On September 17, the European Commission released a draft of the so‑called EU KIDS Act – an acronym for Keeping Internet Digital Spaces Accountable and Trustworthy. While the most heated provisions target social networks and video‑sharing platforms – for example, an outright ban for users under 13 and tighter limits for those under 15 – the text also contains a series of rules that directly affect online games, artificial‑intelligence companions, chat bots, app stores and operating systems. This article unpacks the game‑specific elements of the proposal and explains what developers should start preparing for now.
If the legislation is eventually adopted, it will apply to every online game that is offered to consumers in any EU member state. The definition of "online game" is deliberately wide; it could encompass everything from sandbox platforms such as Roblox to the multiplayer component of a traditionally single‑player, disc‑based title.
The only clear exemption is for games that are sold exclusively on physical media and have no online functionality whatsoever. A more ambiguous scenario concerns games that are not formally marketed in the EU but still attract European players through foreign storefronts, such as US‑based platforms. In those cases the regulation might still reach the service if it is deemed to be directed at EU users. There are, however, explicit carve‑outs for services whose primary purpose is education and that are run by schools or other educational bodies, as well as for certain non‑profit offerings.
One of the central duties imposed on online‑game providers is to avoid encouraging excessive use by minors – defined in the act as anyone under the age of 18. In practice this means that mechanisms that reward players for logging in daily, or that penalise them for taking a break (for instance, by removing bonuses after a period of inactivity), are likely to be prohibited for under‑18 users. While a technically compliant workaround might be imagined, the draft makes clear that such features are viewed unfavourably regardless of how they are framed. These concerns echo recent debates around age‑rating systems such as PEGI in Europe and USK in Germany, but the KIDS Act takes a less nuanced approach.
The rule does not merely target obvious loyalty schemes; it reaches any design element that effectively pressures a minor to return at regular intervals. In a persistent‑world environment, even the simple fact that the game world continues to evolve while a player is offline can be interpreted as a subtle penalty, because progress may be missed.
Consequently, designers will need to shift their thinking from "Should we give a daily reward?" to a broader question: "Does the overall experience create a schedule that draws minors back on a set timetable?" This is a far more complex assessment and may require redesign of core loops. The draft also mandates that contact between minors and other users be limited. To meet this requirement, game operators will generally need to implement robust parental‑control tools and age‑verification solutions.
Unlike social networks, which must use a dedicated EU‑wide age‑verification system, games are allowed to rely on proportionate methods such as age estimation, provided these methods are accurate, privacy‑respecting and non‑discriminatory. Pure self‑declaration will never be sufficient. When a game is distributed through an app store, part of the verification burden can be shifted upstream, because the store already checks a user’s age; with the user’s consent, that information may be passed to the game.
Another notable element is the proposed "Code of Conduct" for age‑rating and online games. The intention appears to be to turn the largely voluntary PEGI framework into a legally binding regime.
This raises questions about how the act will interact with national youth‑protection laws that already impose mandatory rating requirements – a situation that is especially relevant in Germany, where the USK system is backed by statutory law. The KIDS Act does not seem to leave much room for the USK to continue regulating excessive‑use or open‑communication risks, potentially creating a jurisdictional clash between EU‑wide rules and existing German legislation.
Monetisation transparency is also resurfaced in the proposal. Recital 40 states that minors should be given clear information about the real monetary value of virtual‑currency transactions and be protected from variable‑reward systems, a clear reference to loot boxes and similar mechanisms.
However, the operative part of the draft limits these economic‑transaction rules to social networks, video‑sharing platforms, AI companions and generic chatbots, not to online games. This inconsistency may be a drafting oversight or a deliberate decision to address game‑related transparency through the Code of Conduct rather than direct regulation.
The act further expands its scope to cover "general conversational chatbots" and, crucially for the gaming sector, "AI companions." Such systems must avoid design features that foster emotional dependence and must not, by default, reuse personal data from prior interactions with minors unless it is strictly needed for safety. If an online game incorporates an AI companion within a social‑network or video‑sharing environment, the companion may not be auto‑activated, must not be prominently displayed, and minors must be given a clear opt‑out option at any time. Not every AI feature in a game will fall under the definition of an AI companion. The key test is whether the system simulates or facilitates a social, emotional, or interpersonal relationship with the user.
For example, large‑language‑model‑driven non‑player characters (NPCs) that converse with the player’s avatar rather than directly with the human player might be exempt, because the interaction is framed as in‑world dialogue. Yet the definition also captures tools that merely enable such relationships, so developers should carefully assess whether their AI‑driven features could be interpreted as creating a bond with the real user. In addition to age‑verification, the act requires providers to give minors effective content‑control tools. Players must be able to manage the type of content they see, submit feedback, and have those choices applied instantly and persistently.
Interaction controls must also let minors decide how they communicate, create, and engage with others within the service. All safety mechanisms, informational prompts, and user‑interface elements must be easy to find and presented in language that minors can understand.
Enforcement will fall under the existing EU Digital Services Act (DSA) for most services, while AI companions and chatbots will be overseen by the AI Act. Non‑compliance could lead to fines of up to 6 % of a company’s global annual turnover. Each member state will need to designate a regulator to monitor compliance for online‑game operators. The KIDS Act is intended as a directly applicable EU regulation, meaning it would take effect across all member states without the need for national transposition.
The draft is still subject to approval by the European Parliament and the Council of the EU, a process that typically lasts one to two years. Nonetheless, the Commission appears eager to move quickly, having already presented a detailed text rather than a high‑level outline. Given the global focus on regulating social media and digital environments for children, the KIDS Act is unlikely to disappear.
Its provisions could have a tangible impact on core gameplay mechanics, reward structures, and AI features. Game developers therefore have a narrow window to evaluate the potential effects and adapt their designs. The strategic question for designers is whether to rely on robust age‑assurance systems and restrict certain features for minors, or to build games in a way that all features are universally accessible without creating age‑based penalties. Both approaches have trade‑offs in terms of user experience, compliance cost, and market reach.
Dr. Andreas Lober, a partner at ADVANT Beiten, advises leading game companies on youth‑protection and consumer‑law matters. He acknowledges contributions from colleagues Lennart Kriebel and Fabian Eckstein in shaping this analysis.