You tap the screen for a quick break and half an hour later you realize your feed has a theme. I remember reading an internal probe that tracked a teen through streams of sadness and self-harm — and how the company’s own tests had left him in a control group with fewer safety nudges. That moment is exactly what Brussels says it will no longer tolerate.
I’ve followed the Digital Services Act fight for months; you probably have a feed that can prove why it matters. The European Commission now says TikTok’s recommendation engine funnels users into “rabbit holes” and has ordered ByteDance to change course or face penalties that could amount to 6% of annual turnover.
A European official spoke bluntly at a press briefing about business incentives.
The EC’s headline finding: TikTok’s recommendation algorithm increases exposure to clusters of content — such as material about body image, isolation, or self-harm — without adequate assessment of harm. Henna Virkkunen, the EC’s lead on tech policy, framed the issue as a business model problem: engagement-driving features like infinite scroll, autoplay, push notifications and hyper-personalized recommendations are designed to keep eyes on the app.
The Commission argues the problem is not whether posts are illegal; it is that the continuous serving of the same material can amplify harms, especially among minors and vulnerable users. Regulators warn this effect can intensify rumination — the repetitive, hamster wheel-style thinking linked in studies to anxiety and depressive symptoms.
What are “rabbit hole” algorithms?
At its simplest, a rabbit hole recommendation repeatedly serves more of what you just watched. The EC says TikTok’s system does that in a way that can trap young users in narrow, harmful loops. You’ve seen the mechanics: the more you engage, the more the feed narrows.
A UK charity survey found almost half of teenage girls had seen harmful content in the last week.
Groups like the Molly Rose Foundation (named after a teenager whose father has campaigned about social media’s role in suicide) say those numbers show reach, not just frequency. The Guardian and other outlets have amplified those findings; an internal TikTok review obtained by Bloomberg reportedly showed a tragic case where a 16-year-old who later took his life had not been included in safety-treatment groups during experiments.
Scholars disagree about cause and effect. A University of Pennsylvania study on YouTube suggested users, not algorithms, sometimes drive filter bubbles in the political sphere — and that outcome doesn’t automatically translate to content about body image or self-harm. The evidence is growing, but it is not uniform.
A Reuters-style headline is already in the pipeline: regulatory escalation.
Brussels’ preliminary findings in February flagged infinite scroll and autoplay as likely DSA violations; Friday’s step says TikTok broke the law by not conducting adequate harm assessments and not fixing recommendation flows. The Commission now expects concrete fixes. If TikTok refuses, fines could hit 6% of global turnover — a levy calculated on worldwide revenue rather than a single market.
Can the EU actually fine ByteDance 6% of turnover?
Yes. The DSA gives the Commission powers to sanction systemic failures by very large online platforms. The percentage is steep by design: regulators want penalties large enough to change corporate behavior rather than be treated as an operating cost. TikTok has denied the conclusions, calling some of the Commission’s statements inaccurate and defending its safety tools.
A lobbyist in Brussels reminded me that platforms don’t operate in a single legal universe.
TikTok has filed a separate challenge in California alongside Google and Meta against a state law that bans “psychologically exploitative features” for under-16s. That legal fight in the U.S. will run parallel to the EU’s enforcement. Expect pile-on campaigns from industry groups, algorithm researchers, child-safety NGOs, and major outlets like The Guardian and Reuters as each side tests legal, technical, and PR arguments.
At the heart of this is a simple product question from a parent and a regulator.
How should recommendation systems weigh engagement against wellbeing? The Commission argues you can’t outsource that decision entirely to machine learning tuned for clicks and watch time. TikTok counters that it already offers user controls and content moderation tools and that the Commission mischaracterizes its systems.
The algorithm is a siren, the Commission suggests, drawing users in with increasingly narrower content. The app’s infinite scroll is a treadmill that keeps pace with attention metrics and rewards more of the same; both metaphors explain why regulators see a structural problem rather than a series of isolated posts.
A data scientist I spoke with said fixing algorithms is as much technical as it is ethical.
Possible remedies include tighter default settings for minors, broader safety-treatment cohorts in testing, throttles on recommendation reinforcement, and transparent auditing tools. Platforms like TikTok could use third-party auditing frameworks, standardized metrics for wellbeing, or sandboxed experiments similar to those discussed in academic papers. Watch for increased involvement from researchers who use tools like CrowdTangle-style data, and from policy figures pushing mandatory transparency.
An internal memo, a regulator’s press release, and a lawsuit sit on my desk figuratively.
On one side stands the European Commission with legal teeth and public appetite for enforcement. On the other, a platform that drives massive attention through personalized feeds and says it protects users. You should expect both to escalate: Brussels with fines and orders, TikTok with litigation and public-relations pushes. Google and Meta are watching — and filing their own suits in the U.S. — because the precedent will matter to every algorithmic feed that monetizes attention.
This article contains discussions of suicide. If you or someone you know is having a crisis, please call the National Suicide Prevention Lifeline at 1-800-273-8255 or text the Crisis Text Line at 741-741.
Will regulators force a redesign that changes how you use social apps, or will the industry treat fines as a cost of doing business and carry on?