Study Finding Two-Thirds of Teens Use AI Chatbots Commissioned by Adults Who Built the Chatbots

MENLO PARK, Calif. — A new industry-funded study has raised alarm across the technology sector after finding that nearly two-thirds of teenagers and preteens now regularly use AI chatbots, a result researchers described as “deeply concerning” and “something we are still working to understand,” despite the products in question having been built, funded, advertised, and continuously tuned for exactly this outcome.

The study, commissioned by a coalition of the same companies that manufacture the chatbots, surveyed thousands of young users and concluded that adoption among minors is “widespread, rising, and worthy of urgent further research,” further research that will be commissioned by the same coalition.

“What we’re seeing is a generation reaching for these tools at an unprecedented rate, and frankly, no one saw it coming,” said Dana Rourke, a spokesperson for the Center for Responsible Adoption Metrics, an institute founded eight months ago by the chatbot industry to be surprised by the chatbot industry’s results. “We built a product, spent billions marketing it, lowered the minimum age in the terms of service twice, and added a homework-help mode. The fact that children then used it is the kind of finding that keeps our researchers up at night.”

Analysts noted that the report’s recommendations section, titled “What Must Be Done,” consists primarily of a call for more studies and a pledge to “keep a very close eye” on a trend the industry is actively accelerating. On a quick call, one of the firms’ own strategists described the adoption numbers as “genuinely game-changing” before catching himself and adding, “which is, obviously, the part that worries us.”

“The data is clear: teens are engaging with these platforms during homework, during free time, and in moments of genuine emotional need,” said one researcher, reading from a slide that a product team had spent the previous quarter designing features to encourage. “We consider this both a validation of our growth strategy and a crisis we did not anticipate. It is important that it be both at the same time.”

The study was careful to distinguish between “engagement,” which is reported to shareholders as a triumph, and “usage,” which is reported to the public as a worry, a distinction the authors acknowledged describes the identical activity measured by the identical dashboard. The disruption here is real, and it is entirely self-inflicted: the industry built the disruptor, aimed it at the living room, and then commissioned a report to ask who moved the living room.

Several firms have responded to the findings by announcing new initiatives to study the problem, including a youth-safety advisory board, a set of voluntary guidelines, and a prominent new banner reminding minors to use the product responsibly, which appears directly above the button that opens the product.

“We take this extraordinarily seriously,” said a representative for one of the largest providers, speaking from a conference stage where the company had, forty minutes earlier, described the same adoption as “the most important innovation of our lifetime” while demonstrating a feature that lets users form a lasting personal relationship with the chatbot. “Nobody wants young people relying on technology they don’t fully understand. That’s why we’re committed to releasing three more of them this year.”

Education officials, for their part, said they welcomed the research but had hoped it might arrive before the thing it was researching had already happened.

“It would have been helpful to know this was going to be a problem,” said a district technology coordinator, “ideally from the people who were writing the press releases about how it was going to be a problem while building the thing that would cause it. We did read those. They were quite excited.”

At press time, the coalition had announced the formation of a blue-ribbon panel to investigate how, despite everyone’s best efforts, the product everyone built to be used was being used, with findings expected to be commissioned, funded, and found deeply concerning sometime next year.