AI Policies in 32 of the Largest US School Districts (2026 Tracker)

School AI rules are being written right now, district by district, and they disagree with each other. This tracker records what 32 of the largest US school districts say about student use of generative AI, using each district’s own documents where we could open them and local news or education press where we could not. Every row links to its source.

What did we find?

Fewer than half of these large districts have AI rules in force, and those that do mostly allow AI with conditions. These findings are computed from the 32 primary rows of the dataset (one per district).

1. 14 of 32 have a published AI policy or guidance in force: 6 board policies and 8 guidance documents. The Prince George’s County policy exists on the district’s site but returned an error to us, so its text is unverified.
2. Student use is mostly conditional: partial 16, yes 1 (Charlotte-Mecklenburg), no 2 (LAUSD for 2026-27; Northside on its network), unknown 13.
3. 7 of the 14 in-force districts require students to disclose AI use: Chicago, Hillsborough, Montgomery MD, Charlotte-Mecklenburg, Philadelphia, Hawaii and Loudoun. Orange County requires it in part.
4. 8 of 32 blocked or restricted AI at some point: NYC, LAUSD, Miami-Dade, Clark County, Wake, Northside, Cobb and Austin. NYC (2023) and Miami-Dade reversed course, and Miami-Dade now uses Gemini for about 100,000 high-schoolers. NYC and LAUSD tightened again for 2026-27.
5. Named tools: Gemini is named in 4 districts and ChatGPT for Teachers in 3, staff-only where the district says so.

What do the policies actually say?

The details vary more than the headline status suggests. A few representative rules, quoted from the sources in the table:

• Chicago: “Students must obtain permission from their teacher before using GenAI tools for academic work.”
• Hawaii: “Any use of AI to aid assignments, projects, or research must be declared.”
• Hillsborough: “Students cannot access generative AI before eighth grade without supervision.”
• Orange County FL: graded work only “unless you have written permission from the teacher,” and teachers may not use AI detectors.
• Charlotte-Mecklenburg: passed unanimously; no AI use “as a substitute for effort or problem-solving skills.”
• Houston: under 14, pre-approved tools with supervision; 14 and older, parent consent plus teacher permission.
• NYC (2026-27): “Software that uses student-facing generative AI is not permitted in grades 2K-8”; grades 9-12 are limited to vetted programs.
• LAUSD (2026-27): “restricted all student access to generative artificial intelligence platforms,” after previously allowing students 13+ to use approved tools after digital citizenship training.

Age and grade limits are the most common condition. Disclosure is the second. If you are a student, the practical takeaway is to ask your teacher before using AI for graded work and to say how you used it. Our guide on how to use AI for homework without cheating covers how, and how to cite ChatGPT shows the citation formats.

Full tracker: all 32 districts

One row per district, showing the current status as of October 14, 2026. “Partial” student use means use is allowed with conditions. “Unknown” means the source we read did not say. Sorted alphabetically.

Notable history: blocks, reversals and a chatbot collapse

The current status often hides a back-and-forth. These are the additional rows in the dataset, kept because they explain how districts got where they are.

Three stories stand out. NYC blocked ChatGPT on school devices and networks in January 2023 and reversed in May 2023, when Chancellor David Banks wrote of “knee-jerk fear”; by 2026-27 it had a moratorium on student-facing generative AI for younger grades. LAUSD launched an AI chatbot called “Ed” in March 2024 under a contract of about $6 million over five years, about half paid, and it went offline in June 2024 when the vendor, AllHere, furloughed staff. Miami-Dade moved from a chatbot block to Gemini for Education for about 100,000 high-school students in 2025, then directed a districtwide tiered framework.

How we built the tracker

Method:

• Districts: 32 of the largest US public school districts.
• Snapshot: October 14, 2026.
• Sources: district documents (board policies, guidance pages, guidebooks) where we could open them. Many district sites blocked automated access, so a large share of rows cite local news or education trade press such as Chalkbeat, K-12 Dive, The 74, EdSurge, Axios, WFAE, Spectrum News, Loudoun Now, Community Impact and GovTech. One row (Miami-Dade’s Gemini rollout) cites Google’s blog, a vendor source.
• Rows: 43 rows covering 32 districts. In each district, the first row is the current status; further rows are notable history.
• Coding: student_use_allowed is Y, N, partial (allowed with conditions) or unknown. disclosure_required is Y, partial or unknown.
• Every row cites the URL we actually opened. The one exception is Memphis-Shelby County, where we found nothing to open.

Coverage caveats

Read the gaps before you cite the numbers.

• “No public policy found” is not “no policy.” It means we could not find and open one. Several of these districts are in states with new requirements: Florida’s Board of Education approved AI rules in September 2026 that require parental consent and district policies by July 1, 2027; Maryland districts faced a fall-2026 deadline; Tennessee requires districts to submit AI policies to the state.
• Weak or provisional rows: Prince George’s County (policy text returned an error; known only from snippets), Broward (hub labeled “working draft”), Philadelphia (from news coverage because the district site refused connection), Hillsborough (policy number and date from Axios), Houston (official guidebook link returned 404; row cites a Houston Landing explainer), Montgomery MD (the PDF we read may be pre-final text; adoption is confirmed), Jefferson County KY (secondary rewrite of a WDRB story).
• Not confirmed as final: Cy-Fair, San Diego, Palm Beach and Fairfax. Austin’s 2023 restriction rests on a single Brookings mention.
• Conflicting sources: Wake County has a 2024 report allowing ChatGPT for students 13+ with citation and a 2025 report saying ChatGPT is restricted and Gemini preferred. We kept both.
• Student rules only. We coded what students may do. Several districts have staff-only tools (for example ChatGPT for Teachers) that do not change student access.

Download and reuse the data

The full dataset is free to use. Download ai_policies.csv (UTF-8, 43 rows, 10 columns: district, state, status, doc_title, doc_url, date, student_use_allowed, disclosure_required, approved_tools_named, notes).

License: Creative Commons Attribution 4.0 International (CC BY 4.0)↗. You can copy, adapt and republish it, including commercially, as long as you credit “The Homework Site, School AI Policy Tracker (snapshot 2026-10-14)” and link to this page.

Teachers writing their own rules can start from our classroom AI policy template and the guide to writing a classroom AI policy. For the evidence on AI detection, which one district (Orange County FL) now bars teachers from using, see do AI detectors flag honest students?. Parents can find a plain explanation in is using AI cheating?, and our AI & School hub collects every related guide.