How to Read School AI Policy Documents in 20 Minutes

Illinois published 409 pages of AI guidance for schools in July. DC’s model policy runs a fraction of that. Florida’s new rule fits on a few pages but carries the force of law. Your district’s version is probably somewhere in between, buried in a board packet or a parent portal PDF. You can read school AI policy documents in 20 minutes if you know which nine questions they have to answer, and this checklist, built from those three official documents, tells you where to look and what a bad answer looks like.

Key Takeaways

• Nine questions determine whether a district AI policy protects your kid: scope, prohibited uses, student-facing access by grade, data handling, notification, opt-out, integrity rules, training, and review dates.
• DC’s September 1, 2026 model policy uses a red/yellow/green “stoplight” and covers staff only; if your district copied it, student rules live somewhere else.
• Florida’s rule, adopted September 16, 2026, is the strictest template: parental notification before AI tools, a non-AI alternative on request, no “emotional companion” AI, and 30-day interaction logs.
• Illinois’s guidance is 409 pages, non-mandatory, and openly AI-drafted in places; look for the tool-selection questions, which are the useful part.
• The most common gap in district policies is silence on student-facing AI in grades K–5 and on who sees your child’s prompts.

What a district AI policy actually is

A district AI policy is a board-approved document that says which artificial intelligence tools staff and students may use, for what, under whose supervision, and with what data protections. It is different from state guidance, which recommends, and from vendor terms of service, which govern one product. Most districts in 2026 have one of three things: a real policy (voted on, dated, with a review cycle), an “administrative procedure” (written by staff, not the board), or a paragraph pasted into the acceptable-use agreement you signed in August.

The reason this matters right now is timing. Four states (Maryland, Idaho, Oklahoma, Virginia) mandate that districts adopt AI policies. Florida joined the list on September 16, 2026, when the State Board of Education voted to require all 67 districts and 28 state colleges to have AI rules in place by July 1, 2027. DC’s Office of the State Superintendent of Education released its model policy on September 1. Illinois published its guidance on July 9. Between those dates, a lot of districts copied a template, changed the letterhead, and posted it.

Copying is fine. But templates leave holes where the local decisions were supposed to go, and those holes are where your kid’s actual experience gets decided by whichever teacher or vendor moves first.

How to read school AI policy language: the nine questions

The three official documents cover different ground. DC’s policy addresses staff use only. Florida’s rule reaches students, parents, and vendors. Illinois’s guidance covers everything but mandates nothing. Read together, they show the full set of questions a complete policy answers.

1. Scope: who does this apply to? DC’s model policy says explicitly that it “addresses staff use only, explicitly excluding student use and tool procurement.” That is not a flaw; it is a boundary. If your district adopted a DC-style staff policy, ask where the student rules are. If the answer is “the acceptable-use policy from 2019,” there are no student rules.

2. Prohibited uses: what is red? DC’s stoplight puts four things in red: physical surveillance of students and staff, student discipline decisions, teacher performance evaluations, and determining IEP or Section 504 eligibility. Yellow (limited use with safeguards) includes reviewing and grading student work, drafting IEP language, and monitoring digital activity on district devices. Green (permitted with human review) includes drafting lesson plans, customizing student materials, and analyzing data sets. A policy that has no red list has not made any hard decisions.

3. Student-facing access by grade. Florida’s rule requires “extra age-appropriateness review” for tools used in pre-K through grade 5 and bars tools “designed or configured to meet a student’s social or emotional needs.” Katy ISD in Texas banned student AI in K–6 and built supervised access for grades 7–12. NYSUT’s resolution asks for no student-facing AI in Pre-K–2. Whatever your district decided, it should say so by grade band. Silence on elementary grades is the most common gap.

4. Data: what is collected, who sees it, how long is it kept? Florida requires districts to “retain AI interaction records for at least 30 days” and to “maintain a public list of approved tools.” DC ties staff use to FERPA, COPPA, CIPA, IDEA, HIPAA, and DC’s Protecting Students Digital Privacy Act of 2016. Look for the words “personally identifiable information,” “training,” and “retention.” If the policy never says whether student prompts can be used to train a vendor’s model, the vendor’s terms decide.

5. Parental notification. Florida’s rule is the clearest standard in the country: districts must notify parents before children use AI instructional tools, identifying “the platform, the grades and subjects involved” and how students interact with it. Most districts have nothing like this. If yours does not, you will find out about a tool when your kid mentions it.

6. Opt-out and the alternative. Florida requires schools to provide a “comparable non-AI instructional alternative” when a parent opts out. New Mexico’s Public Education Department went further in August after five districts refused the Amira reading tool over voice-data concerns: schools may skip the voice-recording feature, use paper tests, or substitute their own assessment. Check whether your policy names an alternative. “Opt-out available” without an alternative usually means a worksheet in the hallway.

7. Academic integrity and detection. Florida’s rule says policies “must address plagiarism.” The unwritten question is whether the district uses AI detectors, which have a documented false-positive problem. A Wake County freshman was flagged at 62–87% by a detector in May 2026 for work she wrote herself. A good policy names the process for an accusation (who reviews, what evidence, whether the student can explain the work orally), not just the tool.

8. Teacher training. DC recommends staff training before AI use, “demonstrated AI literacy proficiency, annual training renewal.” Florida requires teacher training on “AI risks, limitations, safety and compliance.” The NPR/Ipsos poll in June 2026 found only about a third of teachers report formal guidelines at all. If the policy requires training but no one can tell you when it happened, it has not happened.

9. Review date and owner. Illinois’s guidance was written partly with ChatGPT, Claude, and Gemini, and the state says so. That transparency is a model. Your district’s policy should have a date, an owner (a named office, not “the district”), and a review cycle. Tools change quarterly. A policy from 2024 without a review clause is describing a world that no longer exists.

The 20-minute checklist

Print this, open the PDF, and work down the list. Twenty minutes is realistic for a 6–12 page policy. Illinois’s 409 pages are guidance, not a district policy; if your district posted the whole thing, you are looking for the wrong document.

Minute	Check	Where it is in DC / IL / FL	Good answer	Red flag
0–2	Board-adopted, dated, owner named	Header / signature block	Vote date within 12 months, named office	”Draft,” no date, no owner
2–4	Scope: staff, students, or both	Section 1 or “Purpose”	Both, or a pointer to the student policy	Staff only, no pointer
4–7	Red list of prohibited uses	DC “stoplight” red tier	Discipline, IEP eligibility, surveillance, evaluations	No prohibited list
7–9	Student access by grade band	FL pre-K–5 review; Katy K–6 ban	Explicit grade bands with supervision rules	Silence on K–5
9–12	Data: PII, training, retention	FL 30-day logs; DC FERPA/COPPA list	Named vendors, no model training, retention period	”We take privacy seriously”
12–14	Parental notification	FL notice of platform, grade, subject	Notice before first use	Notice only in annual handbook
14–16	Opt-out plus alternative	FL “comparable non-AI alternative”; NM paper option	Named alternative	”Contact your principal”
16–18	Integrity process, detector policy	FL plagiarism clause	Human review, oral defense option, no detector-only accusations	Detector score treated as proof
18–20	Training and review cycle	DC annual renewal	Annual training, annual review	No date, no training record

Nine checks in 20 minutes. If the policy passes six or more, it is above average for 2026. If it fails the data, notification, and opt-out rows together, your district has a staff memo, not a policy.

Why the checklist leans on Florida, DC, and Illinois

These three are useful because they represent the three ways a state can act. Florida regulates: its rule has deadlines, compliance requirements, and reporting to the Department of Education. DC models: it wrote a template that local education agencies “can tailor and adopt as they see fit.” Illinois advises: its guidance offers frameworks for tool selection and AI literacy but “provides recommendations rather than mandates,” according to Capitol News Illinois’s July coverage.

The Illinois document is worth a second look for one section. Its tool-selection questions (what data does the tool collect, does it train on student inputs, who reviews outputs, what happens when it is wrong) are the questions your district should have asked vendors. If your policy lists approved tools without evidence those questions were asked, ask them yourself.

There is one more reason to read policies from other states. The U.S. Department of Education’s August 20, 2026 Dear Colleague Letter tells districts to evaluate AI by “measurable learning outcomes, not screen time.” That is a federal signal that policies should say what the tool is for, not only what it is not for. A policy with a red list but no learning goal is half a policy.

What to do at home once you have read it

Ask the three questions the policy did not answer

Every policy leaves gaps. Email the principal with the three rows from the checklist that were blank, phrased as questions with a yes/no answer. “Does any tool used in my child’s grade store their prompts, and for how long?” gets an answer. “What is your approach to AI?” gets a paragraph. The research on parent-teacher communication is consistent that specific questions get specific answers; our guide to questions that get real answers at parent-teacher conferences covers the phrasing.

Match the policy to what your kid actually sees

Policies describe intent. Kids describe reality. Ask your child which tools opened on the school device this week. Common Sense Media’s August 2026 survey of 1,017 teens found 37% do not understand their school’s AI rules and 44% of AI users had a tool blocked at school, with 59% of those switching to a personal device. If your kid’s description and the policy disagree, the policy is not being enforced, and that is information.

Keep the vendor list

Florida requires a public list of approved tools. Most states do not. Ask for the list anyway and keep it. When a new tool appears mid-year (a Gemini tab in Google Classroom, a Khanmigo login, a reading app that records voice), you can check whether it went through any review. Our guide on what schools can see on Chromebooks explains the monitoring side of the same question.

Read the opt-out clause before you need it

If the policy offers an opt-out, find out what the alternative is now, not in the week your child is assigned an AI tool. New Mexico’s districts pushed back on Amira because a superintendent said a child’s voice “is effectively biometric data.” That objection landed because parents had raised it before the assessment window, not after.

What not to do

Do not treat a long policy as a strong one. Illinois’s 409 pages are guidance, and much of it is background. Do not treat a ban as a complete policy either; NYC’s Pre-K–8 ban says nothing about what a fifth grader does on a home laptop. And do not assume a vendor’s privacy page overrides a weak district policy. The district contract governs, and if the contract is silent, the vendor’s default terms apply.

What to Watch For Over the Next 3 Months

• Week 4: You have read the policy, sent three specific questions, and received at least one written answer. If you got no reply, escalate to the board’s public-comment period; Florida’s timeline means many districts are revising policies this fall anyway.
• Month 2 red flags: A new tool appears on your child’s device with no notice; the school cannot name the alternative for an opt-out; a teacher references an AI detector score as the sole basis for an integrity concern.
• Month 3 self-check: Can you name every AI tool your child used at school this quarter, what data each collects, and who at the district owns the policy? If yes, the checklist worked. If not, the gaps are the next email.

Frequently Asked Questions

My district only has a paragraph in the acceptable-use policy. Is that a real AI policy?

Not by 2026 standards. An acceptable-use paragraph typically covers what students may not do. A policy also covers what staff may do, what data is collected, how parents are told, and how to opt out. If your state mandates district adoption (Maryland, Idaho, Oklahoma, Virginia, and now Florida), a paragraph will not satisfy the requirement.

Does a “non-mandatory” state policy like DC’s mean my school can ignore it?

Yes, legally. DC’s model is “guidance only, not a legal requirement.” In practice, most local agencies adopt the template with small edits because writing one from scratch is harder. The useful move is to compare your district’s version to the state model and ask why anything was deleted.

Can I opt my child out of AI tools if my state has no rule?

Often, but it depends on the district. Florida’s rule guarantees an opt-out with a comparable non-AI alternative. Elsewhere, ask in writing. Districts that have dealt with biometric data (like New Mexico’s Amira dispute) tend to have a process; districts that have not may improvise. A written request creates a record either way.

How do I know if a tool trains on my kid’s data?

The district contract or data-processing addendum says so, and the policy should reference it. Vendor claims vary: Google’s Utah agreement says conversations “are not used to train our AI models”; other tools are less explicit. If the policy and contract are silent, assume the vendor’s consumer terms apply and ask the district to confirm in writing.

What if my kid’s teacher uses AI to grade?

DC’s model puts grading in yellow: allowed with a human making the final decision. Florida’s rule requires teacher training on AI limits. Ask the teacher directly whether AI produced the score or the comments, and whether they reviewed it. A teacher who reviewed the output is following most 2026 guidance. A score with no human review is a policy question for the principal.

---

About the author

Ricky Flores is the founder of HiWave Makers and an electrical engineer with 15+ years of experience building consumer technology at Apple, Samsung, and Texas Instruments. He writes about how kids learn to build, think, and create in a tech-saturated world. Read more at hiwavemakers.com.

---

Sources

1. DC Office of the State Superintendent of Education. (2026, September 1). “OSSE Releases AI Model Policy to Guide Responsible Staff Use in Schools.” https://osse.dc.gov/release/osse-releases-ai-model-policy-guide-responsible-staff-use-schools
2. Illinois State Board of Education. (2026, July). Artificial Intelligence Guidance. https://www.isbe.net/Documents/AIGuidance.pdf
3. Capitol News Illinois. (2026, July 15). “Illinois State Board of Education issues AI guidance, written with help from AI.” https://capitolnewsillinois.com/news/illinois-state-board-of-education-issues-ai-guidance-written-with-help-from-ai/
4. Florida Department of Education. (2026, September 16). “Florida Sets National Standard for Responsible AI in Classrooms—From VPK Through College.” https://www.fldoe.org/newsroom/latest-news/florida-sets-national-standard-for-responsible-ai-in-classroomsfrom-vpk-through-college.stml
5. WTOP News. (2026, September). “No, robots won’t be teaching your kids. DC releases new AI guidelines for teachers.” https://wtop.com/dc/2026/09/no-robots-wont-be-teaching-your-kids-dc-releases-new-ai-guidelines-for-teachers/
6. GovTech. (2026, August 11). “New Mexico Allows Schools to Opt Out of Controversial AI Tool.” https://www.govtech.com/education/k-12/new-mexico-allows-schools-to-opt-out-of-controversial-ai-tool
7. Common Sense Media. (2026, August 18). Teens in the AI Era: Schoolwork and the Skills That Matter. https://www.commonsensemedia.org/research/teens-in-the-ai-era-schoolwork-and-skills-that-matter
8. Pursuit. (2026). “AI in Education News: Policies and Innovations” (running log incl. ED Dear Colleague Letter, Aug 20, 2026; Katy ISD, July 24, 2026). https://www.pursuit.us/news/ai-in-education-news-policies-innovations