Brussels Demands TikTok Redesign Its Feed to Break Addiction Loops

The Commission's Core Argument

The European Commission issued preliminary findings on 6 November stating that TikTok's recommendation engine violates the Digital Services Act. The regulator's argument centres on what it describes as self-reinforcing content loops: the platform's algorithm interprets engagement signals to surface similar material, and sustained interaction with themes related to sadness, body image or isolation can lead users into prolonged exposure that regulators believe poses mental health risks. The Commission used the phrase "rabbit holes" to describe these patterns, focusing attention on vulnerable adults and minors who may lack the cognitive tools to moderate their own consumption.

The findings form part of a broader investigation launched in early 2024. In February, Brussels said TikTok's design features qualified as addictive under the DSA; in July, officials challenged the platform over inadequate privacy protections for accounts held by children. The latest charge is more specific: it alleges that ByteDance, TikTok's parent company, failed to conduct sufficient risk assessments of the algorithm's potential to amplify harmful content and did not implement adequate mitigation measures.

If the preliminary findings are confirmed and TikTok does not address the concerns to the Commission's satisfaction, ByteDance faces a penalty of up to 6 per cent of annual global revenue. For a company whose estimated revenue exceeded USD 100 billion in recent financial reports, the upper bound of the fine would represent a significant enforcement action.

What the Regulator Wants Changed

The Commission's statement outlined a series of design changes it expects TikTok to implement. First, it wants the For You page to rely more heavily on explicit user preferences rather than inferring intent from passive signals such as watch time or replays. Second, it called for viewing duration to be downweighted or removed as a recommendation input, arguing that time spent is an unreliable proxy for genuine interest and can inadvertently reward content that holds attention through negative emotion rather than value.

Third, the regulator demanded that user controls become more accessible and easier to use. The Commission noted that the current "not interested" button may not be sufficient to alter the feed in a meaningful way, suggesting that users need clearer, more granular tools to shape what they see. Finally, it wants TikTok to test its algorithm regularly and transparently, submitting evidence of those tests to demonstrate compliance.

These demands represent a departure from the DSA's earlier enforcement actions, which focused on transparency reports and content moderation processes. By prescribing changes to the recommendation system itself, Brussels is asserting regulatory authority over the core product experience, a move that has implications for every platform that uses algorithmic curation.

TikTok's Defence

TikTok rejected the findings in a statement to news agencies, calling the Commission's characterisation "fundamentally inaccurate" and arguing that it distorts both the platform's architecture and its commitment to user safety. The company pointed to existing controls that allow users to filter content, set screen-time limits and refresh their For You page, framing those features as evidence of direct and intuitive control.

The company has not yet published a detailed technical rebuttal, but its response suggests it will challenge the Commission's interpretation of how the algorithm functions. One point of contention is likely to be whether engagement signals such as watch time can be separated from user preference in a meaningful way. TikTok's recommendation model, like those of most social platforms, treats repeated viewing, shares and completion rates as indicators of interest; removing those signals could fundamentally alter the feed's relevance and, by extension, user retention.

Another area of dispute is the threshold for harm. The Commission's preliminary findings do not cite specific case studies or quantitative data linking TikTok's algorithm to measurable mental health outcomes. Instead, they rest on a precautionary logic: that the design creates conditions under which harm is more likely. TikTok is expected to argue that this standard is too vague to guide product development and that it imposes a burden of proof the platform cannot meet without clearer definitions.

Why This Case Matters Beyond Europe

The DSA, which came into full effect in 2024, is the European Union's primary tool for regulating large online platforms. It imposes obligations on companies designated as Very Large Online Platforms, requiring them to assess systemic risks, provide transparency reports and give users more control over algorithmic curation. TikTok was designated a VLOP in April 2023, bringing it under the DSA's most stringent requirements.

The Act's extraterritorial reach means that design changes mandated in Europe often become de facto global standards. Companies rarely maintain separate product architectures for different jurisdictions, particularly when the alternative involves fragmenting the user experience. If TikTok introduces a mode that allows users to disable watch-time signals or opt into chronological feeds, those features are likely to appear in other markets, either because regulators elsewhere adopt similar rules or because users demand parity.

Asia-Pacific regulators have taken note. Singapore's Infocomm Media Development Authority has been consulting on updates to its Online Safety Code, and officials have cited the DSA as a reference model. South Korea's Korea Communications Commission has opened discussions on algorithmic transparency, particularly for platforms popular with teenagers. Australia's eSafety Commissioner has called for stronger age-verification and content-filtering tools, and the TikTok case provides a template for how those requirements might be enforced.

The case also signals a shift in regulatory strategy. Early internet governance focused on illegal content; the DSA extends that remit to the design layer, treating recommendation systems as a distinct source of risk. This approach has precedents in consumer product safety, where regulators assess not only the materials used but also the ergonomics and user interface. Applying the same logic to software is new, and the TikTok findings will test whether courts and appeals bodies uphold it.

The Broader Context of Platform Accountability

TikTok is not the only platform facing scrutiny over algorithmic recommendations. Meta's Instagram has been the subject of internal research, leaked in 2021, showing that the company was aware of links between certain content types and negative self-perception among teenage girls. YouTube has faced questions over its recommendation of conspiracy theories and misinformation. X, formerly Twitter, has been criticised for amplifying polarising political content through its algorithmic timeline.

What distinguishes the TikTok case is the speed and directness of the regulatory response. The Commission opened its investigation in February 2024 and issued preliminary findings within nine months, a timeline that reflects both the DSA's streamlined enforcement procedures and the political priority Brussels has placed on child safety and mental health. The findings also mark the first time the Commission has demanded specific changes to a recommendation algorithm as a condition of compliance, rather than accepting transparency measures or voluntary commitments.

California's attempt to regulate addictive features offers a parallel case study. In early November, Google, Meta and TikTok filed a lawsuit challenging a state law that would ban social media companies from allowing users under 16 to access features deemed addictive, including infinite scroll, autoplay and push notifications. The companies argue that the law is unconstitutionally vague and that it infringes on their First Amendment rights. The case is pending, but it illustrates the tension between prescriptive regulation and platform autonomy that the TikTok-EU dispute also embodies.

What Happens Next

The preliminary findings trigger a formal process under the DSA. TikTok has the right to respond in writing and to request a hearing before the Commission makes a final determination. That process typically takes several months. If the Commission confirms the findings, it will issue a decision requiring specific remedial actions and setting a deadline for compliance. Failure to comply would trigger the fine.

TikTok can appeal any final decision to the Court of Justice of the European Union, a process that can take years. In the meantime, the company is likely to implement interim measures designed to demonstrate good faith, such as piloting new user controls or publishing research on the effects of its recommendation system. Whether those measures satisfy the Commission will depend on their scope and whether they address the core concern: that the algorithm, as currently designed, amplifies content in ways that regulators believe create mental health risks.

The outcome will shape not only TikTok's product roadmap but also the regulatory expectations placed on every platform that uses machine learning to curate content. If the Commission's findings are upheld, the precedent will be clear: recommendation systems are not neutral tools but design choices subject to public interest obligations, and regulators have the authority to require changes when those systems create foreseeable harm.

For ByteDance, the stakes extend beyond the fine. The company is already navigating restrictions in the United States, where TikTok faces potential divestment requirements, and in India, where the app has been banned since 2020. A protracted dispute with Brussels would add another front to a regulatory battle that is testing the limits of platform governance in an era when algorithms, not human editors, decide what billions of people see each day.