98 Students from 50 States Wrote You an Acceptable AI Use Policy

Nearly 100 high school students representing all 50 states met up inside a replica chamber in Boston to draft a coherent AI policy for how AI should be handled in K12 schools.

The bill's shortened name, the STUDENTS FIRST ACT, ran five sections. By the end, the result was an AI policy "written by students, for students," according to Jeffrey C. Riley, Day of AI's executive director.

What does the bill actually entail? Does it really address real-world AI governance questions that the adults haven't answered?

Let’s break it down.

There's still no national AI policy for schools

As of writing this article, US states and districts are forced to create their own AI policy, which has led to uneven improvising. At least seven states, including Ohio, have already enacted comprehensive school AI policy requirements, while other states have yet to start.

That gap is the real significance. Right now, it feels like the wild west of AI management in K12. Although advancements in AI-powered web filtering are unprecedented, the pace of managing bad AI usage struggles to catch up.

That’s why 98 students came to address real-world AI governance questions the adults haven't answered. And they want us K12 IT leaders to listen.

The heart of this document is its 15 student-focused provisions, which cluster into four themes.

Four highlights from the bill on AI governance in education

1. Academic integrity is the spine

This is the most ambitious and most operationally demanding part. AI is banned outright on all graded assessments:

• Students can't paste assignment prompts directly into AI tools.

• Students can't use AI for the actual act of writing (editing, brainstorming, and studying become permissible only after 8th grade).

• Students can't use it on artistic or creative work.

• Students need a teacher's explicit permission for any AI-assisted assignment.

If AI was used during drafting, the student must prove independent mastery through discussion, a handwritten test, or an oral defense.

The distinction they're drawing is sophisticated (that AI supports cognition versus AI replacing it). The catch is enforceability: outside a proctored, device-locked classroom, an assessment ban on a take-home Chromebook is an honor code.

2. Due process for the accused

Some say this is the smartest thing in the bill.

The bill states students get the right to appeal a wrongful-use accusation, and (critically) a detector flag alone can't convict anyone; a human has to investigate first.

This matters because AI-writing detection tools are claimed unreliable, that their errors aren't random. They disproportionately false-flag English-language learners and neurodivergent students.

Written by the generation actually on the receiving end of these accusations, the safeguard is pointed and deliberate.

3. Literacy and access as rights

The students framed these as entitlements, or rights to each student.

• States must provide AI literacy the moment devices enter the classroom; closing the dangerous gap where kids get the tool before any guidance.

  
  

• Every student must have access to at least one school-approved, filtered, and monitored AI tool, which puts a floor under access so it doesn't simply track family wealth.

And AI supports can be written into IEP and 504 plans, turning assistive AI into a right for the students it helps most rather than a teacher-by-teacher favor.

4. Privacy and harmful use

Students must be told what data is collected about them and can refuse unnecessary collection; a principle aligned with where actual law is heading, though the load-bearing word "unnecessary" is left undefined.

And the bill prohibits using AI to generate deepfakes or harassing content targeting anyone in the school community, students and staff alike.

That last provision is the least hypothetical line in the entire document: schools are already dealing with AI-generated fake imagery of classmates, and the students were living in the actual present when they named it.

The three debates worth remembering

Three floor fights revealed more than the final text does, because each shows a clean principle colliding with a messy reality.

1) When should AI literacy start, and who pays?

A push to begin instruction the moment devices appear ran headlong into school-funding inequity.

One Missouri student argued a draft was too permissive because it never required proficiency before use; others from California and Tennessee countered that under-resourced schools couldn't absorb a new instructional mandate.

The students had reinvented the unfunded mandate in real time, then tried to patch it by requiring federal funding — which is roughly where such fixes go to die. The episode is a near-perfect miniature of why AI may widen the equity gap before it narrows it.

2) Should parents get a veto?

An early draft handed parents a complete veto over their child's AI access.

It didn't survive.

One New Hampshire delegate made the limiting-principle case that schools can't legislate parenting, while a North Carolina delegate made the sharper argument that a total veto could actively harm students because many adults don't understand AI well enough to make the call.

They introduced a third axis into the usual parents-versus-institution debate: competence.

The family section was ultimately softened from mandates to suggestions, which was practically wise, though it somewhat waved away the families with real, informed data objections.

3) What should school chatbots do in a crisis?

This was the weekend's most charged exchange.

If a student turns to a school AI tool for emotional support and discloses something sensitive — a mental-health crisis, or questions about their identity — should that stay private?

The privacy argument was not naive; for a vulnerable student in an unsupportive home, confidentiality can be a matter of safety.

But an Iowa delegate argued that genuine crisis requires the technology to alert a human even over the student's wish for secrecy, calling it potentially "lifesaving." The senate landed there: the chatbot should not try to handle the request itself and should escalate to school officials.

In effect, the students invented mandatory-reporter logic for software — a conceptually correct move that nonetheless leaves enormous operational and legal questions (who responds after hours, false-positive rates, the risk of outing a student, and the tangle with FERPA and state reporting law) compressed into a single sentence.

Our takeaway from the student AI policy

The honest way to describe the Students First Act is a first of its kind, built on a very old form.

The machinery (a replica chamber, subcommittees, floor debate, a model bill) is essentially the model-legislature template that YMCA Youth and Government has run since 1936 in all 50 states, alongside Boys State, Girls State, and Model UN.

That tradition even has precedent for youth-drafted bills becoming real law in some states. So students writing legislation, and even influencing statute, is not new.

The reason 98 teenagers could produce something this coherent in a weekend is precisely that they inherited a well-refined civic scaffold and pointed it at the most current problem imaginable.

What is genuinely unprecedented isn't that students wrote a bill.

Look at the fact that the School Superintendents Association (AASA) decided a student-written bill on this subject was worth placing in front of K12 decision-makers nationwide as a serious starting point.

The kids using a Senate replica is tradition, but the adults treating the output as policy-grade input is definitely a break from it.

Now, as a document, it is imperfect, obviously, and its authors said so on the floor.

Its weaknesses share a single shape: a sound principle with no enforcement mechanism or definition underneath it, which also describes a fair amount of real legislation.

But that imperfection is arguably the point.

These students did not deliver the final word on school AI policy. They delivered a remarkably thoughtful place to begin, and demonstrated that the "student agency" educators have talked about for decades can actually be operationalized when adults are willing to hand over the pen.

Students' message to school IT directors and K12 leaders

The most useful thing we as school district leaders can do is put their student AI policy of 15 provisions next to our own AI policy. Identify any contradictions, and ask our teachers and students what their perspective is on the matter.

Meanwhile, find any contradictions. And if we do, find out why our AI policy decides differently than what this student AI policy thinks.

The answer to "should students have a seat at this table" was settled in Boston. And now, the kids are finished.

The question now is whether the adults meant what they said. And yes, their Students First Act passed overwhelmingly, 82-16.