An AI tools policy for schools that your filter, your board and your E-Rate file all agree on
Each of 11,853 AI tools has a K-12 decision: block, allow with controls, or allow. Every decision carries its reason, so the same answer reaches the web filter, the board packet and the letter home to parents.
Counts as of October 2026. The list grows as new AI tools appear.
What an E-Rate reviewer, the board and parents each see
An AI decision in a school is read by three groups with different questions. A good AI tools policy gives all three the same facts, in the form each one needs.
The E-Rate reviewer
Reviewers look for an internet safety policy and proof that a technology protection measure is running. AI image, video and companion tools can produce the harmful-to-minors content CIPA covers, so your filter has to know them.
- AI categories in the filter, with dates
- A written reason for every block
- Policy text that matches the filter settings
The school board
The board adopts the policy and answers to the community for it. It needs rules it can defend, change and review once a year.
- A policy ready to adopt, with annexes
- Category rules with a rationale for each
- A yearly report on reviews and changes
Parents and guardians
Parents ask which tools their child uses and what happens to what the child types. They want plain words, not a list of domains.
- A one-page letter with the three decisions
- Which vendors may not train on student data
- How consent works for children under 13
For the CIPA side of this, see our guides to the CIPA internet safety policy and what USAC asks for in a CIPA audit.
AI decisions now have to be explained, not only enforced
A full AI block removes reading support, captioning and translation that some students rely on. Opening everything lets in essay writers, deepfake tools and companion chatbots.
Student data is the first question
Parents ask whether a tool learns from what their child types. As of October 2026, 671 AI tools state that they train on user data, and 215 more do so by default unless the user opts out.
States are writing AI rules
Several states now expect a written AI policy from every district. Some add parental consent and a non-AI alternative for students.
A list is not a reason
A blocklist answers what is blocked. Reviewers, boards and families also ask why, and each decision here comes with its reason in writing.
What lawmakers are asking of schools
A short summary of AI rules for K-12 schools, as reported in October 2026. Check the current text with counsel before you rely on it.
Consent, alternatives and an off switch
State Board rules approved in September 2026 require parental consent before a student uses AI, and a non-AI alternative. AI that simulates an emotional relationship is banned, staff must be able to switch AI off at once, and district policies are due by 1 July 2027.
A policy in every school
Each district, community school and STEM school needed an AI policy by 1 July 2026. The state published a model policy that districts can start from.
Policies kept and reported
Local boards of education and public charter schools must keep an AI policy and report it.
Written policy before 2027-28
Districts need a written policy listing approved and prohibited uses, with data protection for students. Parents can opt their child out.
Companion chatbots and minors
California signed a companion chatbot law in September 2026. Colorado's law applies from 1 January 2027.
Congress and COPPA
Bills on AI companions for minors are before Congress. The amended COPPA rule applies from April 2026.
Five parts, one set of rules
The database says what each AI tool is and what it does with data. The rules turn that into a decision, and the kit turns the decisions into a policy the board can adopt.
AI tools database
11,853 AI tools in 18 categories and 165 subcategories, rebuilt daily.
K-12 policy rules
A decision and a reason for every tool, from 18 category rules and 165 subcategory rules.
K-12 AI Policy Kit
Board policy, parent letter, staff and student guidance and a review checklist.
13 risk factors
Three core risks plus ten factors checked against vendor terms.
AI apps for MDM
2,101 AI apps for iPad and Android, with their app IDs.
Three decisions, split across 11,853 tools
The counts below are from the build of 17 October 2026. Most AI tools are blocked for K-12, a large group is open under conditions, and only a few are open outright.
Block
Not reachable on school networks or devices, and not used for schoolwork on any device.
Allow with controls
Only with a school account, a signed data-privacy agreement and a teacher's approval for the class.
Allow
Low-risk tools that do not write open-ended text, such as mind-mapping, and safety tools such as content moderation.
The five steps behind every decision
Hard blocks
Abusive purpose, a high risk rating or high data-sovereignty exposure means block, with no exceptions.
Baseline rule
Each subcategory label brings its K-12 rule, or its all-sector rule; the strictest wins. With no subcategory rule, the category default applies.
Tightening
Training on user data by default moves a tool two steps stricter. Training unless users opt out, medium risk and medium sovereignty each move it one step.
AI-enabled relief
An established product that added AI features is placed under controls rather than blocked. Hard blocks never get this relief.
Education-tier relief
If training was the only reason for a block and the vendor's education tier does not train, the tool is allowed with controls on that tier.
Example decisions, as the verdict file shows them
Each row is a real record. The reason column is what you can read out to a parent or put in the audit file.
| Tool | Decision | Reason given | Trains on user data |
|---|---|---|---|
| ChatGPT | Block | General assistants & chatbots (K-12 rule) | Trains unless opted out |
| Claude | Block | General assistants & chatbots (K-12 rule) | Trains unless opted out |
| Microsoft Copilot | Allow with controls | General assistants & chatbots (K-12 rule); AI-enabled platform, control rather than block | Trains by default |
| Grammarly | Allow with controls | Grammar, style & proofreading; trains unless opted out; AI-enabled platform, control rather than block | Trains unless opted out |
| illumi | Allow | Mind-mapping & diagramming (K-12 rule) | Does not train |
| Sightengine | Allow | Content moderation | Does not train |
18 category defaults for K-12, with the reason for each
Every category starts from a K-12 default. Subcategory rules then open what schools need, such as captioning inside a blocked audio category.
| Category | K-12 default | Why, in short |
|---|---|---|
| Education & Learning | Controls | Tutors, quiz makers, course builders and grading aids are made for schools. A signed student data-privacy agreement is the condition. |
| Text & Language | Block | Open text generators can write the assignment for the student. Open chat also shows minors unfiltered output and asks for personal details at sign-up. |
| Code & Development | Controls | Computer science is a K-12 subject, and code assistants are now part of how code is written. |
| Image & Visual | Block | Open image generation is the main route to explicit or harassing images of real people, classmates included. |
| Video | Block | Text-to-video, avatar and dubbing tools create synthetic media of real people, with more convincing results than still images. |
| Audio, Voice & Music | Block | Voice cloning and voice changers make impersonation and bullying easy. Speech-to-text and text-to-speech are opened as assistive technology. |
| Lifestyle & Entertainment | Block | Companion chat, dating, astrology and games carry the highest risk for minors: emotional dependence, sexual content and data collection. |
| Search, Knowledge & Docs | Block | Answer engines hand over a finished answer instead of sources. Chat-with-PDF tools summarize the set reading. |
| Models & Infrastructure | Block | Raw model access and hosting skip the safety filters that consumer apps apply. A student with an API key has an unfiltered model. |
| Agents & Automation | Block | Agents act on accounts, browse and send data with no person checking each step. |
| Productivity & Collab | Controls | Documents, slides and task boards that schools already run for staff and older students. |
| Design, 3D & Creative | Controls | Design, CAD, 3D and game design are taught, so the teacher picks the tools a class uses. |
| Robotics & Embodied AI | Controls | Robot kits and computer vision appear in STEM lessons, with low risk and clear teaching value. |
| Security & Detection | Controls | Security and fraud tools are for staff. Detectors for AI content, deepfakes and harmful content are allowed outright. |
| Data, Analytics & Research | Block | Business analytics and scraping have no student use and often take in files that may hold student records. |
| Business & Vertical | Block | Finance, HR, legal and healthcare software for professionals, not pupils. |
| Marketing, Sales & SEO | Block | No classroom use. These tools gather contact data and send messages at scale. |
| Customer Support | Block | Helpdesk bots and call-center agents are business software. |
Subcategory rules: where learning and access are protected
Of the 165 subcategory rules, 112 block, 48 allow with controls and 5 allow. The rows below show how the rules keep accessibility and learning tools open inside stricter categories.
| Subcategory | Rule | Why |
|---|---|---|
| Subtitles & captions | Controls | Gives deaf and hard-of-hearing students access to video and speech. |
| Text-to-speech | Controls | Reads text aloud for students who struggle with reading. |
| Speech-to-text & transcription | Controls | Assistive technology for students with dyslexia or motor impairments. |
| Translation & localization | Controls | Helps students who learn in a second language, with controls that keep personal data out of prompts. |
| AI tutors & personalized learning | Controls | The core classroom use, with a data-privacy agreement and design suited to the age group. |
| Mind-mapping & diagramming | Allow | Organizes the student's own ideas, with little data or content risk. |
| Writing & long-form | Block | Writes complete essays, the central academic-integrity risk. |
| Paraphrasing | Block | Rewords copied text to get past plagiarism checks. |
| AI companions & character chat | Block | Emotional dependence and sexual role-play with minors; blocked in every sector. |
| Face swap & effects | Block | A direct deepfake risk; blocked in every sector. |
The policy the board adopts says what the filter does
Every kit document uses the same three decisions and the same five steps as the rules. When a reviewer compares the policy with the filter settings, they match.
AI Acceptable Use Policy
Ready for board adoption, in 14 sections with two annexes.
- Purpose, scope and definitions
- The three decisions
- How tools are classified and how the rules apply
- What "allow with controls" requires
- Student data and privacy, including FERPA, COPPA and CIPA
- Academic integrity and staff use
- Requesting a tool, exceptions, roles, violations and yearly review
- Annex A: K-12 category rules
- Annex B: Approved AI Tools List
Letter to parents
One page in plain language.
- The three decisions, with examples
- How student data is protected
- Consent for students under 13
- AI in schoolwork
- How families can help at home
Staff guidance
Six sections for teachers and staff.
- What you can use
- Using an approved tool
- What never goes into an AI tool
- Requesting a new tool
- Handling suspected AI misuse fairly
- What to do when something goes wrong
Student guidelines
Two versions by age.
- Grades K-5: 8 short rules and a line to sign
- Grades 6-12: 9 guidelines on approved tools, the AI level for each assignment, disclosure, privacy, respect, checking facts, wellbeing and reporting
Tool review checklist
One page for IT and curriculum staff.
- Describe the tool
- Check the hard blocks
- Find the baseline rule
- Tighten for data use and risk
- Apply relief and record the decision
Rules and filter files
The rules as data and the decisions as files your systems can import.
- Category rules, subcategory rules and decision layers
- Google Admin blocked URLs
- Chrome policy JSON
- DNS RPZ zone and hosts file
- Block, allow and controls domain lists
In the kit's own words
This policy sets out which generative AI tools may be used in the district, under what conditions, and how new tools are reviewed. It has three goals, in this order:
- Protect students, who are minors.
- Protect academic integrity.
- Keep vetted, useful tools available for teaching and learning.
CIPA (Children's Internet Protection Act) requires schools that receive E-rate funding to have an internet safety policy and to filter internet access for content that is harmful to minors.
An AI tool that cannot be placed in any category is blocked until it is reviewed. A tool with the verdict allow with controls may be used only after it has been added to the Approved AI Tools List.
No exceptions for hard blocks. Tools blocked under Step 1 (abusive purpose, high risk rating or high data-sovereignty exposure) cannot receive an exception.
Urgent action. The [Technology director] may block any tool at once if it puts students or data at risk, and then refers it for review.
- Ask your child to show you the AI tools they use for school.
- Remind them never to share personal details or photos with AI chatbots.
- Remind them that a chatbot is not a friend or a counselor.
Tip: remove identifiers first, for example "Student A" instead of a name. If you enter personal data by mistake, tell [Data privacy officer] the same day.
- I ask first. I use AI only when my teacher says it is OK.
- I use school tools. I use only the tools my school gives me, with my school login.
- I am kind. I never use AI to make pictures, videos or voices of real people, or to be mean to anyone.
Hard blocks: tick any that apply. Any tick means Block. Stop here.
- Abusive purpose: built to harm or deceive, such as fake or sexual images of real people, impersonation or fraud.
- High risk rating: readily makes sexual, violent or hateful content or fakes of real people, or connects students with strangers.
To fit the AI policy into your wider rules, read our guides to an acceptable use policy for schools and presenting web filtering to the school board.
What every AI tool does with student data
Three core risk factors sit on every tool. Ten more are checked against each vendor's published terms, with the clause quoted and a link to the terms.
1. Data sovereignty
Where data is stored and processed, which laws apply and who else can reach it.
2. Abusive purpose
296 tools are flagged as NSFW, deepfake, uncensored or nudify tools as of October 2026. All are hard blocks.
3. Trains on your data
Whether the consumer tier learns from what users type and upload.
4. Opt-out available
Can a user stop training on their data?
5. Opt-out scope
Does the opt-out cover all data, or only data from now on?
6. Enterprise no-training
Do paid or education tiers keep customer data out of training?
7. API no-training
Is API traffic kept out of training?
8. Feedback exception
Are ratings and thumbs up or down still used for training?
9. Retention
How long are prompts kept for monitoring?
10. De-identified use
May "anonymized" data still be used?
11. Human review
Can vendor staff read conversations?
12. Zero-retention option
Is there a mode that stores nothing?
13. Subprocessor training
May third parties train on the data?
Training on user data, across all 11,853 tools
Counts as of October 2026. Under the K-12 rules, a tool that does not say whether it trains cannot be simply allowed.
One record, as a district receives it
trains_consumer_default yes
optout_available yes
optout_scope future_only
feedback_exception yes
deidentified_use permitted
human_review yes
enterprise_no_training yes
api_no_training yes
k12_verdict block (General assistants & chatbots, K-12 rule)
Each field also carries the vendor's clause and a link to its terms. That is the evidence a privacy officer files with the review.
The same decisions on iPads and Android tablets
A web filter does not see what an installed app does. The app lists carry the same tools and categories into the device management your district already uses.
AI apps
Each linked to its AI tool and category.
Android apps
Listed by package name.
iOS and iPadOS apps
Listed by bundle ID.
New apps checked
App stores are checked every week for new AI apps.
Import files
Ready for the tools school IT teams already run.
Microsoft Intune: iOS restricted appsMicrosoft Intune: Android restricted appsJamf configuration profileAndroid Management API blocked appsBundle ID and package lists18 category filesMore on device controls in our page on app management on school devices.
The base: 11,853 AI tools, classified and kept current
The AI tools list is drawn from our database of 120 million classified domains. It is rebuilt every day, so new tools reach the decisions and the filter files quickly.
18 categories, 165 subcategories
From AI tutors and captioning to companion chat and face swap. Schools can decide by type of AI rather than all or nothing.
Rebuilt daily
About 300,000 newly registered domains are reviewed each day. New AI tools enter the list and get a K-12 decision.
One source for every control
Web filters, firewalls, DNS filters, Chromebooks and MDM read the same decisions.
Formats
CSVJSONhostsEDLPACRPZDNS filter imports
From board vote to filter settings in one term
Most districts already have a CIPA policy, a filter and an MDM. The AI policy slots into all three without a new system.
1. Adopt
Fill in the bracketed fields, check the text with counsel and take the policy to the board. Annex A records any rule the board changes.
2. Enforce
Import the block list into the web filter, the Google Admin and Chrome files into Chromebook settings, and the app files into MDM.
3. Inform
Send the parent letter, brief staff with the guidance and give students the guidelines for their grade band.
4. Review
Use the checklist for new tool requests and report to the board once a year. Keep the reasons with your E-Rate records.
AI tools policy for schools: common questions
Does CIPA require schools to block AI tools?
CIPA does not name AI tools, but it requires E-Rate schools and libraries to filter visual depictions that are obscene, child sexual abuse material or harmful to minors. AI image, video and companion tools can produce that content, so many districts cover them in the same policy and filter.
What does a district receive?
A block, allow-with-controls or allow decision for each of 11,853 AI tools as of October 2026, with its reason and 13 risk factors. It also receives the K-12 AI Policy Kit, ready-to-import filter files and AI app lists for MDM.
Are all AI tools blocked?
No. Tutoring, translation, captioning, reading support and coding tools are allowed with controls, and a few low-risk tools are allowed outright.
Can our board change a rule?
Yes. Each rule is published with its rationale, so the board can move it, record the date and reason in Annex A, and see which tools change.
How do vendor data terms affect a decision?
Training on user data by default moves a tool two steps stricter, and training unless users opt out moves it one step. Abusive purpose, high risk and high data-sovereignty exposure are hard blocks.
How often is it updated?
The tool list and the K-12 decisions are rebuilt daily, and app stores are checked weekly for new AI apps. Vendor terms are checked again when they change.
Related pages on AI in schools
Give your board an AI policy the filter already follows
Licenses for single districts, regional service agencies and platforms that serve many schools. Tell us how many students and devices you cover and which systems you run.
Product names in the examples, including ChatGPT, Claude, Microsoft Copilot, Grammarly, illumi, Sightengine, Microsoft Intune, Jamf, Google Admin, Chrome and Android, are trademarks of their owners. They appear only to identify the tools and systems discussed, and CIPA Web Filtering is not affiliated with, sponsored or endorsed by these companies.