase compliance costs, as platforms may need to invest more heavily in independent assessments, technical monitoring and documentation of algorithmic risks.

Meta and Other Platforms Could Face Similar Scrutiny

The European action against TikTok is part of a wider regulatory effort examining how social media recommendation systems influence user behaviour. The European Commission is separately investigating similar concerns involving Facebook and Instagram, both operated by Meta.

This is significant because the problem is not exclusive to one platform. Most major social media companies rely on automated recommendation systems to determine which content users encounter.

YouTube, Instagram, Facebook and other services use different combinations of viewing history, engagement, user preferences and content characteristics to personalise recommendations.

Their commercial models also depend heavily on maintaining user attention. The more time people spend on a platform, the greater the opportunities to display advertisements and promote additional services.

The European investigation could therefore establish a precedent for examining whether engagement-based recommendation systems adequately account for their potential consequences.

The Digital Services Act gives regulators a framework to demand that large platforms identify systemic risks and introduce appropriate safeguards. This approach could eventually influence how companies design recommendation systems across several markets. It also places pressure on technology companies to demonstrate that their algorithms are not simply optimised for commercial performance without adequately considering user welfare.

The Larger Question Is Who Controls Digital Attention

The TikTok investigation raises a fundamental issue about the power of technology companies over everyday information consumption. Social media platforms are no longer merely providing spaces where users communicate and share content. Their recommendation systems increasingly determine what people discover, which subjects receive attention and how frequently particular ideas or experiences appear.

This influence is particularly important because users generally have limited knowledge of how recommendation systems operate. They see the videos presented to them but cannot easily understand why one subject receives repeated attention while another disappears from their feed.

Greater transparency could help address this imbalance. Platforms may need to explain more clearly how user behaviour influences recommendations and provide controls that allow individuals to make meaningful changes to their content experience.

However, transparency alone will not resolve the commercial conflict. The companies developing these systems still have strong incentives to maintain engagement, and regulators must determine how far those incentives should be allowed to shape product design.

The European Union’s action against TikTok could become an important test of whether governments can hold technology companies accountable for the consequences of their recommendation systems. The outcome may influence not only the future of social media but also the wider relationship between artificial intelligence, commercial interests and individual control over digital information.

(Adapted from Politicco.eu)