Yes, with limits. Students should read your AI policy before you adopt it, should draft the parts that are about their own work, and should be told plainly about the rules the law has already settled. The reason is not fairness, though fairness is a good reason. A student is the only person who can tell you whether your policy works, because the student is the one it is written for.
In July 2026, about a hundred high school students did what most adult committees have not. They wrote a national AI policy for schools, argued over it in a replica of the U.S. Senate chamber, and passed it. The adults who run school systems are now being asked to read it.
What did students write when they were asked?
They drew the line at doing the work, not at using the tool, and the line they drew was firm. The STUDENTS FIRST Act of 2026 was drafted at America's Youth AI Festival in Boston on July 17 to 19, with 98 students from all 50 states, according to Education Week, and hosted by Day of AI, MIT RAISE, AASA (The School Superintendents Association) and the Edward M. Kennedy Institute for the U.S. Senate. AASA's release of August 3, 2026 reports that it passed 82 to 16.
The provisions repay a slow read. Students would ban AI from generating written or artistic assignments at every grade, and from graded assessments. From ninth grade they would allow it as a supplementary aid for brainstorming, studying and editing, with teacher permission. They wrote in a right to appeal an accusation of improper AI use, with human review. They would bar schools from using AI to determine grades or discipline on its own. And they would let students show mastery through oral defenses, handwritten work or in-person discussion.
Notice what is missing: a demand for freedom. Nothing in that list says "let us use it for everything." It says: tell us exactly where the line is, and do not punish us on a guess.
The students were also unsentimental about their own limits. Education Week reports that they found the parent provisions hardest to draft, because a school has no authority over a home. And the group was not a sample of American teenagers. AASA says select students served as Student Senators, so treat the Act as evidence of what a motivated group of students will write when given a real task, not as a poll of what students think.
What do students know that your policy does not?
They know whether your rules exist. The strongest evidence comes from asking a representative group instead of a motivated one. RAND's American Youth Panel, a nationally representative survey of 1,214 young people aged 12 to 29 fielded in December 2025 and published on March 17, 2026, found that AI use for homework among students in middle school and above rose from 48 percent in May 2025 to 62 percent in December. In the same survey, 67 percent agreed that heavier AI use for schoolwork will harm critical thinking skills, more than ten points higher than ten months earlier.
Then comes the finding that should interest anyone who has just finished a policy. RAND's press release says only about one third of students reported that their school had a schoolwide AI policy. The RAND report adds that higher-grade students were more likely to say the rules depended on the teacher, and more likely to worry about being accused of using AI to cheat.
Adults are not much better informed. An EdWeek Research Center survey of 693 teachers, principals and district leaders, conducted from September to November 2025, found that a majority said their school or district had no policies on generative AI use or did not know whether any existed.
So here is my reading of the evidence, which is interpretation and not a finding. A policy can be adopted, minuted and published and still not exist in the part of the school where it has to work. Students are the quickest reliable instrument for finding that out. The evidence here is American, and the logic travels to other systems better than the numbers do.
The Read-Back Test
The Read-Back Test is three questions I ask students to answer about a draft school AI policy before it is adopted: What does this let me do tomorrow? What does it stop me doing, and could I explain why to a friend? And if someone accused me of breaking it, what would happen next? A policy that students can read back correctly is working. One they cannot is a document the school has written for itself.
I offer it as my suggested way of checking a policy, not as a validated instrument. Each question catches a different failure. The first finds the policy that is so cautious it permits nothing, which students then ignore. The second finds the rule nobody can explain, which is the rule that gets broken without anyone meaning to. The third finds the most consequential gap of all, the missing process, and it is the one the students in Boston wrote into their Act as a right to appeal. If you have not yet written down what happens when a student is accused of using AI, a student reading your draft will find that hole in about ten seconds.
Which parts should students draft, read back, or leave alone?
Students should draft the rules about their own work, read back the rules about their rights, and be told, not asked, about the rules the law has already settled. Asking students to vote on a safeguarding duty or a data protection obligation is not consultation. It is theater, and children can tell.
| Part of the policy | Students' role | Why |
|---|---|---|
| When AI helps learning and when it replaces it | Draft | They know what the shortcut feels like and where it tempts them |
| What counts as their own work, and how to disclose AI help | Draft | A rule they wrote is a rule they can explain |
| How an accusation is handled and how to appeal | Read back | They hold the stakes, you hold the duty of fairness |
| Examples, wording, where the rules are published | Read back | Only a reader can tell you the sentence is unclear |
| Safeguarding, student data, legal duties | Be told | These are settled by law and by your leadership team, not by vote |
For the second row, the same logic sits behind how a head of department should handle AI in coursework: a declaration only works after a conversation about what it means.
What I Tell Leadership Teams
That the order of the work is usually backward. Across the leadership teams I work with, the pattern I see most often is that the policy is drafted by adults, shown to governors, explained to parents in a letter, and presented to students in an assembly. The students are the last to see it and the first to be held to it. I taught secondary students before I advised anyone on policy, and the lesson from the classroom holds: students will tell each other exactly where a rule is unfair or unclear, in the corridor. The only question is whether any of that reaches you.
The fix is a small change of sequence. Run the read-back before the vote, not after the complaint. It also pairs naturally with the family version of the same exercise, which is the Kitchen Table Test for the letter you send home: one check from the student's chair, one from the parent's.
What a 45-minute read-back looks like
Here is a hypothetical example, labeled as one: a secondary school with a draft policy that its leadership team likes. It asks a dozen students from different year groups, chosen by the deputy head to include the confident and the quiet, to give 45 minutes. Each student gets the draft one hour earlier. No teacher from the department that wrote it is in the room.
The facilitator asks the three questions of the Read-Back Test and writes what students say, word for word, without correcting them. Then comes the part that stings: a student who cannot answer question three, "what would happen next if I were accused," is not a student who needs retraining. That is the policy's defect, and it goes on the leadership team's list with an owner and a date.
Three limits keep this honest. A dozen students is a sounding board, not a mandate. Students can give you the pupil's view of a rule, but they cannot give you the legal view, and they should not be made to carry the decision. And if you ask and then change nothing, you will have taught the group that asking is a ritual, which is worse than never asking.
Where this goes next
The students in Boston described their Act as a starting point. They are right, and the same goes for yours. Treat the first read-back as version one, schedule the second for the end of the year, and keep the record of what changed because of it. Your school's AI steering group is the natural owner. If you want a plain document to test, a short school AI policy is easier for a student to read back than a long one.
If your leadership team is working out who should be in the room when your AI rules are written, this is the kind of work I support through AI training for schools and leadership sessions, and through keynotes for education leaders when a whole staff or system needs the conversation.
Sources and further reading
- AASA (The School Superintendents Association), Students from All 50 States Produce National Framework for AI in America's Schools, press release, August 3, 2026.
- Education Week, Lauraine Langreo, Students Created a National AI Policy Framework. Here's What It Includes, August 7, 2026.
- RAND Corporation, Heather L. Schwartz and Melissa Kay Diliberti, More Students Use AI for Homework, and More Believe It Harms Critical Thinking: Selected Findings from the American Youth Panel, March 17, 2026 (1,214 young people aged 12 to 29, December 2025 survey).
- RAND Corporation, Student Use of AI for Homework Rises as Concerns Grow About Critical Thinking Skills, press release, March 17, 2026.
Dan Fitzpatrick is the founder of The AI Educator and writes and speaks on how leaders should respond to AI. More about Dan.
