What 98 Teens Want in a National AI School Policy

By /
Screenprint illustration of a hand raising a gavel whose head is a cracked smartphone above a panel and a room of people raising their hands

6 min read · 1,485 words

This article was written with AI. It was drafted from the sources it cites and checked against the full text of those sources before publishing. How we make articles

Ninety-eight high school students representing all 50 states gathered in a replica of the U.S. Senate chamber in Boston (Source; Source), and over three days, July 17 through 19, 2026, they wrote, debated, and passed the Students First Act at America’s Youth AI Festival (Source). The bill passed overwhelmingly, NPR reported (Source). This version came from the students it would govern, and AASA, the School Superintendents Association, plans to share it with more than 10,000 school and district leaders, Education Week reports (Source); NPR puts the figure at 10,000 school leaders (Source).

What the STUDENTS FIRST Act Actually Requires

The student rules are stricter than a student-written bill might suggest. The bill says students should never use AI for the act of writing, though editing, brainstorming, and studying with AI are permitted after eighth grade; students may not use AI to complete artistic assignments; AI use is strictly prohibited on all graded assessments, with a zero for a student caught using it unless the teacher specifies otherwise; and students “must have permission from a teacher to use AI to assist in an assignment,” and students who use AI “must cite how and where they used it or else it will be considered plagiarism” (Source). Teachers, in turn, must use one standardized AI detector for written graded assessments and lockdown browsers for graded online assessments (Source).

Two further provisions give the bill its student-rights character. The bill’s student-rights section, whose amendments the STUDENT subcommittee proposed and the full Student Senate added by majority vote, reads: “States shall require schools to provide AI literacy instruction when students start using devices in the classroom”; students “should have the right to appeal any convictions of wrongful AI use”; and those suspected of using AI “should be guaranteed a thorough examination of their work by two school officials” (AASA). High school students can also refuse to use AI tools in class and must be given an alternative assignment (Source). Before the final vote, Connecticut’s Bena Mejia Romero told her fellow senators the bill “is about who gets a seat at the table” (Source).

Read together, the rules restrict more than they permit. AI is prohibited on all graded assessments; after eighth grade, editing, brainstorming and studying with AI are permitted, but using AI to assist on an assignment needs a teacher’s permission; and literacy instruction starts when students first use devices in the classroom (AASA). That boundary matters for what follows.

Who Wrote It, and Why Nobody Has to Obey It

Legally, the document sits closer to model legislation than to policy. Government Technology describes it as nonbinding (Source), and NPR notes that the act “isn’t actually law, any more than the teens are actually senators” (Source). The students produced the document over three days at the festival (Source).

AASA, the School Superintendents Association, plans to share the final text with its network of more than 10,000 school and district leaders (Source). Government Technology’s framing, students drafting what they think national policy should look like, highlights how unusual the provenance is (Source).

That combination may explain why a nonbinding text could still move practice. Students supplied legitimacy, having drafted rules for their own work, and superintendents supply reach through the AASA network.

Where District AI Policy Stands

New York City Schools Chancellor Kamar Samuels told principals to hold off on new technology purchases until a long-overdue AI policy is ready, writing in an email that “any policy changes … may potentially impact your technology buying decisions this summer” (Source). New York’s pause is not a rejection of AI; the city’s own policy is still being written.

On this article’s reading, a district that adopts or adapts the Students First Act gains three things at once: a defensible usage line, AI-literacy instruction tied to students’ first use of school devices, and an appeal process that predates the first accusation.

Timing gives the document unusual use. A national text shared with more than 10,000 school and district leaders (Source) before local rules are written could become a default first draft.

Some Law Schools Are Restricting Devices

Cross-jurisdictional comparison sharpens the picture. At the University of Chicago Law School, first-year students will be prohibited from using electronic devices in some classrooms this fall, with some potential exceptions, the Chicago Tribune reported, an approach that blocks AI by blocking the hardware it runs on (Source). Inside Higher Ed reports that some law schools restrict technology to help students develop oral argument skills, which AI likely won’t take over from human lawyers soon (Source).

On this article’s reading, the two sectors chose different instruments. The K-12 student framework is rule-based and assignment-scoped: it strictly prohibits AI use on all graded assessments, has teachers use an AI detector for written graded assessments and lockdown browsers for graded online ones, and gives students the right to appeal any conviction of wrongful AI use (AASA). The University of Chicago Law School’s device ban, which the Chicago Tribune reports will keep first-year students from using laptops, tablets and phones in the classroom, works through hardware instead (Source). Neither the students nor the law school, in its first-year core classes, chose permissiveness; what the students added is a right to contest an accusation.

The Literacy-Over-Limits Counterargument

A serious objection deserves a serious answer, and in its strongest form it is an argument for teaching AI literacy rather than writing AI rules. A version of it came from inside the drafting: Missouri’s Herrera argued an early draft let students use AI before they were proficient, and her subcommittee’s proposed solution was to “require schools to teach AI literacy as soon as devices are introduced in the classroom” (Source). On this view, banning AI from writing assignments trains students for a world that no longer exists, and every hour spent policing usage is an hour not spent teaching competent use.

The bill meets that critique only halfway. Its literacy mandate attaches to a student’s first classroom device, and editing, brainstorming, and studying with AI are permitted after eighth grade. But it also bans AI on all graded assessments and requires a student to have a teacher’s permission before using AI to assist in an assignment (Source). On this article’s reading, the sharpest line in the bill is around assessment: the students’ bill bans AI on all graded assessments.

What Administrators Should Do With the Text

We suggest four steps for a district. Read the final bill PDF against your current handbook. Define the appeal path before adopting any detection tool, since the bill pairs mandatory detectors with a right to appeal and a review by two school officials. Schedule literacy training to coincide with device checkout, the trigger the text specifies. Compare whatever emerges with New York City’s pending policy.

From Distribution to Adoption: Will a Nonbinding Text Stick?

Distribution is not adoption, and the distance between the two is the least evidenced link in the case for the Students First Act’s real-world influence.

Its organizers, including AASA, say they intend it to inform conversations among districts, educators and policymakers as schools continue developing AI governance policies, a framing that lowers the political cost of borrowing from it (Source).

AASA plans to share the final framework with more than 10,000 school and district leaders through its own network (Source). None of this guarantees a single adoption, but it converts the bill’s influence from speculation into a testable prediction: count how many 2026–27 handbooks echo the phrase “right to appeal.” If the count is zero by spring, the skeptics win; if it is not, a replica Senate will have legislated anyway.

References

Scroll to Top