The EU Wants to Break Up Kids and Their Chatbots

Chatbot developers must now secure guardian consent for users under 13 in the EU, or redesign their services to comply with new safety mandates.

The European Union has proposed new rules under the EU Kids Act to regulate children's access to AI chatbots. Children under 13 will require guardian supervision to use chatbots, and platforms must integrate child safety features or face significant fines. A key provision prevents chatbots from remembering interactions across separate conversations to avoid accumulating sensitive data or reinforcing harmful patterns. The proposed legislation also prohibits platforms from automatically activating or promoting AI chatbots for minors, such as pinning them to contact lists. These measures are targeted at conversational AI companions deemed risky for minors' health and well-being, following concerns about potential negative impacts on development, relationships, and exposure to dangerous advice. The EU is also preparing a Digital Fairness Act to address addictive design more broadly.

This regulatory move signals a growing global concern over the ethical implications of AI's interaction with minors, potentially setting a precedent for other regions. The EU's focus on specific risks like addictive design, emotional dependency, and the provision of dangerous advice suggests a more nuanced approach to AI governance than broad prohibitions. It also highlights a tension between child protection and the desire for universal safeguards against potentially harmful AI features, as noted by advocacy groups. The emphasis on built-in safety and restrictions on data retention indicates a shift towards platform accountability for the AI products they deploy to younger audiences.

The implications extend beyond chatbots to broader AI design principles, with the EU's upcoming Digital Fairness Act targeting addictive interfaces across digital services. The legislation's scope, targeting AI companions and general conversational chatbots, suggests a focus on AI types perceived to have the most direct and potentially intimate influence on users. The EU's stance on generative AI, including requirements for labeling synthetic content and addressing deepfakes, underscores a commitment to mitigating the societal risks associated with advanced AI technologies, particularly their misuse against vulnerable populations.