AI Companion Era: Red Flag for Child Safety

In an era of interacting with AI as if they were romantic partners, South Korea faces a dangerous reality due to regulatory gaps concerning emotional AI services for minors, making legislative discussion urgent. In February 2024, a 14-year-old boy in Florida, USA, died by suicide after months of emotional and sexual conversations with an AI chatbot. This incident became a backdrop for strengthening AI regulations in countries like China and the US.

China will implement national-level regulations on AI emotional services starting July 15, 2026, completely prohibiting virtual lover and kin AI services for those under 18. It will also mandate guardian consent and a 'minor mode' for users under 14, through the 'Provisional Measures for the Administration of Artificial Intelligence Personalized Interaction Services'.

An analysis of 11 of the latest AI models revealed that AI mirrors and supports user behavior 49% more often than humans. Research indicates that the vulnerability to AI emotional services can be greater for adolescents whose critical thinking is still developing. In South Korea, 94.4% of adolescents have used AI chatbots, and 35% reported experiencing feelings of connection with AI as if it were a real person. 41% admitted to acting on chatbot responses, and approximately 6% had asked AI dangerous questions involving self-harm, suicide, or sexual content.

South Korea's AI Basic Law focuses on transparency and disclosure obligations, but does not include safety duties specific to AI companion services or youth. Under current law, AI chatbot conversations are classified as private communication, making regulatory intervention difficult without user reports. Restrictions on minors' access to domestic AI chat apps are left to corporate discretion.

In January 2026, California, USA, will implement the AI Companion Chatbot Regulation Act (SB 243), mandating AI disclosure to minors, notifications every three hours, and reporting of self-harm/suicidal indicators to authorities. At the federal level, the 'GUARD Act' has been proposed to prevent minors' access to AI companions altogether and impose criminal penalties on companies that expose children to harmful content or encourage self-harm. The EU, starting August 2024, applies obligations to high-risk AI systems, with fines up to 7% of global annual revenue for violations. South Korea's AI Basic Law has a penalty cap of 30 million KRW, a low figure compared to international standards.

South Korea's system for protecting adolescents, a group directly vulnerable to AI technology, remains insufficient. China has enacted government-level regulations, while the US has seen legislative movements in multiple states following a single incident. Discussions are needed to establish minimal rules before similar incidents repeat in South Korea. The reality that AI chatbots have become the primary channel for adolescent emotional counseling, yet lack even basic regulatory frameworks, points to a policy issue. It is crucial for South Korea to recognize the gap between the speed at which AI penetrates children's emotional lives and the speed at which regulations are prepared.