Policy Resource

Sample AI Use Policy Language for Online Asynchronous High School Courses

A course-agnostic, tiered policy framework for high school programs where learners work independently, at their own pace, and without synchronous teacher oversight. Pick the tier that fits your course's learning outcomes, copy the student-facing language into your LMS, and use the instructor commentary to adapt the details to your discipline.

An AI policy is not a surveillance posture. It is an instructional design decision. The right default is restrictive in process-dominant courses, and it loosens deliberately where the learning outcomes require students to reason with AI rather than around it.

Grades 9–12 Online Asynchronous Any Discipline WCAG 2.2 AA UDL · SEL · OER-Friendly

How to Use This Resource

This page is written for instructors, curriculum designers, and program leaders. It includes a clearly marked student-facing excerpt you can lift directly into a syllabus, course overview page, or LMS module.

Work through the page in this order: understand the course-first design principle, pick a tier, read the recommended async default, then adopt the three shared protocols (Receipts, Evaluation, Conversation). Finish by pasting the student-facing block into your course and adjusting the bracketed course details.

"The right AI policy is the one that protects the learning your course is actually trying to produce. That is a design decision, not a compliance posture."

The Course-First Design Principle

Before you write an AI policy, answer one question about your course: is the graded outcome a process or a product?

Process outcomes ask learners to develop a skill by struggling through the work themselves. Reading comprehension. Math reasoning. Writing analysis. Scientific explanation. Language production. In these courses, AI assistance during the work erodes the exact cognitive load that produces the skill.
Product outcomes ask learners to complete a task that a competent adult professional now routinely completes with AI assistance. A marketing brief. A research summary. A project proposal. A design artifact. Code review. In these courses, prohibiting AI is both unrealistic and pedagogically unhelpful.

Most high school courses are process-dominant, which is why the recommended default for an async high school program is restrictive. But an Intro to AI elective, a CTE capstone, a professional-writing course, or a portfolio-driven senior project may legitimately live in a different tier. Make the call on instructional grounds, then communicate it clearly and consistently.

Working Rule

If the task is what the learner is being graded to grow into, protect it from AI. If the task is what the learner will use AI to perform after graduation, teach them to use it well.

The Three-Tier Policy Framework

Every course in your program adopts one of three AI postures. Tiers are not ranked by virtue. They are calibrated to learning outcomes. A single program can run all three tiers simultaneously as long as each course states its tier plainly and enforces it consistently.

Tier 1

AI-Restricted

Use when: the graded outcome is a cognitive process (analysis, reasoning, explanation, composition, problem solving, interpretation, translation). Recommended default for core ELA, mathematics, science, world language, and social studies.

Permitted

Self-discovery prompting: the learner asks AI to ask them questions that surface their own interests or stuck points.
Post-draft mechanics check: after a complete original draft exists, AI may be used to flag grammar, spelling, punctuation, or unit-conversion errors. No rewriting, restructuring, or expansion.
Accessibility accommodation use documented in an IEP, 504, or district-approved accommodation plan.

Not Permitted

Brainstorming, outlining, drafting, paraphrasing, or summarizing any portion of the learner's work.
Having AI explain a reading, solve a problem, or produce a worked example substituted for the learner's reasoning.
Running the learner's draft through AI to "improve" it beyond mechanics.

Required Documentation

Show Your Receipts for every permitted use, including mechanics checks.

Tier 2

AI-Supported

Use when: the course includes real-world task outcomes where AI fluency is a secondary skill, but the primary graded outcome still depends on the learner's thinking. Examples: PBL courses, career exploration, AP Seminar-style research, CS foundations.

Permitted With Documentation

Concept clarification: asking AI to explain a term, algorithm, or concept encountered in course materials.
Brainstorming partners for the learner's own ideas (learner proposes, AI expands, learner selects). Submitted ideas must be recognizably the learner's.
Revision feedback on a completed original draft (structural feedback, not rewriting).
Study support: practice problems, flashcards, quiz items on course content.
Post-draft mechanics check.

Not Permitted

Generating the learner's first draft, thesis, hypothesis, or interpretation.
Submitting AI-produced text without substantial learner authorship on top.
Any use that bypasses a learning outcome the assignment is designed to assess.

Required Documentation

Show Your Receipts for all AI use plus a learner-authored reflection on what AI produced versus what the learner kept.

Tier 3

AI-Integrated

Use when: the course is designed to teach AI-fluent work itself. Examples: AI literacy electives, generative media production, prompt engineering capstones, CTE courses where professional practice assumes AI collaboration, authentic client-facing senior projects.

Permitted With Documentation

AI as collaborator at any stage, including ideation, drafting, iteration, and revision.
Producing a final artifact that is the product of human-AI collaboration, provided the learner can articulate, defend, and revise their contributions.

Non-Negotiable Requirements

A process log or receipt for every AI-assisted step of the project.
A reflection component explaining what judgment calls the learner made, what AI suggestions they rejected, and why.
A live or recorded oral defense where the learner explains their work without AI assistance. This is the load-bearing assessment in a Tier 3 course.

Not Permitted

Submitting AI output without the required reflection or defense.
Misrepresenting AI-generated content as unaided human work.

Tier Comparison at a Glance

What each tier allows, prohibits, and requires
Dimension	Tier 1: Restricted	Tier 2: Supported	Tier 3: Integrated
Brainstorming / Ideation	Not permitted	Permitted with documentation	Permitted with documentation
Outlining / Structure	Not permitted	Not permitted	Permitted with documentation
First-Draft Generation	Not permitted	Not permitted	Permitted with documentation
Summarizing / Paraphrasing Readings	Not permitted	Not permitted	Permitted with documentation
Post-Draft Structural Feedback	Not permitted	Permitted with documentation	Permitted with documentation
Post-Draft Mechanics Check	Permitted with documentation	Permitted with documentation	Permitted with documentation
Self-Discovery Prompting	Permitted with documentation	Permitted with documentation	Permitted with documentation
Oral / Live Defense	Recommended for large summatives	Recommended for capstones	Required
Receipts Required	For all AI use	For all AI use	Process log + reflection

Recommended Default for Online Asynchronous High School

For an async high school program where most courses are assessed on process outcomes, Tier 1 (AI-Restricted) is the recommended default. This is not because AI is bad. It is because asynchronous learners complete work without teacher-led scaffolding, and unconstrained AI use in that setting tends to collapse the productive struggle that is doing the actual teaching.

Three calibrations matter for an async program specifically:

Discipline variance is real. A math fluency module, a world-language speaking assessment, and an ELA analytical essay all live in Tier 1 but have different boundary cases. Publish a short course-level appendix that names the three or four AI scenarios most likely to come up in that specific course, and state whether each is permitted.
Accommodations are not violations. Text-to-speech, speech-to-text, translation for access, grammar correction for dyslexia or dysgraphia, and any AI use specified in an IEP, 504, or district-approved accommodation plan are access, not academic dishonesty. Treat them that way in both policy and practice.
Equity of access is a program issue. Not every student has equal access to paid AI tools, stable internet, or parental support for navigating AI. A restrictive default partially neutralizes this; program leaders should still audit AI-adjacent assignment components for hidden resource assumptions.

Implementation Note

Publish the tier on the course landing page, in the syllabus, and in the first module's orientation content. Repeat it on every assignment page with a two-sentence per-assignment note. Learners in async courses have no synchronous moment to ask "wait, what's the rule?" The rule has to be findable.

Shared Protocol: Show Your Receipts

Regardless of tier, any permitted AI use is documented in a short receipt appended to the submitted work. This is a professional practice, not a punishment. It also makes conversation-based enforcement possible without surveillance software.

1. Tool Used: Name the AI tool and version if available (for example, ChatGPT, Gemini, Claude, Copilot, Grammarly AI, NotebookLM).
2. Prompt Given: Copy the exact prompt or request. If the interaction was a conversation, include the key turns.
3. AI Output: Copy or summarize what the tool returned. For long outputs, include the first meaningful passage and describe the rest.
4. What You Did With It: Explain specifically how you used the output. For mechanics checks, list which corrections you accepted and which you rejected and why. For Tier 2 or 3 uses, describe what you kept, what you changed, and what you threw out.

Tool Clarification

Built-in red/blue underline spell check in word processors is not an AI tool under this policy and does not require a receipt. Translation tools used for access, when documented in an accommodation plan, are governed by that plan rather than the AI policy.

Shared Protocol: AI Evaluation Activities

In every course and every tier, at least one unit should include a structured activity in which learners critically evaluate AI-generated content. The point is not to teach learners to use AI. The point is to teach them to see what AI gets wrong, what it misses, and where human judgment remains load-bearing.

Discipline-specific AI evaluation activities
Discipline	Activity Example	What the Learner Evaluates
English / Language Arts	Compare an AI-generated literary analysis to a learner-authored analysis of the same text.	Depth of textual evidence, specificity, tone, interpretive risk, handling of ambiguity.
Mathematics	Hand a learner an AI-generated worked solution containing a subtle conceptual error. Ask them to find and explain it.	Procedural fluency masking conceptual misunderstanding, invalid algebraic moves, misapplied theorems.
Science	Evaluate an AI-generated explanation of a scientific concept known to produce common misconceptions.	Common misconception patterns, oversimplification, plausible-sounding but unsupported mechanisms.
Social Studies / History	Evaluate an AI-generated historical summary for what it flattens, whose perspectives are absent, and what sourcing it gestures at without citing.	Dominant-narrative bias, missing voices, pseudo-sourcing, anachronism.
World Language	Evaluate an AI-generated translation or cultural explanation for register, idiomatic accuracy, and cultural nuance.	Formal register errors, idiom literalism, erasure of cultural context.
CTE / Career Pathways	Evaluate an AI-generated client deliverable (resume, business plan, marketing copy, code snippet) against an industry rubric.	Professional accuracy, ethical and legal issues, and the gap between generic AI output and client-specific work.
Research / Capstone	Evaluate an AI research summary for source quality, fabricated citations, and missing recent scholarship.	Hallucinated sources, citation style integrity, recency, source credibility.

Shared Protocol: Conversation-First Response to Suspected Violations

AI detection software is not reliable, it disproportionately flags neurodivergent writers and multilingual learners, and it should not be the primary mechanism for enforcing this policy. A better approach is authentic assessment design plus a conversation-first response when something seems off.

When an instructor suspects that submitted work was generated or substantially shaped by AI in a way the tier does not permit, the response is:

Open a conversation, not an accusation. Ask the learner about their process. A learner who authored the work can describe how their ideas developed, what they tried that did not work, and what choices they made.
Ask for an oral or written process explanation. In an async program this can be a short recorded response or a timed written reflection. The purpose is not gotcha. It is surfacing whether the thinking in the artifact lives in the learner.
Decide next steps together. If unauthorized AI use is confirmed, options include redoing the assignment with documented authentic process, submitting a revised version with receipts, or, for repeat patterns, engaging school academic integrity procedures. The primary goal is learning, not punishment.
Protect the learner's honest disclosure. A learner who self-discloses AI use that crossed the line should receive a substantially different response than one who denies it under questioning. Honesty is a learned skill. Rewarding it here teaches it.

Why this matters in an async program

Async learners are disproportionately students with accommodations, learners balancing work or caregiving, credit-recovery learners, and students for whom the traditional classroom did not work. A surveillance-heavy AI enforcement posture replicates the exact dynamics that pushed many of them out of brick-and-mortar schools. Design for trust with verification, not suspicion with theater.

Student-Facing Language You Can Copy

The blocks below are written directly to learners at roughly an 8th to 10th grade reading level. Paste the one that matches your course into your orientation module, syllabus, or policy page. Replace the bracketed items.

Student-Facing · Tier 1 (Restricted)

AI Use in This Course: [COURSE NAME] (Tier 1)

This course is designed to build [e.g., your analytical writing, your math reasoning, your ability to explain scientific ideas in your own words]. That means the thinking is the assignment. AI tools (ChatGPT, Gemini, Copilot, Claude, Grammarly AI, and similar services) cannot do that thinking for you without skipping the learning.

What you can do with AI in this course:

Self-discovery prompting. You can ask AI to ask you questions to help you figure out what you think. AI asks, you answer. You never ask AI to give you ideas.
Post-draft mechanics check. After you have written your own complete draft using your own ideas, you may use AI to flag grammar and spelling. You may not have it rewrite, expand, restructure, or "improve" your work.
Accommodation use that is part of your IEP, 504, or approved accommodation plan. That use is protected. It is access, not a policy violation.

What is not allowed: having AI brainstorm, outline, draft, paraphrase, summarize, explain a reading, or generate any part of what you submit.

Show Your Receipts: if you use AI at any point (including a grammar check), you add a short receipt at the end of your submission: the tool you used, the prompt you gave it, what it returned, and what you did with it.

If you are unsure whether a use is allowed, ask before you use it. Guessing wrong is a harder conversation than asking.

Student-Facing · Tier 2 (Supported)

AI Use in This Course: [COURSE NAME] (Tier 2)

This course lets you use AI for defined tasks where it genuinely helps your learning, but the thinking in your submitted work still has to be yours. You are graded on [e.g., the quality of your analysis, your project decisions, your evidence and reasoning], not on whether you used AI.

You can use AI to: explain a concept you met in course materials, brainstorm ideas after you have your own starting point, get structural feedback on a completed draft, generate practice problems or study questions, and check grammar and spelling after drafting.

You cannot use AI to: generate your first draft, thesis, hypothesis, or interpretation, or produce work you submit without substantial thinking of your own on top.

Show Your Receipts every time you use AI: the tool, the prompt, the output, and a short note on what you kept and what you changed.

Student-Facing · Tier 3 (Integrated)

AI Use in This Course: [COURSE NAME] (Tier 3)

This course is designed to teach you how to work with AI well. That means AI is a permitted collaborator at every stage of your project. It also means the bar for what counts as your work is higher, not lower.

You can: use AI to brainstorm, draft, iterate, and revise your project.

You must: keep a process log or receipts for every AI-assisted step, write a reflection explaining what you chose to keep, what you rejected, and why, and complete a short oral or recorded defense of your work without AI assistance. The defense is the centerpiece of how you are assessed.

You cannot: submit AI output without the required reflection or defense, or represent AI-generated content as work you did without assistance.

How AI Can Help You Write (Student-Facing)

Here are five honest ways AI can support your writing without doing it for you. Each use case names which tier allows it. If your course is Tier 1, only the uses marked Tier 1 are permitted. If your course is Tier 2 or Tier 3, more of these become available. When in doubt, check your course's policy statement.

Tier 2 or higher

Brainstorming & Outlining

AI can help you surface angles you hadn't considered, and organize scattered notes into a structure you refine. The ideas you submit still need to be yours; use AI to expand your thinking, not to replace it. Tier 1 courses do not permit this use.

Tier 2 or higher

Research Assistance

AI can help you locate relevant sources and summarize long articles so you can identify key points faster. Always verify that any cited source actually exists (AI sometimes fabricates citations) and always return to the original source to confirm what it really says. Tier 1 courses do not permit AI summarizing for you.

Tier 1 and higher

Grammar & Spelling (Post-Draft)

After you have written your own complete draft using your own ideas, AI can flag grammar, spelling, and punctuation errors so you can fix them. You decide which corrections to accept. AI should not rewrite sentences, restructure paragraphs, or "improve" your ideas. Permitted in every tier with Show Your Receipts documentation.

Tier 2 or higher

Feedback & Revision

AI can ask you questions about the structure, coherence, and flow of your completed draft and point out parts that are unclear. You decide what to revise. The revisions must be your own words. Tier 1 courses do not permit this use because it blurs into AI shaping the work.

Tier 2 or higher

Self-Assessment & Goal Setting

AI can help you evaluate your own progress, identify patterns in your writing, and set personal goals for your next assignment. This promotes metacognition without outsourcing the work. Tier 1 courses may permit a narrower version of this (AI asking you reflective questions about your own draft) as part of self-discovery prompting.

The Honest Version

AI can do any of these things well. The question is not whether AI can help you. The question is whether using AI in that way replaces the thinking your course is assessing. If the answer is yes, the tier says no, even if the use feels minor. If the answer is no, document the use and keep going.

Equity, Access, and UDL Notes

A good AI policy is not accidentally an accessibility or equity policy. It has to be designed that way.

Accessibility: AI-powered assistive technology documented in an IEP, 504, or district accommodation plan (screen readers, speech-to-text, text-to-speech, advanced grammar tools for dyslexia and dysgraphia, translation for language access) is not governed by this policy and must not be treated as a violation.
Cost and access: Do not design assignments that require a paid AI tier, and do not let optional AI use become a hidden prerequisite for success. If you permit AI in Tier 2 or Tier 3, make sure a free-tier path works.
Multilingual learners: AI translation for understanding course materials is access, not policy violation. AI translation used to generate the final graded artifact in a language course is. Name that line explicitly.
Neurodivergent learners: AI detection tools disproportionately flag autistic writers, learners with ADHD, and multilingual learners. Do not use them as primary evidence in an integrity conversation. Rely on conversation and process.
UDL provision: Offer learners a choice of artifact format when possible (written, recorded, visual). A learner who can demonstrate their thinking in an oral defense has done the cognitive work regardless of format.

"Equity is the scope of this policy, not a feature of it. If the rule punishes the wrong learners for the wrong reasons, rewrite the rule."

Need Help Adopting This in Your Program?

I design and audit AI policies for online and hybrid schools. If you'd like help calibrating tiers to your courses, writing discipline-specific appendices, or facilitating faculty training around this framework, let's talk.

Book a Consultation AI Policy Services