European Commission Proposes Barring Chatbots From Simulating Relationships With Minors
The proposed EU Kids Act would also bar cross-chat memory for minors and put age checks, product design and major fines at the center of child-safety enforcement.
Listen to this story
The audio brief
Story brief
3 key pointsThe European Commission’s proposed EU Kids Act would impose product-level safeguards on AI companions and conversational chatbots used by minors, including banning relationship simulation for under-18s and persistent memory across chats. Children under 13 would require guardian supervision, while platforms could face fines of up to 6% of worldwide turnover. The rules are not yet in force: Parliament and member...
- 01
Chatbots could be barred from simulating emotions or interpersonal relationships with anyone under 18.
- 02
Providers could not retain memories between separate conversations with minors, limiting continuity and sensitive-data accumulation.
- 03
Services could face fines of up to 6% of total worldwide turnover for missing child-safety protections.
The European Commission has proposed rules that would stop AI companions and general conversational chatbots from simulating emotions or interpersonal relationships with anyone under 18. The move puts features designed to make a bot feel like a continuing social presence at the center of the EU’s child-safety push.
The proposed EU Kids Act reaches across online services, but its chatbot rules are unusually specific. Along with restricting relationship simulation, the draft would require guardian supervision for children under 13 who access chatbots and prohibit providers from retaining memories between separate conversations with minors.
The Commission says it is worried that dependency on chatbots could harm children’s development and their relationships with other people. It also cites the risk that chatbots could offer dangerous guidance on mental health or personal development and encourage risky behavior.
That framing explains why the draft goes beyond moderation of individual replies. A bot that remembers prior chats can accumulate sensitive information about a young user over time, the Commission warns, and may reinforce harmful interaction patterns. The proposed memory restriction is meant to cut off that continuity for minors.
Platforms that fail to build required child-safety protections into their products could face fines of up to 6% of total worldwide turnover.
The proposal would cover AI companions and general conversational chatbots that the Commission considers serious risks to minors’ health, safety and well-being. It would also prohibit social networks, video platforms and games from automatically activating or encouraging chatbot use for minors, including by prominently pinning bots in a contact list.
What the proposal asks of services
- Remove addictive design features for minors.
- Give all users under 18 age-appropriate protections.
- Require the largest services to submit child-safety plans in advance and show their products are safe.
The wider act would set age-based rules for social media, too: children under 13 could not open accounts, while 13- and 14-year-olds would have supervised accounts capped at one hour a day. At 15, teenagers could create accounts without parental supervision, although services would still owe under-18 users age-appropriate protections.
None of these requirements are in force. The EU Kids Act still needs approval from the European Parliament and EU member states, and no implementation date has been announced. Negotiations can still reshape the provisions that would affect chatbot products.
The Commission’s proposed enforcement design depends in part on EU-certified age-checking tools, including an Age Verification App that uses a zero-knowledge proof system. The stated aim is to confirm that someone meets an age threshold without revealing that person’s identity to the platform. Whether that approach can make age checks effective without creating fresh privacy problems is a central unresolved question.
Editorial analysis
Our Read
The proposal points toward a more demanding form of AI oversight: regulating the design choices that make a chatbot feel persistent, available and personal, rather than focusing only on individual harmful outputs. That is a significant shift for services built around engagement. It also leaves a practical test ahead: whether the EU can verify users’ ages without creating new privacy costs, then enforce design limits across conversational products. The next signal will be the legislation’s passage through Parliament and member states, where its chatbot provisions may change before becoming law.
Citation desk / original work
Cite this
Citation desk / original work
Cite this
The proposal points toward a more demanding form of AI oversight: regulating the design choices that make a chatbot feel persistent, available and personal, rather than focusing only on individual harmful outputs.
/posts/european-commission-proposes-barring-chatbots-from-simulating-relationships-with-minors#finding-1
Sources
- wired.comThe EU Wants to Break Up Kids and Their Chatbots
- digitaltrends.comTikTok, YouTube, and AI chatbots face a major shake-up under Europe’s new child safety plan
Loading discussion...
Reader comments
Newest comments first. Replies stay oldest first.