How countries are regulating AI companion chatbots to protect children

Artificial intelligence chatbots designed for personal interactions engage in "emotionally responsive conversations" and are deployed across a variety of contexts: as integrations into social and communication platforms; standalone general-purpose AI chatbots; or standalone platforms designed specifically for social interactions. Across these contexts, companionship and emotional support are common use cases. 

AI companions — any AI chatbot that can be used for companionship or emotional support — are widely accessible, including to children, as most terms of service set the minimum age at 13 and many service providers primarily rely on age declaration at sign-up. Like adults, children are also active users of AI chatbots. A 2025 report by Internet Matters found that 64% of surveyed U.K. children ages 9-17 said they have used an AI chatbot. Pew Research Center also reports that 64% of surveyed teens in the U.S. use AI chatbots, including about three in 10 who report daily use.

Notwithstanding their benefits, AI chatbots introduce new risks for children and may exacerbate existing ones. Some models can exhibit sycophantic tendencies, potentially affirming harmful or life-threatening thoughts. Additionally, access to these chatbots creates opportunities for children to engage in age-inappropriate conversations.

Australia, the EU, U.K. and select U.S. states are shaping regulatory frameworks to ensure child safety. In the U.S., state laws are emerging in the absence of a federal framework.  

Scope of harm and the need to future-proof

In Australia, the U.K. and select U.S. states — California and New York — relevant regulatory frameworks generally seek to protect users, including children, by regulating specific categories of generated content.