The Numbers: AI Chatbot Use Already Has a Toehold Before Age Ten
Common Sense Media's 2026 census, released November 3, found that 49% of US children aged 9 to 12 have used an AI chatbot or companion, alongside 77% of teenagers aged 13 to 17. Across the full sample of 1,211 nationally representative children aged 9 to 17, 65% have used one. This is the first year the annual census tracked AI chatbot use alongside television, online video, gaming and social media, and the numbers land well ahead of where most school AI policies currently sit.
The age breakdown is the detail worth sitting with. Use climbs from 36% among nine-year-olds to 55% among eleven-year-olds, meaning a meaningful share of elementary-age students are already interacting with AI chatbots before most districts have finished drafting acceptable-use guidance for that age band. Smartphone ownership tells a similar story, rising from 24% at age nine to 57% at age eleven, so the access that enables chatbot use is arriving earlier than policy conversations have generally assumed.
Heavy Use Without Heavy Enjoyment
About a quarter of children aged 9 to 12 use AI chatbots at least weekly, compared with 47% of teenagers, so frequency is substantial even in the youngest band surveyed. But frequency and enjoyment diverge sharply. Only 15% of younger children and 11% of teenagers say they enjoy these tools a lot, which Common Sense Media summarized this way: heavy use doesn't always mean heavy enjoyment. Founder and CEO Jim Steyer put the broader point even more directly, noting that time spent with media and the value young people get from it are two different measures entirely.
That distinction matters for anyone sizing the actual engagement or dependency risk of AI tools in a young user base. Median daily entertainment media use, excluding homework, runs about 8 hours for teenagers and 5 hours for younger children, days Common Sense describes as crowded with screen media, leaving little time for other activities. High usage numbers alone should not be read as evidence that chatbots are delivering proportional value, and institutions building policy around usage metrics need a sharper measure than time-on-tool.
The Access Gap Nobody Closed Along the Way
AI exposure is not landing evenly across income levels, even as smartphone access converges. Home computer access sits at 92% in higher-income households versus 83% in lower-income households, and personal computer ownership splits 56% to 46% along the same line. Smartphone access, by contrast, is similar across income groups, meaning most children's primary path to an AI chatbot is likely the phone in their pocket rather than a household computer their family chose and can more easily monitor.
That matters for any school or vendor assuming that managed, school-issued devices are the primary access point worth governing. If students are reaching chatbots mainly through personal smartphones, device-level controls on school-issued Chromebooks or tablets address only part of the exposure. Institutions and vendors designing safety features around managed devices should treat that control as a necessary layer among several, given where the actual access gap sits. A policy that governs only the school-issued laptop while leaving the personal phone unaddressed is governing the smaller half of a student's actual AI exposure, not the larger half most families experience at home.
Why This Is a Policy Gap, Not Just a Data Point
Most district AI acceptable-use policies were written with secondary and high school students in mind, reflecting an assumption that elementary-age exposure was marginal. This census says that assumption no longer holds. More than a third of nine-year-olds and over half of eleven-year-olds have already used an AI chatbot, which means policy committees that have not extended guidance down into elementary grades are governing a problem that has already outgrown their scope.
Dr Supreet Mann, Common Sense Media's director of research, suggested some young people treat AI chatbots as a practical tool, much like a search engine, while others hold more mixed feelings about them. That split view inside the same age band argues for guidance that accounts for how differently children in the same grade actually use these tools, a harder policy document to write but a far more accurate one.
What Edtech Vendors Building for This Age Band Should Do Differently
For edtech vendors building chatbot-based products aimed at elementary and middle school users, this data is a market-sizing correction. Demand already exists in the 9-to-12 age band at a scale most product roadmaps probably did not assume twelve months ago. But the enjoyment gap is a warning against assuming usage equals satisfaction or retention, and vendors should design evaluation metrics around comprehension and task value rather than session count or time-on-tool, both of which this data shows can run high without being meaningful.
Vendors should also build for the access reality the census describes, assuming a meaningful share of their youngest users reach the product through a personal phone rather than a managed school device. Safety features, consent flows and usage reporting designed only for managed-device environments will miss a real portion of the actual user base, and procurement teams evaluating these products should ask vendors directly how their safeguards function outside a school-managed device.
The Governance Checklist for District Technology Leaders
District technology and curriculum leaders reviewing AI policy this year should treat elementary grades as already in scope, not as a future phase. That means extending acceptable-use guidance, parent communication and staff training down to grades where a third or more of students already have hands-on chatbot experience, rather than waiting for a planned middle-school rollout to catch up with behavior that has already happened. A policy revision cycle built around secondary schools alone, which is how most districts currently structure the work, will keep missing the exact age band this census says needs attention first.
It also means asking vendors the access question directly: does a given tool's safety design assume a managed device, and what happens when a student reaches the same chatbot from a personal phone at home. Common Sense Media's census makes clear that usage, income and device access move independently of each other in ways a single blanket policy cannot address, and district leaders who build that into their next policy revision will be ahead of a curve most of their peers have not caught up to yet. Boards that treat this census as a one-time data point rather than a trigger for a formal policy update will be writing their next revision reactively, after a specific incident forces the question, rather than on their own schedule.
