The 98 Teens Who Wrote the AI Policy Congress Never Did

No federal law tells American schools how to handle artificial intelligence. Congress has not passed one, and none looks imminent. So over the weekend of 17 to 19 July, 98 teenagers did the job themselves.

Meeting inside the full-scale Senate chamber replica at the Edward M. Kennedy Institute in Boston, students from all 50 states debated, amended and ultimately passed the STUDENTS FIRST Act by a vote of 82 to 16. The event, called America's Youth AI Festival, was organised by the Kennedy Institute, AASA (The School Superintendents Association) and Day of AI, a nonprofit that grew out of MIT's RAISE AI literacy initiative and is run by Jeffrey Riley, a former Massachusetts education commissioner.

The bill has no legal force. But AASA has said it will distribute the text to roughly 10,000 school superintendents nationwide, many of whom are writing their own AI policies with no federal or, in most states, even statewide template to follow. That makes this exercise worth reading closely, not as a curiosity, but as a preview of what a genuinely deliberated, student-informed policy looks like when the people most affected by it are given the pen.

What the bill actually says

The five-section act is more specific than most district policies currently in force. Its core provisions:

• Writing after year 8 is off-limits. Students would be barred from using AI to produce written work from ninth grade onward, though brainstorming, editing and studying with AI tools remain permitted throughout.
• Tests are AI-free, full stop. No AI use is allowed on any graded test, at any grade level.
• Suspicion triggers a conversation, not a verdict. A student suspected of misusing AI can be required to give an oral defence of their work, and any case flagged by a detection tool must be personally investigated by a teacher — not acted on from the software output alone. Contested cases go to two staff members, not one.
• AI literacy starts on day one of device access. Schools must teach students about misinformation, bias, privacy, plagiarism and the environmental footprint of AI as soon as they are given classroom devices, not as a bolt-on unit later.
• Teachers must publish their rules each term. Educators can use AI themselves for lesson planning, practice materials and feedback, but must post a clear, written classroom AI policy every semester so students know where they stand.
• Chatbots that detect distress must escalate, not comply. If a school-provided AI tool identifies a student in emotional crisis, it must alert staff rather than simply continue the interaction. Iowa delegate Dylan Dornack, who pushed the provision, called it potentially "lifesaving."
• Parents get a voice, not a veto. An early draft gave parents full authority to block their child's AI access; delegates struck that down after arguing, as one put it, that limiting students "because a lot of adults don't even know what AI is" would leave them unprepared. The final text asks parents to encourage balanced use and reinforce academic honesty at home, but stops short of granting opt-out control.
• Vendors owe disclosure. The bill asks AI companies to disclose bias in their tools and affirms, in its preamble, that no algorithm replaces a teacher's human connection with students.

"We missed the mark with social media and phones," said Ohio delegate Hunter Wurzel, one of the student senators, framing the weekend's work as an attempt to get ahead of a technology shift the last generation of policymakers arguably didn't.

Why this matters more than a student-government exercise

Compare this to what's actually landed on teachers' desks this year. Ohio's mandate, which took effect 1 July, requires districts to have a policy but leaves the substance almost entirely up to local boards. California's model policy, released in June, is deliberately non-prescriptive. Most district policies NeuralClass has reviewed this year default to vague language — "use AI responsibly" — that leaves individual teachers to improvise rules on the fly, then absorb the fallout when a detector flags an essay wrongly.

The STUDENTS FIRST Act, by contrast, makes hard calls. Banning AI writing outright after eighth grade is a stricter line than most districts have drawn. Requiring a teacher to personally investigate before escalating a detector flag is a direct, if unofficial, answer to the due-process gap that surfaced in the Palo Alto detector lawsuit and in JCQ's UK guidance: a score is not evidence, a human check is. And putting the onus on teachers to publish policy every term, rather than once a year in a handbook nobody reads, closes a communication gap that shows up repeatedly in surveys of teacher AI use.

None of that guarantees the bill is right. Banning AI for all writing after year 8 will be difficult to enforce and arguably conflicts with the same document's insistence on AI literacy — it is hard to teach responsible use of a tool students are simultaneously forbidden from touching for its primary purpose. The provision letting chatbots escalate signs of distress raises its own privacy questions that 98 teenagers in a weekend workshop were not equipped to fully resolve, however well-intentioned.

The value here isn't that teenagers got the policy perfect. It's that they produced something specific enough to argue with — which is more than most adult-written district policies manage.

What to do

Read the bill text before dismissing it as a stunt. AASA's distribution to 10,000 superintendents means a version of this language may well land in front of your school board this autumn. If your district's current AI policy is a single vague paragraph, use the STUDENTS FIRST Act's structure — literacy timing, detector due process, semester-published classroom rules — as a discussion starter with colleagues, even if you disagree with specific provisions like the blanket writing ban.

What to watch

Whether AASA's distribution produces any districts that adopt language close to the original text, or whether it gets diluted the way most model policies do once lawyers and boards get involved. Also worth tracking: whether any of the 98 delegates take this further, since student advocacy groups that emerge from events like this sometimes outlast the news cycle that covered them.

Sources: NPR, KPBS Public Media, AASA.