Roughly half of U.S. teenagers now use AI chatbots like ChatGPT, Copilot, and Character.AI for schoolwork, information, or simply companionship, according to Pew Research Center, and a growing body of research shows some teens using these tools as substitutes for real friendships and relationships, with usage patterns researchers describe as resembling addiction. Digital safety researchers say the trajectory closely mirrors social media’s early years, when platforms scaled to hundreds of millions of young users years before meaningful regulation caught up. YourNewsClub notes the specific pattern that makes chatbots different from the social media precedent everyone keeps citing: a chatbot doesn’t just compete for a teenager’s attention the way a feed does, it simulates a relationship, which changes both the nature of the dependency and what an effective regulatory response would need to address.
Current legislation aimed at protecting minors online has been slow to catch up to that distinction. The U.K.’s teen social media law includes only narrow language restricting under-18s from AI “romantic companion” chatbots built for sexual roleplay, and the U.S. House recently passed the KIDS Act to restrict AI chatbot interactions with children, though it remains stalled awaiting Senate approval. Kaitlyn Regehr, an associate professor of digital humanities at University College London, said the pattern amounts to policymakers repeating the same mistake with AI that they made with social media a decade earlier – treating a new technology as safe by default until the harm is already widespread and hard to legislate away. YourNewsClub marks the gap between the two most-cited legal responses so far, the U.K.’s narrow romantic-chatbot restriction and the U.S.’s stalled KIDS Act, as evidence that no jurisdiction has yet produced a framework addressing chatbot dependency broadly, as opposed to its most extreme and easiest-to-legislate expression.
Sonia Livingstone, a professor at the London School of Economics who studies children’s digital rights, has separately argued that legislation isn’t moving fast enough relative to how quickly chatbot usage among minors has scaled, echoing Regehr’s concern that lawmakers are again defining the problem too narrowly – focusing on the most visible harms, like romantic or sexual roleplay, while leaving the broader pattern of emotional dependency and what researchers call cognitive de-skilling largely unaddressed in current or pending legislation.
Maya Renn, whose work focuses on the ethics of computation and access to power through technology, frames the dependency mechanism: “A social media feed optimizes for your attention through content selection. A companionable chatbot optimizes for your attention by simulating understanding and validation directly, in a one-on-one exchange that feels like it’s responding specifically to you. That’s a more psychologically potent mechanism than an algorithmic feed, and regulation written for feed-based platforms doesn’t automatically transfer to a technology built around simulated relationship.”
Jessica Larn, who studies macro-level technology policy and infrastructure impact of AI, places the regulatory-lag angle: “Every major technology cycle produces the same sequence: mass adoption among minors, years of research documenting harm, and legislation that finally arrives once usage patterns are already deeply entrenched. The chatbot cycle is moving faster than social media’s did, which means the regulatory lag, if it repeats at the same pace, will land at an even larger scale of existing usage than it did with social media.” Your News Club tracks the gap between adoption speed and legislative response as the actual metric worth watching over the next year, more than any single piece of pending legislation: social media took roughly a decade to go from mass teen adoption to serious regulatory attention, and if chatbot adoption among teens continues at its current pace, that gap could close on a meaningfully shorter timeline.
YourNewsClub clocks the absence of chatbot-specific provisions in most current child-safety legislation, rather than the existence of any single bad chatbot product, as the more consequential problem underlying this entire story: individual companies can and do add safeguards voluntarily, but a regulatory framework built entirely around social media’s specific mechanics has no structural way to catch harms that work through a fundamentally different mechanism.