If your institution has not settled on a college-wide generative AI policy, your students still need to know what is expected in your course. Silence does not create a neutral position. It leaves students to interpret a general academic integrity statement, compare rules from other classes, or guess whether using an AI tool for brainstorming, revising, translating, or drafting crosses a line.
That is a difficult position for students, and it is not especially useful for faculty either. A policy that is unclear before an assignment is submitted becomes much harder to enforce afterward.
You do not need to resolve every debate about artificial intelligence to write a workable syllabus statement. You need to make a smaller set of decisions: What must students learn or demonstrate themselves? Where could AI support that learning? Which uses would bypass the work the assignment was designed to assess? How should students document permitted use? What process will you follow if the boundary is crossed?
Those questions produce a much stronger policy than beginning with whether you are generally “for” or “against” AI.
Begin With the Rules Above Your Course
Before drafting, check the policies that already govern you. Look at the college catalog, academic integrity code, faculty handbook, department guidance, accessibility and privacy rules, and any statements from the provost, academic senate, or teaching and learning center. Ask your chair or dean whether informal guidance is being used while a formal policy is under development.
This step matters because a syllabus can clarify institutional rules, but it should not quietly replace them. For example, you may be able to set assignment-specific expectations, but the official academic integrity process may determine how suspected violations are documented and reviewed. Likewise, your institution may restrict the submission of student work to third-party systems or require approved versions of certain tools.
If no broader AI policy exists, say that your course statement operates within the institution’s existing academic integrity, privacy, and accessibility policies. This keeps the syllabus connected to established procedures.
Start With the Learning, Not the Tool
The most useful question is not, “May students use ChatGPT?” It is, “What intellectual work must the student perform for this assignment to count as evidence of learning?”
Imagine that students write three different kinds of assignments in the same course:
- A low-stakes discussion post checks whether they understood a reading.
- A research proposal assesses whether they can narrow a question, locate credible sources, and design an appropriate method.
- A final presentation assesses whether they can explain a complex idea clearly to a particular audience.
A single sentence allowing or prohibiting AI for the entire course may not fit all three. Generating a discussion response could replace the thinking you intended to observe. Using AI to produce possible research questions might be acceptable if students must critique and refine them. Rehearsing likely audience questions before a presentation might support the learning goal rather than undermine it.
This is why the University of Illinois ATLAS guidance recommends aligning course AI rules with learning goals, assessments, and disciplinary norms. It also suggests distinguishing among allowed, conditionally allowed, and prohibited uses. That framework is practical because it treats AI use as an activity, not simply as the name of a product.
Write Assignment-Level Boundaries
Begin with a short course-level default, then add a clear note to each major assignment. The note should answer four questions:
- Is generative AI allowed for this assignment?
- If it is allowed, for which parts of the process?
- What must remain the student’s own work?
- What disclosure, citation, or documentation is required?
Specific language is better than broad language. “AI may be used responsibly” does not tell a student whether asking for an outline is permitted. “AI is prohibited” may still leave questions about grammar correction, predictive text, translation, transcription, or accessibility tools.
Here are three possible labels:
AI not permitted: This assignment is designed to assess your independent analysis of the assigned readings. Do not use generative AI to generate, outline, draft, revise, or summarize any portion of your response. Standard spelling and grammar checking are permitted.
AI permitted for limited purposes: You may use generative AI to brainstorm possible topics and to identify questions that require further research. You may not submit AI-generated language or treat an AI response as a source. Verify all claims through assigned readings or credible sources, and include the disclosure described below.
AI integrated into the assignment: You will use an approved AI tool to generate two responses to the case. Your grade will be based on how well you identify unsupported assumptions, verify factual claims, compare the responses with course concepts, and produce your own recommendation.
The exact categories matter less than using them consistently. A short label at the top of every assignment can prevent students from searching through the syllabus each time.
Distinguish Disclosure From Citation
Faculty sometimes ask students to “cite AI” when what they actually need is a record of how it was used. These are related, but different, expectations.
A citation identifies material treated as a source according to a required style. A disclosure explains the role the tool played in the student’s process. If you require disclosure, give students a format. For example:
I used [tool and version, if known] on [date] to [brainstorm questions, revise organization, generate code, or another purpose]. I reviewed the output, verified relevant claims using [sources or method], and take responsibility for the submitted work.
For a major project, you may also ask students to retain prompts and relevant outputs. Do not require pages of transcripts unless you will actually use them. The goal is meaningful transparency, not paperwork that neither you nor the student will revisit.
Address Accuracy, Privacy, Access, and Ownership
A complete policy should cover more than cheating.
Students remain responsible for factual errors, fabricated sources, biased output, weak reasoning, and inappropriate language in anything they submit. Say this directly. Permission to use a tool is not permission to outsource judgment.
Also tell students not to enter confidential, proprietary, personally identifiable, or sensitive information into unapproved tools. If an assignment involves field placements, clinical work, student records, unpublished research, or community partners, this boundary is especially important.
If you require a particular AI platform, confirm that students have equitable access and an alternative if disability, cost, privacy, account, or technology constraints make the tool unavailable. HEPI’s 2026 survey of 1,054 full-time UK undergraduates found near-universal use, but it also showed uneven institutional support. The finding should not be generalized automatically to every campus, yet it reinforces a basic point: use is widespread, while access, skill, and confidence are not necessarily equal.
Finally, consider intellectual property. Students should know whether their work will be uploaded to an external service, whether the institution has approved that service, and what alternative exists if they do not consent. Consult institutional guidance rather than improvising legal language.
Explain How Concerns Will Be Handled
Do not promise that an AI detector will decide whether misconduct occurred. Detection tools can be part of a conversation or review where institutional policy permits them, but a score should not substitute for evidence or due process.
Your syllabus can state that suspected unauthorized use will be addressed under the existing academic integrity procedure. You can also explain the evidence you may request, such as drafts, notes, source records, version history, or a conversation in which the student explains the submitted work.
This approach is more defensible than inventing a separate penalty for AI. It also gives students a clearer understanding of the process before a concern arises.
Teach the Policy, Then Revisit It
A paragraph on page nine of a syllabus is not enough. Discuss the policy during the first week, show students two or three realistic examples, and ask them to apply the rule. NC State’s teaching guidance recommends concrete course and assignment directions, including examples of how students should document their use.
Revisit the rule when the first major assignment is introduced. If students ask a question that exposes an ambiguity, clarify it in writing for the entire class. Keep the clarification with the assignment so that everyone receives the same guidance.
At the end of the term, review where the policy created confusion. You may discover that a prohibited-use statement was clear, but the disclosure instructions were not. Or you may find that an assignment no longer measures what you intended once a particular AI use is permitted. In that case, revise the assignment along with the policy.
A Practical Policy-Building Checklist
Before adding the statement to your syllabus, confirm that you have:
- Checked institutional and departmental rules.
- Identified the learning students must demonstrate independently.
- Set a course-level default.
- Added assignment-specific permissions and prohibitions.
- Defined what counts as generative AI for your purposes.
- Distinguished ordinary editing or accessibility tools where appropriate.
- Supplied a usable disclosure or citation format.
- Addressed verification, privacy, access, and responsibility for output.
- Connected suspected misuse to the established academic integrity process.
- Placed the guidance in the syllabus and in assignment instructions.
- Added the semester or date when the policy was last reviewed.
You do not need a perfect policy. You need one that reflects the actual learning in your course, gives students enough information to make a responsible decision, and gives you a consistent basis for responding when questions arise. Write it as instructional guidance, not as a warning label. The strongest AI policy does more than prohibit misconduct. It tells students what responsible work looks like in your discipline.
Sources
- Higher Education Policy Institute, “Student Generative Artificial Intelligence Survey 2026.”
- University of Illinois ATLAS, “AI Use Policy Guidance & Template.”
- NC State DELTA, “Developing Course AI Guidelines & Driving Class AI Discussion.”
