AI in schools / Doorstops
How to write a solid AI policy for your school
Find your state’s rules, see how other schools wrote theirs, and see Doorstop’s policy recommendation.
The best AI policies use AI to further learning, not to cut corners. That means no AI on exams or to write whole essays, teachers setting clear rules for their own class, no punishing students on an AI detector’s word alone, and a review every year with students at the table.
Pick your school or state to see your state’s rules and a policy with your school’s name in it.
No AI on exams or to write whole essays
Exams are where students show what they know, and a whole essay is where they practice thinking it through.
Real Policy Examples
- TeachAI toolkitAI as a tutor or study aid, but not on exams unless the teacher allows it.
- Ohio model policyEach teacher spells out AI expectations in the syllabus.
Our recommended language
Students may not use AI tools during tests, quizzes, or exams. Students may not use AI to write an entire essay or other written assignment.
Teachers decide how AI is used in their class
An art class and a chemistry lab need different rules. Teachers set them; the policy makes sure students always know what they are.
Real Policy Examples
- South Orange-Maplewood, NJ (Policy 2365)Teachers set the conditions for each assignment and tell students up front.
- North Carolina guidebookRed, yellow, and green levels of AI use teachers can put on each assignment.
- Prince George’s County, MD (BP 0123)“Any use of AI must be cited.”
- Peninsula School District, WAStudents cite AI, explain how they used it, and list their prompts.
Our recommended language
Within this policy, each teacher decides how students may use AI in their class. Teachers will state their rules in the syllabus and on any assignment where they differ, and students must follow them.
our district recommends that teachers:
(a) allow AI tools for explaining and understanding concepts;
(b) not allow AI to complete homework problems or produce answers for students;
(c) state which parts of the writing process may use AI, for example brainstorming but not drafting, or rephrasing a few sentences but not inventing arguments or conclusions; and
(d) require students to disclose any AI use on homework and essays, naming the tool and what they used it for.
Teachers may grade with AI, openly, with a way to appeal
AI can save teachers hours, but students deserve to know when it was used and to have a person look again.
Real Policy Examples
- California model policyStudents are told when AI grades their work, the teacher has the final say, and parents can request a human review.
- Oklahoma model policyAI can’t be the primary basis for grades, discipline, or placement.
- Beverly Hills USD (BP 6163.47)A stricter approach: teachers may not use AI to grade or write final comments at all.
Our recommended language
Teachers may use AI tools to help grade student work. A teacher who does must tell students, and remains responsible for every final grade.
A student who believes a grade is unfair or incorrect may ask the teacher to review it within ten school days. If the student is still not satisfied, they may appeal to the head of the teacher’s department, who will review the work without relying on AI.
Every tool is vetted for privacy first
Whatever students type into a chatbot can end up stored, sold, or used to train the next model.
Real Policy Examples
- Loogootee, IN (Neola 7540.09)Vendors sign a data agreement: no training on student data, and deletion on request.
- California model policyVendors may not train on or advertise with student data; no personal accounts.
- Wichita, KS (P1231)No personal information in consumer AI tools.
Our recommended language
Only AI tools approved by our district may be used with students or student data. Before approval, each tool will be reviewed for compliance with FERPA, COPPA, state student-privacy law, and the tool’s own age requirements. Approved vendors must agree not to use student data to train AI models or for advertising, and to delete student data on request and when the contract ends.
our district will keep a public list of approved tools. Staff may not enter students’ personal information into any tool that is not on it.
AI detectors are never the only evidence
Detectors are often wrong: in one study, seven popular detectors flagged essays by non-native English speakers as AI-written 61% of the time on average.
Real Policy Examples
- California model policyNot the sole basis for discipline or grades, and students get criteria to rebut.
- Loogootee, IN (Neola 7540.09)Detectors “shall not serve as the sole basis.”
- South Orange-Maplewood, NJ (Policy 2365)Due process for the student before any finding.
- Washington OSPIGoes further: recommends teachers not use detectors at all.
Our recommended language
AI-detection software may be used as one part of an academic integrity review, but a detection result may never be the only evidence that a student violated this policy.
Before any finding, the student will be told of the concern, shown the evidence, and given a chance to respond, for example by walking through their drafts, version history, or reasoning.
If AI is allowed, every student gets access
An assignment that allows AI shouldn’t reward the students who can pay for the best tools at home.
Real Policy Examples
- Wichita, KS (P1231)If an assignment allows AI tools, they “must be made available to all students.”
- Ohio model policyNo assignment may require a tool the district can’t provide.
Our recommended language
If AI use is allowed or required on an assignment, the AI tools needed must be made available to all students, at no cost, with time to use them at school for students who lack access at home.
Revisit it every year, with students, parents, and teachers
AI changes every few months. A policy from two years ago is already out of date.
Real Policy Examples
- Prince George’s County, MD (BP 0123)The board reviews the policy every year, and students are invited onto the work group.
- Oklahoma model policyAn AI committee reviews it every year and publishes a public AI Impact Report.
- Ohio model policyThe AI workgroup includes students.
Our recommended language
The Board of Education will review this policy at least once a year. Before making meaningful changes, our district will consult students, parents and guardians, and teachers, for example through surveys, a public comment period, or an AI committee with student members.
The whole policy, ready to send
Paste it into an email to your principal or a board member, or bring it to a meeting. Edit anything in brackets.
Artificial Intelligence in Teaching and Learning
our district supports the use of artificial intelligence (AI) to further student learning, not to cut corners on it. This policy sets the rules that apply in every class and the expectations each teacher sets for their own.
1. Tests and essays
Students may not use AI tools during tests, quizzes, or exams. Students may not use AI to write an entire essay or other written assignment.
2. Classroom use
Within this policy, each teacher decides how students may use AI in their class. Teachers will state their rules in the syllabus and on any assignment where they differ, and students must follow them.
our district recommends that teachers:
(a) allow AI tools for explaining and understanding concepts;
(b) not allow AI to complete homework problems or produce answers for students;
(c) state which parts of the writing process may use AI, for example brainstorming but not drafting, or rephrasing a few sentences but not inventing arguments or conclusions; and
(d) require students to disclose any AI use on homework and essays, naming the tool and what they used it for.
3. AI in grading
Teachers may use AI tools to help grade student work. A teacher who does must tell students, and remains responsible for every final grade.
A student who believes a grade is unfair or incorrect may ask the teacher to review it within ten school days. If the student is still not satisfied, they may appeal to the head of the teacher’s department, who will review the work without relying on AI.
4. Privacy and approved tools
Only AI tools approved by our district may be used with students or student data. Before approval, each tool will be reviewed for compliance with FERPA, COPPA, state student-privacy law, and the tool’s own age requirements. Approved vendors must agree not to use student data to train AI models or for advertising, and to delete student data on request and when the contract ends.
our district will keep a public list of approved tools. Staff may not enter students’ personal information into any tool that is not on it.
5. AI detection
AI-detection software may be used as one part of an academic integrity review, but a detection result may never be the only evidence that a student violated this policy.
Before any finding, the student will be told of the concern, shown the evidence, and given a chance to respond, for example by walking through their drafts, version history, or reasoning.
6. Equal access
If AI use is allowed or required on an assignment, the AI tools needed must be made available to all students, at no cost, with time to use them at school for students who lack access at home.
7. Annual review
The Board of Education will review this policy at least once a year. Before making meaningful changes, our district will consult students, parents and guardians, and teachers, for example through surveys, a public comment period, or an AI committee with student members.
Adopted [date] by the our district Board of Education.
Getting it adopted
Where a rule lives decides who can change it. A board policy is voted on by your district’s school board and covers every school. A student handbook rule is set by your principal and covers one school. Start with whichever moves faster where you are; many students start with their principal and take a working handbook rule to the board later.
Bring numbers. A short survey of how students use AI, and how many have been accused of using it when they didn’t, makes the case better than any argument. Nationally, 51% of students worry about being falsely accused of using AI (RAND, 2025).