As of July 15, 2026, regulators on both sides of the Atlantic have intensified scrutiny of digital platforms for two linked concerns: addictive product design and the use of artificial intelligence in recommender and conversational systems. Governments, supranational bodies and enforcement agencies are treating design choices,autoplay, infinite scroll, streaks and reward mechanics,as regulatory touchpoints because they multiply risk when combined with algorithmic personalization and generative AI.
This article synthesizes recent regulatory action, enforcement priorities and policy trade-offs in the European Union and the United States, drawing on official EU guidance and enforcement steps, US federal studies and state litigation trends to show where oversight is focused and what platforms should expect next. Key regulatory instruments include the EU Digital Services Act and AI Act, US FTC studies and rulemaking activity, and a spate of state and federal legislative proposals targeting children’s online safety and algorithmic transparency.
EU regulatory framework and enforcement priorities
The European Union treats platform design and AI as part of a single regulatory challenge. The Digital Services Act (DSA) and the Artificial Intelligence Act create overlapping obligations: the DSA targets systemic risks and manipulative interface practices, while the AI Act limits prohibited AI practices and sets duties for high-risk systems. These instruments have already prompted cross‑border investigations and formal proceedings against major platforms for issues that explicitly include recommender systems and potentially addictive features.
European Commission guidance and staff working documents have clarified that features such as infinite scroll, autoplay and gamified reward mechanics are within the enforcement lens when they produce manipulative outcomes or disproportionate risks for minors. National regulators and the Commission have used DSA information requests and audits to examine whether platforms assessed and mitigated risks tied to their recommender systems.
On implementation timing, the EU has phased AI obligations and accompanying guidance so platforms must align product design and AI governance with concrete legal dates and new compliance tools. The European Commission has also encouraged codes of conduct and issued guidelines on prohibited AI practices and protection of minors to ensure enforcement across Member States.
US regulatory activity: FTC studies, state litigation and Capitol Hill
In the United States, the Federal Trade Commission has used 6(b) studies and staff reports to map the data, algorithmic and monetization practices that fuel engagement-based product design. The FTC’s 2024 staff report on social media and streaming companies documented extensive data-driven personalization and highlighted risks to children and other vulnerable groups,an evidentiary foundation for future rulemaking and enforcement.
Parallel to federal fact‑finding, state attorneys general and private plaintiffs have filed suits alleging that platforms knowingly deployed features that encourage compulsive use and harm minors’ mental health; California and coalitions of state AGs remain especially active. Those actions signal that litigation will remain an important enforcement pathway even while Congress debates new federal rules.
Legislatively, the Kids Online Safety Act (KOSA) and related packages have driven sustained attention to age‑appropriate defaults, transparency about recommendation logic, and duties for platforms that serve minors. KOSA passed the Senate in 2024 and was reintroduced; House negotiators have consolidated child-safety measures into broader packages (sometimes labelled the KIDS Act), reflecting an active policy push to require platform changes by statute rather than only by adjudication.
How AI and recommender systems amplify addictive dynamics
Recommender algorithms and AI chatbots change the scale and speed of attention optimization. By optimizing for engagement signals, systems can create “rabbit‑hole” effects that keep users in feedback loops of ever-more personalized content,an effect that regulators now explicitly associate with harm and systemic risk. Several EU reports and enforcement probes name recommender logic as central to risk‑assessment obligations.
Generative AI introduces new vectors: personalized chatbots, automated content creation, and dynamic nudges can all be tuned to extend sessions or maximize ad exposure. Regulators are therefore considering not only UI affordances but also model‑level properties (training data, fine‑tuning objectives, and reward functions) when assessing whether a product design is manipulative or unlawful.
Empirical research cited by regulators has linked certain product mechanics,variable rewards, streaks, intermittent reinforcement and autoplay,to measurable increases in time‑on‑platform and problematic use patterns among teens. That of evidence has strengthened calls to require default safe settings for minors and explicit opt‑ins for engagement‑maximizing features.
Focus on children: age‑appropriate design and default settings
Protection of minors is a regulatory priority in both jurisdictions. The EU has published guidelines and a suite of non‑binding but influential documents that recommend disabling addictive features by default for children, restricting engagement‑focused recommenders for minors, and hardening age estimation and verification where appropriate. These guidelines are already informing national enforcement and platform compliance programs.
In the US, the combination of federal proposals (KOSA/COPPA 2.0 components), FTC findings, and state laws or litigation has produced pressure for platforms to implement age‑specific defaults,such as turning off autoplay, limiting push notifications during school hours, or disabling monetization features that target youth. The policy debate balances child protection with concerns about privacy, parental authority and free expression.
Practically, regulators expect companies to demonstrate risk‑based design choices: safety‑by‑design documentation, impact assessments for recommender systems, internal audits of reward mechanics, and accessible remediation for minors and parents. Several EU and US guidance documents now reference these kinds of technical and governance controls as minimum expectations.
Enforcement tools, corporate responses and compliance trends
Regulators are deploying a mix of tools: information requests and audits under the DSA, 6(b) studies and potential rulemaking at the FTC, state subpoenas and public‑interest litigation, as well as coordination among agencies and multistate coalitions. In the EU, Very Large Online Platforms face proportionate DSA obligations and possible fines; in the US, the FTC’s agenda includes both competition and consumer‑protection angles.
Platforms have responded variably: some have rolled back or modified prize‑and‑reward experiments, adjusted default settings for younger users, and published voluntary safety features; others have pushed back on legal interpretations that would equate algorithmic ranking with editorial speech. Expect continued negotiation between compliance teams and regulators over scope, technical feasibility and evidence standards.
Market and operational consequences are emerging: compliance costs are rising for companies that operate across multiple jurisdictions, and product teams are investing in explainability, logging, and independent audits. At the same time, adversarial litigation and judicial review,especially in the U.S.,will shape how far agencies can push design‑based obligations without new statutory mandates.
Policy trade‑offs and likely next steps
Policymakers face hard choices. Strong design restraints reduce harms but may impose costs on innovation and raise free‑speech or commerce questions. Narrow, evidence‑based restrictions (e.g., banning specific dark patterns for minors) have gained political traction because they target observable mechanics rather than broad categories of content.
Short term, expect more audits, targeted enforcement actions, and rulemaking signals,not wholesale bans on recommender systems. Medium term, legislative outcomes (KOSA/KIDS Act or state laws) will determine whether obligations become uniform across the United States; in the EU, the DSA and AI Act timelines mean compliance duties will continue to harden through 2026 and beyond.
For companies, the prudent path is to document risk assessments, adopt age‑appropriate defaults, provide transparent opt‑outs for engagement maximization, and invest in independent verification of AI models and UI flows. Regulators are signaling that defensible product design requires both technical controls and governance evidence.
Regulatory attention to addictive features and AI is not a short‑term media cycle: it is the result of sustained policy work, cross‑border enforcement and growing empirical evidence of harm. Firms that treat design and AI governance as compliance afterthoughts will face increasing legal, reputational and operational risk.
Convergence of EU and US concerns,child protection, transparency of recommender logic, and limits on manipulative design,means multinational platforms must plan for a regulatory environment that will demand demonstrable, auditable changes to both interfaces and underlying models.
Authorities on both sides of the Atlantic are therefore likely to move from descriptive inquiries to prescriptive remedies: mandatory default settings for minors, stricter transparency obligations for recommender systems, and targeted prohibitions on the most exploitative design patterns. That regulatory hardening will shape product roadmaps and compliance budgets for the next five years.
