Building Your Course AI Policy

A practical, evidence-informed guide to writing an AI statement for your syllabus that is grounded in research, peer institutions, and the field's leading professional organizations. Optional. Adaptable. Yours to shape.

*A special thanks to Zach Claybaugh, Dom Hamerly, and Jen Stockdale for assisting in generating this resource.

Silence is a policy

Without an explicit course statement, students fill the gap with assumptions; often wrong ones. A clear policy protects transparency, fairness, and your own time.

The gap institutions are racing to close

Generative AI reached classrooms faster than most governance structures could respond. National surveys of higher-ed leadership found that student use is now near-ubiquitous, even as formal faculty guidance remains uneven across departments and courses.[1]

That unevenness has a cost. Faculty who study institutional AI policy note that policies which look comprehensive on paper often fail to change student behavior in practice because the missing ingredient isn't more rule-text, it's communication, repetition, and dialogue around the rule.[2]

Your statement is a teaching tool, not just a rule

The strongest course AI statements borrow the same design principles as the rest of an effective syllabus: student-centered language, precision, transparency, and worked examples. Avoid legal boilerplate.[3]

And a syllabus statement is more effective when it's paired with a stated rationale. Faculty-perspective research emphasizes that policies work best when they're treated as living, discussable documents with room for context, not just static prohibition.[4]

Seven steps to a working draft

1. Reflect before you draft

Don't start from someone else's template. Ask: what counts as "AI" in my course? When does the policy apply? What contexts allow or preclude its use? ~ — Stanford Teaching Commons

CTLE suggestion: Create a table that lists your course objectives as they align with your activities and assignments and for each row ask where and how, if at all, AI can be beneficial or harmful.

2. Place your course on the spectrum

Most policies run from full prohibition, to conditional/limited use, to encouraged or open use. Decide roughly where your course sits before you write a word of syllabus language. ~ — Mercer University CTL

CTLE suggestion: Use the CTLE suggested language below one of the tools provided below.

3. Draft in plain, student-facing language

Name the specific tools or categories your policy covers. Avoid legal register. Consider whether the policy should vary by assignment type.

4. State the rationale, not just the rule

Explain why this policy fits this course's learning goals. Policies paired with a stated rationale are more legible to students and easier to enforce fairly. ~ — Frontiers in Education, faculty case study, 2025

5. Design a fair process for suspected violations

Outline how you'll respond to suspected unauthorized use before it happens — gathering evidence and giving students a chance to explain their process, not assuming intent. ~ Mesa Community College CTL

CTLE suggestion: Avoid wording this punitatively and more as an invitation for dialogue. However, set your boundaries and refer to DU Integrity expectations as a universal requirement. Also emphasize that when in doubt, they should be proactive and ask you.

6. Make it visible and repeat it

Post the policy in the syllabus, on your Canvas assignment pages, and revisit it verbally, especially in week one. One-time mentions don't stick. ~ Kent State University CTL

7. Pilot, then revise each term

Treat the policy as a living document. Tools, institutional guidance, and your own comfort level will all keep shifting so build in a checkpoint to revisit it.

CTLE suggestion: Consider co-creating norms with your class around the policy rather than just stating it as a definitive top-down policy.

What actually goes in the statement

The process above tells you how to arrive at a policy. This is what to put on the page once you get there grounded in the specific components research and practitioner guidance converge on, regardless of how permissive your stance is.

The minimum viable policy: three required elements

A 2026 course-design study reviewing physiology syllabi proposed that any classroom AI statement needs at least three components to function: a clear statement of acceptable use, a requirement for citation or disclosure when AI is used, and a plain statement of the consequences for not following the policy. Fewer than that, and the statement tends to leave students guessing at exactly the moments it matters most.[5]

The fuller version: four to seven components

Teaching-center guidance converges on a slightly longer list once you move past the minimum. Pulling from multiple course-level guides, a complete statement typically includes:

Your rationale — why this policy, for this course, not a generic integrity statement.
What counts as "AI" — name tools/categories explicitly (ChatGPT, Claude, Gemini, Copilot, Grammarly's generative features, etc.) rather than leaving it undefined.
Permitted uses, stated concretely — brainstorming vs. drafting vs. editing vs. full generation, ideally tied to specific assignment types.
Documentation method — how students should disclose use: a citation format, a short acknowledgment line, a disclosure form, or prompt/output screenshots.
Consequences for non-adherence — stated plainly, before it comes up.
An invitation to ask — explicit encouragement to check with you when a use case is unclear, rather than guessing.
A revision clause — a line noting the policy may be updated mid-semester as tools and your own comfort level evolve.

Synthesized from Ohio State University Teaching and Learning Resource Center; Carleton University AI @ Carleton guidance; University of Nevada, Reno teaching resources; University of Idaho AI Resource Guide

You don't have to design this from scratch

Several frameworks exist specifically to help instructors define degrees of permitted AI use, rather than a single yes/no rule.

AI Assessment Scale (AIAS)

Guided policy-builder tools

Crowd-sourced example banks

A word on detection tools

A five-level (now six, with the 2025 "AI exploration" addition) framework running from No AI, through AI-assisted idea generation and editing, to full AI collaboration — matched to your learning outcomes rather than applied uniformly. Adopted by institutions across five continents and translated into 30+ languages.

Perkins, Roe & Furze, Journal of University Teaching & Learning Practice, 2025 — CC BY-NC-SA 4.0, attribution required on reuse

Rather than starting from a blank page, some institutions offer a step-through builder that walks you through each policy element with editable example language — useful as a model even if you build your own.

e.g.

Seaver College's Generative AI Syllabus Statement Tool, referenced by Kent State CTL

NKU AI Acceptable Use Statement Generator: NKU Tool

SJSU AI Policy Builder: SJSU Tool

Several CTLs maintain living, cross-institutional collections of syllabus language organized by permissiveness level which is a faster starting point than writing cold.

e.g. The Sentient Syllabus Project, referenced by University of Pittsburgh CTL; Duke CTL's crowd-sourced language collection

AI-detection software remains unreliable and contested. Most best practices steer faculty toward disclosure-based and process-based approaches like drafts, reflections, oral checks over detector-based enforcement.

Consistent with enforcement-risk guidance from Mercer University CTL

Dominican CTLE Suggested Syllabus Language

Sample AI Syllabus Language

Templates for Faculty Use

A Note to Instructors

The two policy examples below are provided as starting points, not requirements. They are intended to help instructors quickly draft syllabus language regarding the use of artificial intelligence (AI) tools in their courses.

Each instructor should adapt the wording, scope, and specific examples to reflect the learning goals, assignments, and discipline of their own course. Bullet points in each example are illustrative rather than exhaustive — add, remove, or revise them as needed. Instructors may also combine elements of both examples (e.g., permitting AI use for some assignments and prohibiting it for others within the same course).

Example 1: AI Not Permitted

The following is a sample policy that instructors may adapt for their syllabi. Language, scope, and specific restrictions should be tailored to fit the goals and requirements of individual courses.

Artificial Intelligence (AI) Policy

Artificial intelligence (AI) tools, including but not limited to ChatGPT, Claude, Copilot, Gemini, and other generative AI systems, may not be used to complete work for this course. All assignments, examinations, discussion posts, projects, and other submitted work must reflect your own knowledge, understanding, analysis, and writing.

This policy is designed to support the learning goals of this course by ensuring that you develop and demonstrate the skills, knowledge, and critical thinking abilities associated with completing course assignments.

Examples of uses that would not be permitted under this policy include the following. [Instructors may add, remove, or adjust these examples based on the nature of their assignments.]

Generating, drafting, or completing any portion of an assignment or submission
Creating outlines, brainstorming ideas, or organizing content for submitted work
Summarizing, paraphrasing, or analyzing course readings or other sources
Editing, rewriting, or revising your writing
Generating code, solving problems, or providing answers to course activities
Completing any other task that substitutes AI-generated content for your own work

Submitting AI-generated material as your own work is a violation of the Academic Integrity Policy and may be considered academic misconduct. Violations will be addressed according to the procedures and consequences outlined in that policy.

Because AI technologies continue to develop rapidly, questions may arise about whether a specific tool or use is permitted. If you are unsure whether a resource involves AI assistance, consult your instructor before using it. When in doubt, ask before submitting your work.

Example 2: AI Permitted

The following is a sample policy that instructors may adapt for their syllabi. Language, scope, and specific allowances should be tailored to fit the goals and requirements of individual courses.

Artificial Intelligence (AI) Policy

Artificial intelligence (AI) tools, including but not limited to ChatGPT, Claude, Copilot, Gemini, and other generative AI systems, may be used in this course as learning tools when used responsibly and ethically. AI can support learning by helping you explore ideas, ask questions, practice skills, and engage with course materials; however, it does not replace your responsibility to think critically, evaluate information, and produce your own original work.

Unless otherwise indicated for a specific assignment, examples of activities where AI tools may be used include the following. [Instructors may adjust these examples to reflect what is and is not appropriate for their specific assignments and disciplines.]

Brainstorming ideas or developing possible approaches to a topic
Creating outlines or organizing your own ideas
Clarifying concepts, terminology, or course material
Reviewing your writing for grammar, structure, or clarity
Supporting studying, practice, or review activities

Examples of uses that would not be permitted include:

Generating complete assignments, papers, discussion posts, projects, or other work submitted for evaluation
Replacing your own analysis, interpretation, or problem-solving
Creating citations, sources, or factual information without verification
Avoiding engagement with required course readings, activities, or learning objectives

If AI is permitted for a specific assignment, any use must be disclosed according to your instructor's requirements. You are responsible for evaluating and verifying any AI-generated content before using it. AI systems may produce inaccurate information, biased responses, fabricated sources, or content that raises copyright, privacy, or ethical concerns.

Do not enter confidential, sensitive, or personally identifiable information into public AI tools. Information entered into these systems may be stored, processed, or used by third-party services.

Responsible AI use requires transparency, critical evaluation, and an understanding of both the benefits and limitations of these technologies. If you are uncertain whether a particular use of AI is appropriate for an assignment, ask your instructor before submitting your work.

Push it beyond the syllabus to the assignment level

Course-level teaching guidance is increasingly explicit that a single syllabus paragraph isn't enough on its own — the same policy needs to be restated in plain terms on the assignment itself, since that's the moment a student is actually deciding what to do. Notre Dame's teaching center frames this directly: ambiguity should be resolved assignment by assignment, not left to a single blanket statement early in the term.[6]

A practical way to do this: repeat a one-line version of your policy at the top of every assignment prompt, not just the syllabus — for example, "AI use for this assignment: [permitted for outlining only / not permitted / fully open, cite your use]."

Sample disclosure line (Level: limited use)	Sample disclosure line (Level: open use)	Why disclosure formats need to be simple
"This assignment used AI for the following components: [outlining / editing / sentence-level revision]. Tool used: [name]. No AI-generated sentences appear in my final submission without direct revision." Adapted from Skidmore College sample AI syllabi statements	"AI contributions to this work: [brief description]. I take full responsibility for the accuracy, originality, and quality of everything submitted, AI-assisted or not." Adapted from course-level templates, University of Idaho AI Resource Guide	A study of mandatory AI-use declarations at a UK business school found 74% of students failed to declare AI use even when a declaration was required — a strong argument for making your documentation method as low-friction as possible rather than a formal citation exercise. Coursework coversheet non-compliance study, Assessment & Evaluation in Higher Education, 2024

A more detailed Transparency Statement required of students especially for writing intensive assignments might include something like this:

If you do use Generative AI such as Microsoft Co-Pilot or Chat GPT to support your process (e.g. for brainstorming or revision), you must not only cite this usage but also provide an “AI Transparency Statement.” In this statement, you must explain: (1) which tool you used, (2) how you used AI, and 3. why you used AI. In addition to requiring this statement attached to the end of your essay, I will also require that you submit (1) your initial draft, (2) a record of your AI prompts/feedback, and (3) a paragraph reflection on your revisions, explaining how/why this tool was helpful or not helpful to you in the context of your rhetorical choices (Adapted from Eugenia Novokshanova and Michelle Kassorla, “Practical Tips for AI Teaching” (licensed under CC Attribution-Non-Commercial-Share-Alike 4.0)).

What Peer Institutions Are Doing

Teaching centers across the country are landing on remarkably similar guidance — encourage, don't mandate; require transparency; and repeat the message all semester. Click any institution name to open its guidance page.

Institution	Notable approach
Stanford Teaching Commons	Treats AI policy-writing as an extension of general effective-syllabus design: student-centered language, transparency, worked examples.
Univ. of Pittsburgh CTL	Organizes guidance around three pedagogical ideals — transparency, relevance, agency — and recommends posting the policy in every assignment area.
Kent State CTL	Offers three paths (adapt examples, use a guided builder, consult departmental colleagues) plus a rationale-and-visibility checklist.
Columbia CTL	Runs hands-on faculty workshops where instructors leave with a drafted policy and early-semester talking points.
Duke CTL	Curates examples from a crowd-sourced, cross-institutional document; encourages research before drafting.
Univ. of Maryland TLTC	Strongly encourages (not mandates) a course-specific policy; provides sample language by permissiveness level.
Harvard Bok Center	Direct, simple encouragement for every instructor to include a policy outlining acceptable and unacceptable uses.
Mesa Community College CTL	Emphasizes early, repeated communication and a fair, evidence-gathering process for suspected violations.
Mercer University CTL	Frames policy options along a permissiveness spectrum; flags enforcement risk of strict prohibition.

What the field's leading bodies recommend

Beyond individual CTLs, the national organizations that set standards for teaching, learning, and faculty development have issued their own guidance.

AAC&U

American Association of Colleges & Universities

Its national survey work (with Elon University's Imagining the Digital Future Center) is the largest current dataset on faculty AI attitudes — 1,057 faculty surveyed in late 2025 alone.
AAC&U frames AI competencies and literacies as learning outcomes institutions should adopt at the course and program level, not just a compliance issue.
Its Teaching with AI guide (2nd ed., Bowen & Watson) is a widely cited practitioner resource on assignment redesign and academic integrity strategy.
Runs a dedicated annual Conference on AI and Higher Education and a multi-month Institute on AI, Pedagogy, and the Curriculum for cross-campus teams.

AAC&U, Leading Through Disruption (2025) & The AI Challenge (2026); Bowen & Watson, Teaching with AI, 2nd ed.

POD Network

Professional & Organizational Development Network in Higher Education

North America's largest educational-development community — over 1,700 members across roughly 1,300 teaching centers — maintains a dedicated AI in Education Special Interest Group (SIG).
The SIG's scope explicitly includes assignment design, faculty labor implications, bias/equity concerns, copyright, and academic integrity — signaling that POD treats AI policy as a faculty-development issue, not just a compliance one.
POD's evidence-based Center for Teaching and Learning Matrix (with the American Council on Education) offers CTL directors a benchmarking frame that can incorporate AI-related programming and goal-setting.

POD Network, AI in Education SIG; POD/ACE Center for Teaching and Learning Matrix

AAUP

American Association of University Professors

Its 2025 report on AI and Academic Professions recommends more critically informed, holistic professional development for faculty — including how AI is already embedded in ed-tech business models, not just how to use the tools.
Centers faculty governance: policy development should be led by the people who understand teaching and learning conditions firsthand, not imposed by administration or vendors alone.
Raises questions CTLE guidance should anticipate — data use, disclosure, and the right for both students and faculty to identify "AI-aware" versus "AI-free" courses.

AAUP, Artificial Intelligence and Academic Professions (TREP Report), July 2025

What the peer-reviewed literature says

Practitioner guidance is useful, but it's worth grounding your policy in what's actually been studied, including the limits of what's known so far. [Click to drop down]

Students notice your policy even when they don't notice the university's

A mixed-methods study of 124 undergraduates and 7 faculty at two US research universities found that students consistently recognized classroom-level AI policies, while institutional-level guidelines remained unclear to them — less than half of surveyed students considered university guidance clear. This is direct evidence that your course statement, not the institution's, is doing the actual communicative work. Jiang et al., Higher Education Quarterly, Wiley, 2025

A clear classroom policy measurably changes how students use AI

In the same study, students who were aware of their course's AI policy were significantly less likely to use AI tools for writing and research than students who weren't — one of the few findings in this literature that directly links a course-level statement to a behavior change, not just an intention. The same study also found 81.5% of students worried that using AI could get them perceived as cheating by instructors regardless of the actual policy, underscoring how much clarity matters. Jiang et al., Higher Education Quarterly, Wiley, 2025

Students often don't disclose AI use — even when policy allows it

A CHI 2025 study of student and instructor perspectives on undisclosed generative AI use found most non-disclosing students weren't primarily hiding rule-breaking — they avoided disclosure out of fear of judgment (54%), fear of penalty (44%), or because they saw AI use as comparable to using Grammarly or other established tools (66%). Most reported not feeling guilty about non-disclosure at all. This suggests disclosure requirements need to be framed as low-stakes and normal, not confessional, to actually work. CHI 2025 Conference on Human Factors in Computing Systems, ACM Digital Library

Formal declaration requirements are widely ignored unless friction is low

A study of mandatory AI-use declarations attached to coursework coversheets at a UK business school found 74% of students failed to declare AI use despite the requirement being explicit. The takeaway for individual instructors: a disclosure mechanism only works if it's built into the normal submission flow, not added as a separate compliance step students can skip. Assessment & Evaluation in Higher Education, Taylor & Francis, 2024

A simple disclosure form, piloted in one course, surfaced real classroom patterns

A brief research report piloting a mandatory AI Disclosure Form in a single general chemistry and citizen-science honors course found students commonly reported pasting assignment questions directly into AI tools word-for-word — a concrete, course-level finding that shaped how the instructor subsequently redesigned assignment prompts. A small, low-overhead classroom tool surfaced actionable information a syllabus statement alone would not have. Frontiers in Education (STEM Education), brief research report, 2025

Students use AI heavily but don't trust its output

A survey of 132 UK undergraduate and postgraduate students found a "widely used but barely trusted" pattern: students readily use generative AI for essays, outlines, and overcoming writing blocks, yet fewer than 25% regard its outputs as reliable. That gap is worth naming explicitly in your policy's rationale — permitting or restricting AI use isn't the same conversation as teaching students when to trust what it produces. "Widely used but barely trusted," Educational Studies, Taylor & Francis, 2025

The minimum viable classroom policy has three parts

Reviewing course syllabi across physiology programs, this study proposed that any classroom-level AI statement needs at least three elements to function: acceptable use, citation or disclosure requirements, and consequences for non-adherence — while cautioning that even a well-built three-part statement will need periodic revision as tools and classroom norms keep shifting. Frontiers in Education, "Do I need an AI policy in the course syllabus?", 2026

The AI Assessment Scale, as published research

Originally proposed in late 2023 and revised in 2025, the AIAS is one of the few classroom-facing frameworks with a direct publication trail and documented uptake — implemented in hundreds of individual courses and translated into more than 30 languages, per the authors' own account of its adoption. Perkins, Roe & Furze, Journal of University Teaching and Learning Practice, 22(7), 2025

References