EU KIDS Act forces AI companions off by default for children, caps social media by age

The European Commission unveiled the EU KIDS Act on September 17, a sweeping child-safety regulation covering social media, video-sharing platforms, online games, app stores, operating systems, and — for the first time in EU platform regulation — AI companions and conversational chatbots. The rule most directly aimed at AI products: companion chatbots must be turned off by default for minors and are barred from interacting in ways designed to create emotional dependency.

The Act establishes a tiered age system for social platforms. Children under 13 cannot access social media at all, though child-directed video services remain available through parent-or-guardian-managed accounts. Ages 13 to under 15 get restricted access through a parent-managed “mini account” capped at one hour of daily use. From 15 onward, minors can open and manage their own accounts without a capped time limit, though platform-wide safety defaults still apply.

Beyond the age tiers, the KIDS Act mandates “safety by design” defaults across covered platforms: restrictions on infinite scroll and variable-reward mechanics for minors, a ban on push notifications during typical sleep hours, blocking of unsolicited contact from strangers, private-by-default profiles, and accessible blocking and muting tools. Services must verify user age through privacy-protective methods, including compatibility with the EU's own age-verification app, rather than relying on self-declared birthdates.

The chatbot and AI-companion provisions land at a moment when companion apps and persistent-memory conversational AI have moved from novelty to mainstream product category, with multiple platforms explicitly designing for long-term emotional engagement as a retention strategy. By requiring these products to default to off for minors rather than opt-out, the Commission is treating emotional-dependency design in AI companions as a risk category distinct from, and arguably more sensitive than, algorithmic feeds — the previous generation of child-safety regulation focused almost entirely on social media engagement mechanics rather than conversational AI relationships.

The Commission's announcement did not specify enforcement penalties or a compliance deadline in detail, leaving open how quickly platforms serving EU minors will need to implement the default-off chatbot requirement and the broader design mandates. The Act follows a wave of EU and national scrutiny of AI chatbot safety for minors throughout 2026, and extends the bloc's child-protection framework beyond the social-media-centric approach that has defined online safety regulation for the past decade.