On 6 February 2026, the European Commission made a preliminary finding that TikTok had breached the Digital Services Act (DSA) — not because of illegal content, not because of data misuse, but because of the way the platform was designed to keep users scrolling. Infinite scroll, algorithmic amplification of emotionally charged content, push notifications engineered to interrupt, and the deliberate absence of stopping cues: these are the features now at the center of a landmark EU enforcement action. 

For anyone working on AI governance and digital regulation, this moment demands attention. It marks the first time the EU has directly targeted what researchers call addictive design — the architecture of compulsive engagement — as a regulatory violation in its own right.

A New Kind of Digital Harm 

Until now, EU digital enforcement has largely focused on what platforms contain: illegal content, privacy violations, anticompetitive behavior. The TikTok case shifts the question to what platforms do — specifically, how they engineer user behavior at scale. The Commission's investigation focuses on design features that behavioral science identifies as drivers of compulsive engagement: variable reward mechanisms, negativity-biased algorithmic curation, and the elimination of natural stopping points that create what researchers describe as a continuous "attention capture loop."

This is a conceptually significant move. The harm being addressed is not a deceptive transaction, a stolen data record, or a piece of misinformation. It is the gradual, engineered erosion of a user's capacity to disengage — what might be termed cognitive exploitation. The potential fine of up to 6% of global annual turnover signals that the Commission treats this as a structural, not symbolic, matter. 

The Legal Gap the Case Exposes 

The DSA does not use the phrase “addictive design.” The Commission’s case rests primarily on Article 34, which requires Very Large Online Platforms (VLOPs) to identify and mitigate systemic risks to mental health and the rights of the child, and on Article 25’s prohibition of manipulative interface design. But the framework was built with a different kind of harm in mind. 

Classical consumer protection law, including the Unfair Commercial Practices Directive, is designed around discrete transactional decisions: did the platform trick a user into making a purchase they wouldn't otherwise have made? Addictive design doesn't work that way. It doesn't target a transaction. It targets presence; sustained, compulsive engagement that users struggle to terminate even when they want to. 

This exposes a fundamental limitation in how EU law currently conceptualizes autonomy. If autonomy means only informed consent plus transparency — explaining the algorithm, offering opt-outs — then addictive design will remain structurally under-regulated. Disclosing that a platform uses infinite scroll does not neutralize its psychological effects. The Commission's move toward systemic risk analysis represents an implicit, and potentially transformative, shift beyond the consent paradigm. 

A Transatlantic Convergence 

The EU is not alone in grappling with this challenge. In early 2026, opening arguments began in a landmark social media addiction trial in California — a consolidated set of lawsuits brought by families, school districts, and state attorneys general against Meta, YouTube, and others. TikTok and Snapchat settled before jury selection. The US cases are grounded in product liability and design defect theories under state tort law, asking whether engagement-maximizing platforms constitute defective products rather than protected editorial services. 

The two approaches — EU administrative risk regulation and US private tort litigation — reflect genuinely different legal philosophies. The EU model is precautionary and systemic: the Commission does not need to prove that a specific child suffered a specific psychiatric injury. It must show that platform design generates systemic risks to fundamental rights. The US model is reactive and compensatory: plaintiffs must establish individualized harm, causation, and damages. 

Yet both are converging on the same underlying question: can the behavioral engineering of attention be held legally accountable? If US courts accept that addictive design is a defect rather than a protected feature, and EU regulators require concrete interface modifications as risk mitigation, addictive platform architecture may become a transnational regulatory category. The stakes are high on both sides of the Atlantic.