Alpna Bhatia is an Associate Teaching Professor of Economics at the University of Colorado Boulder.
If you have spent time wondering how to write an AI policy for your syllabus, you are not alone. Generative AI has moved faster than most course design cycles, and many of us are being asked to make high-stakes decisions about tools that are rapidly changing and that our students are already actively using. In this fast-changing environment, the syllabus matters more than ever. It is often the first place where students encounter our expectations, our boundaries, and our explanation of what learning in the course will require. Our goal is not to predict every possible use of AI or write a policy that lasts forever. Our goal is to use the syllabus to give students a clear, usable framework for when AI supports learning, when it interferes with learning, and how they should be transparent about their use.
A strong AI policy starts with us, the instructor. We may be fully on board. We may worry that permitting AI will weaken student learning, or that prohibiting it is unrealistic. Many of us are also uneasy about enforcement because AI detection tools are unreliable and can create unfair accusations. Those concerns are signs that we understand the stakes. The most useful reframe, for me, is this: Our AI policies can be pedagogy statements. They tell students what kind of learning matters in our courses, what intellectual work they are responsible for doing themselves, and where AI may function as a tool rather than a substitute.
Before We Write the Policy: Complete an Instructor Self-Audit
Before we write syllabus language about AI, we need to work through a self-audit of our own intentionality: our reasons for embracing it, hesitating about it, or landing somewhere in between. Some of us see real possibilities for access, feedback, revision, and applied practice. Some of us worry about shortcuts, uneven preparation, privacy, accuracy, or the loss of important intellectual struggle. A self-audit helps us name those instincts instead of letting them quietly shape our policy. The purpose is to decide whether AI belongs in our course at all, what role it should play, and what conditions are necessary for its use to support rather than displace learning.
- Purpose: What learning goal, if any, does AI serve here?
- Task fit: If you’re using AI, is AI acting as tutor, coach, brainstorming partner, editor, simulator, or generator for most of the course?
- Student readiness: Do students know enough to evaluate AI output critically?
- Transparency: How should students disclose what they used and how they used it?
- Equity and access: Will paid tools, prior experience, language background, or accommodations create uneven advantages?
- Ethics and privacy: Are students being asked to enter sensitive, personal, or protected information?
If we cannot answer those questions clearly, hesitation can be pedagogically useful.
Write Policy Language Students Can Actually Use
Once we know our learning goals, a syllabus AI policy should be brief, specific, and connected to those goals. It should answer four practical questions for students: What kinds of AI use are allowed? What kinds are not allowed? How should they disclose AI use? What are they still responsible for? The syllabus should not try to answer every AI question for every task. Instead, it should give students the course-level philosophy and a set of categories they will see throughout the course.
A policy that says “use AI responsibly” sounds reasonable, but students may not know what that means. A more useful version says something like: “You may use AI to brainstorm examples, but not to draft your final analysis. If you use AI, explain what you asked it to do, what you kept, what you changed, and how you checked the result.” That kind of specific language is important to teach students where the boundary is between support and substitution.
The syllabus does not need to list every possible AI decision, but it can give students a course-level map. I like to name broad categories students will see across the course: For example: AI prohibited for quizzes, exams, or foundational problem sets; AI limited for brainstorming, outlining, grammar support, or study practice; and AI integrated for assignments where students must use, critique, document, and reflect on AI output. The reason being that not every assignment in the course falls under the same AI rule. In fact, a single course may need several categories, and naming those categories can help students understand that our decisions are intentional rather than arbitrary, even when rules vary by task. Further, while we may want a comprehensive course policy, students make decisions assignment by assignment. Students may reasonably ask, “If AI can do this, why do I need to learn it?” One of the clearest ways we can make an AI policy meaningful is to explain why some work must be done without AI, while some can be done with it (if you are okay with it).
A simple disclosure requirement can also reduce confusion: “If you use AI, include a short note explaining the tool used, the purpose of use, and what you changed or verified.” I like this move because it shifts the conversation from catching students to teaching transparent academic practice. A short disclosure note might look like this: “I used ChatGPT to brainstorm examples of opportunity cost for this discussion post. I selected one example, checked it against the course definition, and rewrote the explanation in my own words. The final interpretation and wording are my own.”
I do not want disclosure to feel like a confession box. I want it to feel like part of the learning process: here is how I used the tool, here is what I checked, and here is what I am still responsible for. That is what I mean by visibility, not surveillance. If students think any mention of AI will automatically get them in trouble, they will hide their use or avoid asking questions. But if we explain that disclosure is a normal part of responsible academic work, students are more likely to see it as a way to show their process and their judgment.
Finally, I do think we need to connect our course language to institutional academic integrity procedures. Students also need to know that confusion, misuse, and misconduct are not always the same thing. A student who is not sure how to cite an AI tool, a student who uses AI beyond the limits of an assignment, and a student who intentionally submits AI-generated work as their own may call for different responses. Making that distinction helps us be fair and follow campus procedures when a real violation may have occurred.
For reference, here’s an example of what the beginning of an AI policy may look like:
In this course, my goal is for you to develop [specific skill or knowledge]. Because that learning matters, some assignments must be completed independently and without generative AI. These assignments will be clearly marked.
Other assignments may allow AI for specific purposes such as brainstorming, outlining, feedback, revision, or comparison. When AI is permitted, you remain responsible for the accuracy, evidence, reasoning, and integrity of your submitted work.
If you use AI, include a brief disclosure naming the tool, describing how you used it, and explaining what you revised, verified, or contributed yourself. This disclosure is intended to make your learning process visible, not to penalize permitted use. Using AI in ways not permitted by the assignment may be treated as academic misconduct.
Make the Policy Visible
I teach online. In an online course, students often meet our expectations through the syllabus, the learning management system, and assignment pages rather than through a first-day conversation in the classroom. That means we cannot bury the AI policy in a long academic integrity section and hope students find it when they need it. The course shell has to do some of the teaching, and making AI policy, which may differ for students course to course, assignment to assignment, visible is important. This clarity matters in every course, not just online.
Put the policy in multiple places. Include it in the syllabus, the orientation module, and each major assignment. Students taking (online) courses often navigate by task, not by rereading the full syllabus.
Record a short explanation. A five-minute welcome video can explain not only the rules but our reasoning. Students are more likely to follow a policy when they understand the learning purpose behind it.
Use a low-stakes check-in. In week one, ask students to answer a short prompt: “Which part of the AI policy is clearest? Which part might you need help applying?” This surfaces confusion before a major assignment.
Design for visibility. Use draft checkpoints, short video or audio explanations, annotated problem-solving, or reflection notes to check students’ AI practices. These are not just integrity measures; they are good online teaching practices because they help us see how students are learning.
Revisit expectations at the point of use. Before a paper, project, discussion, or exam, remind students what AI use is allowed for that task. Students may be juggling different AI rules across courses, so repetition is not redundancy; it is clarity.
I do not think we need one perfect AI statement for every course. The best syllabus policies are about clarity, trust, and learning. The AI policy should help students see that knowledge still matters in an AI-rich world because AI makes judgment more important, not less. When students know why it matters, and when a tool may or may not support that work, the syllabus becomes part of teaching.
MEET THE AUTHOR
Alpna Bhatia is an Associate Teaching Professor of Economics at the University of Colorado Boulder. Her work focuses on designing engaging, student-centered learning experiences, with an emphasis on student metacognition and equitable teaching practices that help students use AI intentionally and thoughtfully. She served as a CU Boulder AI Ambassador in 2025–2026 and will continue this work as a CTL Faculty Fellow and mentor in the CU Boulder AI Literacy Ambassadors Program in 2026–2027. Known for a structured and supportive approach that balances clarity with accountability, Alpna’s work spans teaching innovation, scholarship of teaching and learning, and student mentorship, including leadership in Women in Economics and the Economics Club.