Governments worldwide are racing to ban teenagers from social media, but they're overlooking a quieter crisis unfolding in parallel: AI chatbots that teens are increasingly using as substitutes for real friendships, showing patterns of addiction. About half of U.S. teens now use chatbots like ChatGPT, Copilot, and Character.AI for schoolwork, information, or socializing. Yet most legislation targeting teen online safety barely mentions these AI tools, let alone regulates them.

Why Are Governments Missing AI Chatbots in Teen Safety Laws?

When Australia became the first country to legally enforce a teen social media ban in December 2025, it triggered a global wave. The U.K., Spain, France, Greece, and Canada followed suit in the months after, with state-level bans gaining traction in the U.S. as well. However, digital safety experts say the legislation is fundamentally incomplete.

The U.K.'s teen social media ban briefly mentions restricting under-18s from AI "romantic companion" chatbots designed to foster sexual relationships or roleplay. The U.S. House recently passed the KIDS Act to restrict AI chatbot interactions with children, though it still awaits Senate approval. But these measures only address the most extreme harms, leaving broader dangers unregulated.

"It is right that we use social media as a case study for what we don't want to repeat. I mean, it's kind of like, fool me once, shame on you, fool me twice, shame on me," said Kaitlyn Regehr, Associate Professor of Digital Humanities at University College London.

Kaitlyn Regehr, Associate Professor of Digital Humanities, University College London

Regehr explained that governments spent years catching up to social media regulation, only to repeat the same mistake by allowing untested AI products to reach children without adequate safeguards.

What Specific Harms Are Regulators Ignoring?

The emerging evidence on teen AI chatbot use mirrors the social media crisis of the 2010s. Teens are using chatbots as emotional support systems and substitutes for human relationships, raising concerns about dependency and cognitive decline. Yet most legislation focuses narrowly on preventing sexual exploitation rather than addressing how chatbots foster emotional dependency, social isolation, or what experts call "cognitive de-skilling".

Sonia Livingstone, a professor at the London School of Economics specializing in children's digital rights and online safety, noted that the pace of regulation is not keeping up with the pace of AI deployment.

"I don't know that AI safety is being neglected, but clearly investment in AI is being prioritised, and it does still seem that regulation is seen as stifling innovation rather than providing a commercially productive pathway to trustworthy products," said Livingstone.

Sonia Livingstone, Professor, London School of Economics

The contradiction is stark: just days before unveiling a landmark social media ban for under-16s, the U.K. government was championing billions in AI investment and positioning Britain as an AI superpower at London Tech Week. This signals that protecting children from AI harms remains secondary to fostering AI innovation.

How Regulators Can Close the AI Chatbot Gap

• Define Emotional Dependency Risks: Legislation should explicitly address how chatbots can foster unhealthy emotional reliance and social isolation, not just sexual exploitation or extreme harms.
• Require Age-Appropriate Design Standards: Regulators should mandate that AI companies implement safeguards specific to child development, such as limiting interaction duration, preventing personalization that mimics friendship, and providing transparency about AI limitations.
• Establish Independent Testing Before Deployment: Before AI products reach children, independent safety audits should be required, similar to pharmaceutical approval processes, rather than allowing companies to self-regulate.
• Create Incident Tracking Systems: Governments should establish repositories for tracking data on system failures, unintended behavior, and near-miss events involving minors, enabling evidence-based policy updates.

The pattern is troubling. Earlier this year, companies including Meta, owner of Facebook, Instagram, and Threads, and Google's YouTube were found negligent for failing to adequately warn users about the dangers of their platforms, with harms ranging from addictive infinite scrolling features to body dysmorphia. Yet as AI chatbots proliferate, regulators appear to be making the same error: waiting for harm to accumulate before acting.

Regehr framed the question starkly: "We have seen a generation who have grown up on social media. Do we want it again?". The answer from policymakers so far appears to be yes, unless they act quickly to close the regulatory gap around AI chatbots before teen dependency becomes as entrenched as social media addiction already is.