The Numbers Behind the Headline
Fordham's survey, fielded from February through March 2026 and released this fall, is one of the larger nationally representative looks at how students actually use the devices their schools issued them for classwork. Just over three-quarters of respondents said they misuse technology during school hours in some form, and roughly one in five said they used AI specifically without teacher permission at some point during the year. Students reported spending an estimated 900 minutes a week, more than 62 percent of total in-class time, actively on school-issued devices rather than on the paper materials or discussion a lesson plan might assume.
The breakdown of AI misuse is the part that should concern anyone building or buying AI-literacy curriculum for a district right now. Among students who used AI for schoolwork specifically, 48 percent used it to solve math problems outright, 36 percent to summarize or avoid reading assigned books entirely, 29 percent to draft essays or reports from scratch, and 26 percent to get help on quizzes or tests in real time. That is not casual experimentation with a new tool. It is direct substitution for the specific cognitive work each assignment was originally designed to produce in the student.
Off-Task Use Is Not Limited to AI
AI misuse sits inside a much wider pattern of off-task device use that predates generative AI by many years. Over half of students use their devices for social media during the school day, a similar share use them to message friends throughout class, and about a third use them for games or for YouTube content entirely unrelated to the assignment in front of them. Two-thirds confirmed YouTube is accessible on their school-issued devices at all, which suggests filtering policy broadly, not just AI-specific policy, remains the weaker link in a great many districts today.
This matters for how technology leaders should scope the problem internally before buying a new tool. A district that treats AI misuse as a standalone risk to be solved with an AI-specific detector is likely missing the broader device governance gap the Fordham data actually describes in detail. The fix Fordham's researchers point toward is not a smarter AI detection layer bolted onto an existing device fleet; it is tighter, more consistent control over what the device can reach on the open internet in the first place, at the network level.
Phone Bans Show a Clear Dose Response
The survey found a measurable, graded relationship between how strictly a school restricts personal phones and how much off-task device use students go on to report afterward. Social media use was lowest when phones were fully banned, such as locked away in a pouch or held at a central office for the entire school day. Use climbed noticeably when phones were permitted in hallways and at lunch, and climbed further still when the restriction applied only narrowly during class time itself, leaving every other part of the day unrestricted for personal device use.
That gradient gives districts a lever that does not require any new vendor spend at all: enforcement intensity on an existing phone policy correlates directly with measured misuse in the data. It also complicates the politics considerably, since some high schoolers actively oppose full-day bans on safety grounds, citing concerns about reaching family or responding during an emergency such as a school shooting. A September Brookings Institution report cited in the coverage found that cellphone ban enforcement tends to weaken noticeably over the course of a school year, with full bell-to-bell bans proving far more durable than partial ones.
Fordham's Prescription Is Deliberately Low-Tech
Brian Fitzpatrick, Fordham's national research manager and the report's author, wrote that the survey connects "rampant distraction, widespread AI misuse, heavy dependence on screens" as related symptoms of the same underlying access problem, noting that "school-issued devices provide direct access to the platforms driving it." His recommendations skew heavily toward reducing overall device dependence rather than adding new AI-detection software on top of the existing fleet: more supervised, in-class paper-and-pencil and oral exams, stricter consequences for late work, firewalls built directly into school-issued devices, and meaningfully tighter personal-device bans district-wide.
Fitzpatrick added a caveat enterprise vendors selling into this market should sit with carefully: "Though we might quibble about the details, we at Fordham agree with the impulse" behind stricter restriction, even while acknowledging real tradeoffs exist. That is a researcher explicitly endorsing device restriction as the default policy posture, rather than a vendor-supplied AI governance layer, which should temper any sales pitch built primarily around detection tools as the solution to this specific, well-documented problem.
The Pattern Is Already Showing Up in District Policy
The Fordham findings are not landing in a policy vacuum. Los Angeles Unified restricted AI platform access for all students for the entire 2026-27 school year while it conducts a formal review of instructional use, and New York City Public Schools separately paused AI tools outright for elementary and middle school students. Both moves either predate or run parallel to Fordham's release, suggesting districts nationwide are independently converging on restriction as a first move, largely regardless of what any single survey ultimately recommends to them.
For edtech vendors selling into K-12 right now, the strategic read is that the sales conversation is visibly shifting from access and adoption metrics toward containment and proof of controlled, auditable use. A vendor whose pitch still centers primarily on frictionless student access, rather than demonstrable restriction and teacher-mediated use, is increasingly out of step with what district decision makers are now telling researchers and reporters, on the record, that they actually want from a vendor relationship going forward.


