Insights

AI Transparency in Schools: Tool Register

Build AI transparency in schools with a public tool register, family notices, clear ownership, feedback routes, and a practical review cycle.

Published August 16, 2026 By SchoolAmplified Editorial Team 15 min read

Superintendents
Technology and data privacy leaders
Communications and family engagement leaders
Curriculum leaders
School board members

A diverse group of students and educators discussing ideas together in a classroom

15 min read

Make every approved AI use easy to find and explain

Catalogue the use, name its limits, show human oversight, answer questions, and keep the record current.

AI transparency in schools should let a family answer a practical question without searching through board packets, vendor pages, and old emails: What AI is the district using, for what purpose, with whose information, under whose judgment, and whom can I ask about it?

A policy statement alone cannot answer that question. Neither can a list of product names. Districts need two connected records: a detailed internal AI system inventory for governance and a plain-language public AI tool register for students, families, staff, and the community.

In brief: catalogue every approved AI-enabled use, including features embedded in products the district already owns. For each use, name the purpose, users, information involved, human-review boundary, limits, responsible owner, approval status, and next review date. Publish the portion families need in accessible language. Add targeted notice when a use directly involves a child, their work, or their information. Give people a real route to ask questions, report an error, request an accommodation, or understand an available alternative. Then update the record when the product or district workflow changes.

This is a governance and communication framework, not legal advice. Public notice is not automatically consent, and a public register does not replace privacy notices, records obligations, accessibility duties, collective bargaining, procurement review, or rights that vary by jurisdiction and use. District counsel and qualified privacy leaders should determine what must be disclosed, when permission or an alternative is required, and what information should not be published.

Why AI transparency is a district priority now

Families are asking for concrete disclosure, not general reassurance. A May–June 2026 National Parents Union survey of 1,527 parents of public K–12 students, with a reported margin of error of plus or minus 2.7 percentage points, found that:

88% said schools should inform parents if their child will use or interact with AI as part of school.
87% said schools should inform parents if their child's information or assignments will be shared with AI.
84% said schools should disclose whether and how staff use AI for work such as lesson plans, materials, grading, schedules, or policies.

Those results do not tell a district which tool to adopt or whether a particular use is lawful. They do show that families distinguish among student interaction, information sharing, and staff use—and expect visibility into each.

Public policy is moving in the same direction. North Carolina's enacted Session Law 2026-41 directs the state Department of Public Instruction to maintain both a public list of AI tools reviewed under its evaluation framework and a public list of all AI tools used in public school units. The law also calls for transparency with students and families in educator training. Those provisions apply within North Carolina on the law's schedule; districts elsewhere should not treat them as a universal legal requirement. They are, however, a consequential example of AI inventory and public disclosure becoming an operating expectation rather than a communications extra.

Current guidance gives districts a practical foundation:

The NIST AI Risk Management Framework Core calls for mechanisms to inventory AI systems, document risks and impacts, communicate about them, and engage affected people.
Michigan's 2026 AI Starter Guide for Districts includes disclosure norms, prohibited uses, human oversight, and periodic review among its starter practices.
Oregon's 2026 Generative AI in K–12 Classrooms guidance recommends transparent communication with the school community about whether, how, and why AI may be used.
New York City Public Schools' 2026 AI guidance requires tools to meet transparency and explainability expectations and gives families a contact for questions about tools used in their child's school.

The district lesson is straightforward: if the organization cannot produce a current inventory internally, it cannot communicate accurately externally. If it publishes a list without ownership and review, that list will become another stale page families learn not to trust.

Define the use before promising transparency

“We use artificial intelligence” is too vague to govern. “We use Product X” is not much better. One product may contain several AI features with different users, inputs, outputs, and consequences.

Define the unit of transparency as an approved use, not merely a vendor.

For example, these should be separate records even if the same platform powers them:

a teacher drafting a family newsletter from approved district information
a student receiving hints in an algebra practice tool
a principal summarizing anonymous climate-survey comments
a system flagging language in a student account for a safety review
a human resources team generating a first draft of a job description

The purpose, affected people, data, human review, evidence standard, and available alternatives differ. A single vendor row would hide those differences.

Districts should also define what falls inside the inventory. Include generative tools, predictive systems, recommendation or ranking features, automated classification, AI-enabled monitoring, and material AI features embedded in learning, productivity, safety, communications, or administrative products. Do not wait for a vendor to market the feature as “AI.” Ask what the system does and whether an automated model influences content, recommendations, flags, access, or decisions.

This work complements the district's AI governance framework and vendor review process. Governance establishes authority and risk controls. Procurement evaluates a proposed tool. The transparency register shows what the district actually approved and how the live use is bounded.

Build two connected layers

A responsible transparency program needs an internal record and a public record. They should share a source, but they should not contain identical detail.

Layer 1: the internal AI system inventory

The internal inventory supports procurement, privacy, security, instruction, accessibility, legal review, incident response, renewal, and decommissioning. NIST's AI RMF Playbook describes an AI system inventory as an organized database of artifacts that can include system documentation, data dictionaries, incident plans, implementation links, and responsible contacts.

For each approved use, retain at least:

System and feature: product, vendor, enabled AI feature, version when available, and integrations.
Purpose and boundary: problem addressed, permitted task, prohibited task, and non-AI alternative considered.
Users and affected people: roles, grades, schools, programs, and people affected even if they do not operate the tool.
Information flow: input categories, output, storage, sharing, retention, deletion, and connected systems.
Human authority: who reviews the output, evidence they must consult, who decides, and who can override or stop the use.
Evidence and limits: evaluation performed, results, limitations, accessibility conditions, and unresolved risks.
Governance record: procurement, privacy, security, curriculum, accessibility, legal, and labor reviews as applicable.
Lifecycle: approval date, owner, next review, change triggers, incidents, renewal, and retirement plan.

Do not put student records, security configurations, confidential contract terms, sensitive incident details, authentication information, or other protected content in the public register. Transparency requires useful explanation, not uncontrolled disclosure.

Layer 2: the public AI tool register

The public register translates approved uses into information a student, family member, educator, board member, or community member can understand.

Each entry should answer:

What is it? Name the product and the specific AI-enabled feature or use.
Why is it used? State the educational or operational purpose in one sentence.
Who uses or experiences it? Identify relevant roles, grades, schools, or programs.
What information is involved? Describe categories in plain language, not individual records.
What does the AI produce? Explain the draft, hint, recommendation, summary, flag, or other output.
What does a person decide? Name the human-review step and consequential actions the AI is not authorized to take.
What are the limits? State prohibited uses, known limitations, and important conditions.
What options exist? Explain an accommodation, correction route, or alternative when required or offered.
Who owns the use? Provide a district role or monitored contact, not a vendor support form.
When was it reviewed? Show approval, last-review, and next-review dates plus the status: pilot, approved, restricted, paused, or retired.

A public register can link to the district's acceptable-use policy, privacy notices, accessibility information, vendor terms, evaluation summary, board materials, and family resources. It should not force a reader to open six documents to learn the basic facts.

Use the CLEAR transparency cycle

The CLEAR cycle turns AI transparency from a one-time webpage into a maintained district practice: Catalogue, Label, Explain, Answer, Refresh.

C — Catalogue the complete use

Start with purchasing records, data-privacy agreements, single sign-on and rostering systems, app approvals, curriculum licenses, browser extensions, safety systems, productivity suites, and department interviews. Ask vendors and internal product owners which AI features are enabled now, planned, optional, or introduced through an update.

Reconcile the list with real practice. A tool may be approved but unused. Another may be used through a free account without district review. An existing platform may have activated a new AI feature after the original contract.

District Perspective

The work gets easier when teams operate from shared information

Communication, continuity, and implementation improve when the model is more coordinated.

AI transparency starts with a complete internal inventory and a plain-language public register
Families need timely, use-specific notice in addition to a general policy page

SuperintendentsTechnology and data privacy leadersCommunications and family engagement leaders

The work gets easier when teams operate from shared information

Do not publish an unverified draft as a complete inventory. Mark entries honestly: confirmed and approved, under review, restricted, or being retired. Give staff a safe route to disclose a previously unrecorded use so the district can evaluate it rather than drive it further underground.

L — Label the purpose, people, information, and limits

Write every entry around the exact workflow. Avoid marketing language such as “personalizes learning” or “improves efficiency” unless the district has defined and evaluated what that means locally.

Use a bounded sentence:

For these users, the tool uses these information categories to produce this output for this purpose. This district role reviews it before this action, and the tool may not cross this boundary.

If the team cannot complete that sentence, the use is not ready for public explanation—or perhaps for approval.

E — Explain through the right level of notice

One public register is necessary but not always sufficient. Use three levels of communication:

#### Standing disclosure

Maintain the searchable public register, district AI principles, current policies, review process, and a general contact. Make the pages accessible, mobile-friendly, translated according to district language-access practices, and available to people who do not have a portal login.

#### Targeted advance notice

Notify affected families, students, or staff before a material new use when appropriate. The notice should explain the purpose, timing, tool, information boundary, human oversight, known limitations, contact, and any legally required or district-offered choice.

Do not hide a significant use inside a long annual technology form. A safety-monitoring system, student-facing chatbot, AI-supported assessment workflow, or feature that receives student work deserves notice proportionate to its effect.

#### In-context disclosure

Tell people when AI is part of the immediate interaction. A teacher can identify the approved AI role on an assignment. A district document can state when AI supported drafting and that a person reviewed the final version. A staff workflow can show when a summary or recommendation was machine-generated.

The district's AI acceptable use policy supplies common rules. Use-specific notice makes those rules understandable where the AI actually appears.

A — Answer questions and correct the record

Transparency is incomplete if the only action available is “read more.” Every entry needs a monitored route for questions and concerns.

Create a triage path that can handle:

a factual correction to the public entry
a question about information collection or sharing
an accessibility or language-access barrier
a suspected output error or harmful result
a request to understand human review or an available alternative
a report that a listed use differs from what happens in practice
an unapproved or newly enabled AI feature

Publish expected response times and route consequential concerns to a person with authority to investigate, correct, restrict, or pause the use. Do not send families back to the vendor when the district chose and governs the school workflow.

Keep a decision log internally. Record the question, owner, resolution, change made, and whether the public entry or district guidance needs revision. Recurring questions are evidence that the explanation or implementation is unclear.

R — Refresh after changes and report what the district learned

Review on a schedule and after meaningful change. Triggers should include:

a new model, feature, integration, data flow, or vendor term
expansion to a new grade, school, role, subject, or decision
a material incident, complaint, accessibility issue, or evaluation finding
a change to the human-review workflow or prohibited-use boundary
contract renewal, suspension, or retirement

Show a visible “last reviewed” date. Preserve prior internal decisions so staff can understand why a use changed. For a major pilot, publish a short outcome summary: what the district intended, what it measured, what it learned, limitations, and whether the use was expanded, revised, paused, or stopped.

Transparency should include uncertainty. “We have not yet established whether this improves learning” can be more trustworthy than a confident but unsupported benefit claim.

Match disclosure depth to consequence

Not every AI use needs the same public explanation. Use a consequence ladder.

Low consequence: internal drafting support

Example: a staff member uses an approved tool to draft agenda headings from non-sensitive information, then rewrites and approves the document. The register can be concise, but it should still name the tool, permitted data, human responsibility, and owner.

Moderate consequence: learning or communication support

Example: a student receives practice hints, or staff use AI-assisted translation for a family message before qualified human review. Explain affected grades or audiences, information categories, instructional or communication purpose, review, accessibility, and correction process. Provide targeted notice when the use is new or material.

High consequence: flags, rankings, recommendations, or decisions

Example: a system contributes to grading, discipline, special education, employment, safety escalation, service eligibility, or another material opportunity. Public explanation should be more detailed and the internal evidence much stronger. Name what the system does not decide, the evidence a human reviews, recourse, alternatives where applicable, monitoring, and stop conditions.

Some uses may remain inappropriate even with extensive disclosure. Transparency does not cure weak evidence, unlawful data use, inaccessible design, bias, security problems, or the absence of meaningful human authority.

Avoid five common transparency failures

1. Publishing vendor names without uses

A product list tells families what the district bought, not what happens to a child, an assignment, or a decision. Publish at the use level.

2. Treating notice as permission

District Perspective

District leadership needs clearer signals and stronger communication rhythm

Systems feel more credible when guidance and public experience stay connected.

Families need timely, use-specific notice in addition to a general policy page
Every published entry needs an owner, review date, feedback route, and change process

District leadership needs clearer signals and stronger communication rhythm

Notice, consent, opt-out, accommodation, and alternative access are different concepts. State precisely what the district is offering or required to provide. Do not imply a choice that does not exist, and do not hide a required choice behind a general notice.

3. Over-disclosing sensitive details

Do not expose protected records, vulnerabilities, investigation details, or confidential information in the name of transparency. Publish what people need to understand the use and exercise available rights; protect what would create a different harm.

4. Describing the product instead of the district's accountability

“The vendor uses responsible AI” does not identify who approved the use, reviews output, handles errors, or can stop it. The district entry should name district responsibility.

5. Launching a page with no maintenance owner

An outdated register can mislead families and staff. Every record needs a named role, review date, change trigger, and archive or retirement status.

A 30-day launch for districts

Week 1: set scope and ownership

Name an executive sponsor and working owners from technology, privacy, curriculum, communications, accessibility, student services, legal, procurement, and school leadership. Define AI for inventory purposes, approval statuses, public fields, protected internal fields, and escalation routes.

Week 2: build and reconcile the inventory

Gather records across departments and systems. Break vendors into approved uses. Confirm enabled features, users, information flows, human review, owners, and dates. Flag gaps and unapproved uses for review rather than presenting the first list as final.

Week 3: draft and test public entries

Write plain-language entries for a representative low-, moderate-, and high-consequence use. Ask students, families, educators, multilingual staff, special educators, and communications staff to answer the ten public-register questions using only the draft page. Revise where they cannot.

Week 4: publish, notify, and establish review

Publish the register with a clear status and update date. Send targeted notices for material live uses that families may not reasonably know about. Open the feedback route, assign service levels, schedule reviews, and brief school leaders so their answers match the public record.

Do not describe the first release as complete if reconciliation is still underway. Publish the scope and the date of the next update.

Measure whether transparency is working

Page views do not show whether a district is understandable or accountable. Track a small set of operating measures:

percentage of confirmed AI uses with a complete internal record
percentage of public entries reviewed by the due date
time between a material product change and an updated record
percentage of affected audiences receiving required or planned targeted notice
whether test readers can correctly identify purpose, information, human authority, limits, and contact
time to acknowledge and resolve questions or corrections
recurring questions by topic, language, school, and use
unapproved or undocumented uses discovered and brought into review
accommodations, alternatives, corrections, restrictions, or pauses resulting from feedback
differences between the public description and observed practice

Report enough of these findings to show that the register is a live governance mechanism. Avoid publishing small-group data or incident details that could identify a student or staff member.

Where SchoolAmplified fits

AI transparency fails when approved information is scattered across spreadsheets, contracts, policy documents, board materials, emails, and individual memory. A communications team cannot keep families accurately informed if product owners and reviewers are working from different versions of the facts.

District Assist can help authorized staff retrieve current, district-controlled guidance and approved answers from a governed knowledge layer. SchoolAmplified can support clearer family communication, consistent school-level explanations, visible ownership, and human review across recurring workflows. It does not determine whether a tool is lawful, approve a vendor, provide consent on anyone's behalf, audit an algorithm, or replace district counsel, privacy leaders, educators, or decision-makers.

SchoolAmplified's single-source-of-truth approach, trust model, and implementation process support a specific district outcome: when a family or staff member asks how an approved AI use works, the answer should come from current district knowledge with an accountable human behind it—not from vendor marketing, an outdated attachment, or a different explanation at every school.

AI transparency checklist for school districts

Before calling an AI use transparent, confirm that the district has:

defined which AI-enabled systems and embedded features belong in the inventory
catalogued approved uses rather than only vendors
reconciled contracts, app approvals, integrations, and actual staff practice
documented purpose, users, affected people, information flow, output, and consequence
named the person who reviews output and the person who can restrict or stop the use
separated protected internal detail from useful public explanation
published plain-language entries with status and review dates
provided standing, targeted, and in-context notice where each is appropriate
distinguished notice, consent, opt-out, accommodation, and alternative access
described known limitations and unresolved evidence honestly
made public information accessible, mobile-friendly, and available through district language-access practices
created a monitored route for questions, corrections, accessibility needs, and concerns
established service levels and an escalation path with real authority
defined update triggers for product, data, workflow, audience, and policy changes
scheduled periodic reconciliation and review
explained material pilot results, changes, pauses, and retirements
trained principals and staff to use the same current district record
protected student records, confidential information, and security-sensitive detail

AI transparency is not a promise to publish everything. It is the discipline of making the district's actual use understandable, bounded, current, and answerable to the people it affects. A strong public register gives families more than a product name: it shows the purpose, the limits, the human responsibility, and the path to be heard.

Sources and further reading

Published by SchoolAmplified Editorial Team. Read our editorial standards.