Skip to content

EU KIDS Act: what the proposal means for apps and AI chatbots

Filip LuchianencoUpdated 7 min read

Last updated: 2026-10-13

On September 17, 2026, the European Commission adopted the EU KIDS Act, a proposed regulation registered as COM(2026) 681 final. Under Article 2(1), social media, video-sharing platforms, app stores, online games, operating systems, AI companions and general chatbots "accessible to minors" are in scope. On October 2 the Commission announced a call for feedback. The act is a proposal: Parliament and Council have to agree on a text before any of it binds an app.

What is the status of the EU KIDS Act, and when does feedback close?

In the European Parliament's Legislative Observatory file for 2026/0286(COD), the stage on October 13, 2026 was "Preparatory phase in Parliament". No lead committee or rapporteur was named; the committee responsible was listed as "Pending final decision on the referral". Under the ordinary legislative procedure, the steps left are referral to a lead committee, Parliament and Council positions, negotiations between them, adoption by both, and publication in the Official Journal.

No target date for adoption appears in the Commission's press release. From its "Next steps" section:

Considering the clear demand for urgent and comprehensive action at Union level, it is essential that the act is adopted swiftly.

Under Article 43, the regulation would enter into force 20 days after publication and apply 6 months later; Article 5 would apply at once, and Articles 33 and 35 after 12 months. Parliament and Council can change these dates.

Two feedback deadlines appear on Commission pages. "The feedback period closes on 26 November 2026" is in the body of the October 2 news item; "until 3 December 2026 (midnight Brussels time)" is in the link below it. On the Have Your Say page for initiative 19418, where feedback is submitted, the period is "01 October 2026 - 03 December 2026". December 3 is the operative date. On October 13, 2,392 responses were listed there.

Which apps does the EU KIDS Act cover?

Services offered to people in the EU are covered wherever the provider is established (Article 2(2)); for AI companions and chatbots, the test is placing the system on the EU market or putting it into service there (Article 2(3)). An AI companion, under Article 3(5)(c), is an AI system "that provides sustained, personalised interaction or companionship which simulates or facilitates a social, emotional or interpersonal relationship with a user". A general conversational chatbot, under Article 3(5)(d), is a general-purpose AI system "capable of providing assistance across multiple domains and tasks"; specialized customer-service, technical-support, transactional and educational applications are excluded. On those definitions, a support bot in a banking app would fall outside the act and a general assistant app would fall inside it.

Under Article 8(1), social media, video platforms, online games, AI companions, general chatbots and app stores apply the minors' rules to every user and "shall only derogate from those requirements after they have established that the recipient of the service or the user of the system is an adult, by making use of age assurance". Self-declaration is excluded from the definition of age assurance (Article 3(5)(i)); on that definition, a typed-in birth date leaves the minors' rules in place.

Service Main articles Core duties
Social media, video platforms 6, 9–13 No account under 15 where a listed risk applies; guardian-set accounts for ages 13–14, one hour a day; no addictive design; safe defaults
AI companions, general chatbots 14 No designs that create emotional dependency; memory of prior chats off by default; under-13 access only through guardian tools; testing before launch
Online games 15 Addictive design and safe settings rules applied to games; guardian tools; safeguards against luring minors to other services
App stores 16 Age rating for every app; no access to age-inappropriate apps for minors; age assessment; EU age verification solution allowed
Operating systems 4, 29(6) Share an obtained age signal with in-scope providers after user consent; no circumvention

Source: COM(2026) 681 final, Articles 4, 6, 9–16 and 29.

What does the KIDS Act require of AI companions and chatbots?

Under Article 14(1), providers of AI companions and general conversational chatbots must take at least these measures:

  • (a) "avoiding design features and system behaviours that simulate interpersonal relations that are likely to create emotional dependencies";
  • (b) the safe settings of Article 11, including that by default the system "does not use information or analysis derived from a minor's prior interactions in subsequent interactions, except where necessary to protect the minor's safety";
  • (c) transparency of economic transactions under Article 13;
  • (d) access for under-13s "only enabled and controlled by means of the tools for guardians";
  • (e) testing for risks to minors "prior to the system's placement on the market or putting into service";
  • (f) post-market monitoring, except for micro and small enterprises.

From recital 32: "the persistent conversational memory of interactions with minors should be disabled by default except where necessary to protect their safety". Under Article 14(2), a chatbot built into a social network, video platform or game "is not activated automatically", is "not displayed prominently", and minors can opt out at any time. From the press release: "AI companions and chatbots must be turned off by default".

These duties add to the disclosure rule in AI Act Article 50, applicable since August 2, 2026. In California, companion chatbot rules for children apply from July 1, 2027 under SB 1119. In Italy's Character.AI decision, the Garante applied the GDPR to a chatbot's age gate.

What must app stores and operating systems do?

Under Article 16, an app store must age-rate every app, keep minors from apps rated above their age, assess the user's age "in accordance with Chapter V", publish its rating method, and allow the certified EU age verification solution "to be offered in their store". Under Article 28(4), the store, and the services listed in Article 8(1), may keep the age signal at account level "for the sole purpose of avoiding repeated age assurance".

Under Article 29(6), an operating system provider that "has obtained an age signal of a user" through compliant age assurance "shall enable the sharing of the age signal with providers in scope" after the user consents. The technical rules are left to a delegated act (Article 30(3)(b)). Google and Apple already ship age-range interfaces for US state laws: Play Age Signals on Android and Declared Age Range on iOS, compared in our post on app store age signal laws.

What does Article 28 mean for ad and analytics SDKs?

Article 28(1) of the proposal:

Age assurance solutions shall not enable the identification of the recipient nor locate, track, target, advertise to or profile recipients for any purpose.

Under Article 28(2), providers "and any entity acting on behalf or together with such providers in age assurance" may process only the personal data "strictly necessary" to check the age threshold, and may not combine it "with personal data from any other services offered by the provider or from third-party services". From Article 28(3): "Any age assurance measure shall be zero knowledge proof." On that wording, a vendor that runs the age check inside an app, through an SDK or a web view, is an entity acting on the app's behalf in age assurance. Whether an ad SDK that only receives the age result acts "together with" the provider is left open in the draft text.

Under the safe settings of Article 11(1), which also apply to chatbots and games through Articles 14(1)(b) and 15(1)(b), "geolocation and other tracking features", microphone, camera and push notifications are off by default for minors. Under recitals 27 and 29, push notifications "should never be sent to children" during school time or core sleep hours (at least eight consecutive hours between 22:00 and 08:00), except urgent security alerts and messages from guardians.

Who enforces the KIDS Act, and what are the fines?

Enforcement is split by service type in Article 34:

  • Social media, video platforms, video gaming platforms and app stores: Chapter IV of the Digital Services Act (34(1)), with fines up to 6% of worldwide annual turnover under DSA Articles 52(3) and 74(1).
  • AI companions and general chatbots: Chapter IX of the AI Act (34(2)), with fines under AI Act Article 99, capped by Article 34(2) at "6 % of the total worldwide annual turnover" where the provider "acted intentionally or negligently". The caps in Article 99 itself are 7%, 3% and 1%.
  • Video games other than gaming platforms: a national authority designated by each Member State, with penalties set by Member States under DSA Article 52 (34(5)).
  • Age assurance data rules in Articles 27 to 29: data protection authorities, up to the GDPR Article 83(5) cap of €20 million or 4% of turnover (34(6)).

Operating systems, whose age signal duty is in Article 29(6), are absent from the lists in Article 34(1), (2) and (5).

Under Article 35, for very large platforms and AI systems under the Commission's exclusive supervision, the Commission "shall endeavour to" issue preliminary findings within 30 working days (bracketed in the draft) and a final decision within 90 working days.

What does a CITT scan and capture show for the KIDS Act?

Until the act is adopted, any comparison is between an app's current build and a draft rule. When I scan an Android or iOS build with CanITrustThat (CITT), these findings appear in the static report, each with the file it was found in: an LLM provider API host in the code (api.openai.com, api.anthropic.com or generativelanguage.googleapis.com); on Android, an LLM host in the same build as an ad SDK; the Play Age Signals library or Apple's DeclaredAgeRange framework; and a reference to the Play Age Signals API inside an ad, attribution or analytics SDK class. That is presence in the build; whether the code runs is established by a device test.

In a traffic capture on a test device, each request is logged with what it contained (the advertising ID, coordinates and other fields), the host that received it, and whether it was sent before or after consent. That log is the evidence for Articles 11 and 28: whether the advertising ID and location are sent to ad hosts, and which hosts receive requests while an age check runs. What a recipient does with the data after receipt is established only by the recipient's own records. See what a scan checks and the limits of a scan.

I run these investigations for lawyers and for researchers working with lawyers, on one app or a list of apps in bulk. Sign up for a CanITrustThat account to run your own research, let us run an investigation for you, or read the app teardowns.

App privacy law, applied to real apps.

Posts on new laws, fines and studies, and teardowns of the apps those laws apply to.

How we use your address: privacy notice. Prefer a feed reader? Use the RSS feed.