Students in Fifty States Just Wrote an AI Policy. Most Districts Still Haven't.

Ninety-one student senators sat in the real Senate chamber at the Edward M. Kennedy Institute in Boston this July. They split into subcommittees. They drafted language, argued over it, and amended it. On August 3, they passed a bill 82 to 16.

The bill was not about a school mascot or a dress code. It was a national AI policy framework for American public schools, the STUDENTS FIRST Act of 2026, written entirely by students representing all fifty states.

Meanwhile, a few hundred miles and a few weeks apart, dozens of real districts are doing the same work with none of that process. Wake County is targeting an August board vote on its first generative AI policy. Charleston County is launching a districtwide policy this fall. Shawnee Mission, Olathe, and Kansas City, Kansas are each writing rules aimed mostly at stopping cheating and protecting student data. All good goals. None of them involved a single student in the room where the policy got written.

That gap is the story. Not the AI. The room.

What the students actually wrote

The STUDENTS FIRST Act came out of America’s Youth AI Festival, a three-day gathering at UMass Boston and MIT organized by Day of AI, MIT RAISE, AASA, and the Kennedy Institute. Student senators ran the process the way the real Senate does: committees, floor debate, a recorded vote. What they produced reads less like a compliance document and more like a bill of rights.

It requires AI-literacy instruction starting the moment a student first uses a school device, not bolted on in ninth grade. It preserves the right of students and teachers to decline AI use when they judge it is not appropriate for the situation. It requires schools to tell students exactly what data is being collected about them and why, and gives students the right to refuse collection that is not necessary. It flatly prohibits schools from using AI to profile or categorize students by ethnicity, religion, appearance, beliefs, or values. And it draws a specific, workable line on classroom use: AI can help a student brainstorm, study, or edit, with a teacher’s consent, but it cannot generate the original writing a student turns in as their own.

Compare that to what most district policies are actually built to do this fall: catch cheating and cover the district legally. Necessary work. But narrower than what the people most affected by the policy asked for when they were handed the pen.

The invisible room

Every district that writes an AI policy holds meetings. A committee. A legal review. Maybe a survey sent home to families. What almost none of them hold is a room where the students living under the policy help write it.

This is the same gap this site keeps finding wherever a big, visible decision gets made about people who never got asked. The policy is visible: a PDF on the district website, a line item on a board agenda. The process behind it, who was in the room and who was not, stays invisible. And the people missing from that room are usually the ones who will feel the policy first: the student told to put the laptop away, the teacher deciding in real time whether a tool crossed a line nobody defined clearly enough.

The fifty state senators who wrote the STUDENTS FIRST Act were not naive about AI. They allowed it. They just refused to let it touch original work, refused to let it profile them, and refused to let anyone collect their data without saying so. That is not a stricter policy than most districts are writing. It is a more specific one, built by people who actually have to live inside it every day.

What a district can borrow this week

You do not need a Senate chamber to close this gap. You need four things most policies are still missing.

Say when literacy instruction starts. Not “at some point in middle school.” At first device issue, first login, first time a student touches a school-owned account. Waiting until AI becomes a headline in someone’s classroom is waiting too long.

Write down the right to say no. Give students and teachers a real, named ability to decline AI use in a specific assignment or situation, and say who they ask. A policy that only tells people what they must do, never what they are allowed to refuse, will not hold up under a hard case.

Name the data question directly. What is collected, why, and how a student or family can push back on collection that is not necessary. Most districts are writing this in legal language nobody reads. Write it in language a fifteen-year-old could actually explain to a parent.

Draw the brainstorm-versus-generate line clearly. The STUDENTS FIRST Act’s distinction, AI can help you think, it cannot do your original writing for you, is the single most useful sentence in the whole bill. Most district policies still leave that line fuzzy, which is exactly where the arguments start in October.

None of this requires waiting for your state legislature or your general counsel to hand you perfect language. It requires deciding these four things on purpose, instead of discovering them the first time a parent calls about one.

Put someone in the room who will actually use it

The deeper lesson is not about AI at all. It is about who gets a seat when a policy that changes daily life gets written. Ninety-one teenagers proved a workable, specific, humane AI policy is possible in a weekend when the people affected are actually in the room.

Most districts will finish their AI policy this fall without a single student or frontline teacher in that room. That is not a scandal. It is just how policy usually gets written, top-down, by the people with the authority instead of the people with the experience. But it is also exactly the gap that produces a policy nobody trusts and everybody quietly works around.

Before your district finalizes its policy this fall, ask the harder question: who is missing from this room, and what would they tell you if they were in it? The fifty states of students who wrote their own answer already showed what that looks like when someone finally asks.

If your team is building this policy right now and wants the people who will actually live under it in the room while you write it, not surveyed after the fact, training is where that process gets built together.