AI Syllabus Guide

Drop-in syllabus and assignment language for faculty teaching in an age of AI. This guide is organized around the decisions faculty actually have to make: what kind of course you are teaching, what work you are assessing, what AI uses you want to permit or prohibit, and what students must disclose.

The guide draws especially on Kirsten K. Davis’s Guide to Syllabus and Assignment Generative AI Policies and Advice, Korin Munsterman’s GenAI in Legal Education: A Practical Guide for Professors and Students, and teaching materials shared by Professor Cathie Struve on attribution and disclosure in writing-intensive seminars. The Davis and Struve materials are cited here as source materials for this guide, not as Penn Carey Law policy.

This document is not official Law School policy and does not authorize AI use. Faculty should state course and assessment rules clearly, consistent with current Registrar and University guidance and any applicable clinic, placement, client, supervisory, or professional requirements.

Jump to sample language

• Traditional lecture course with a final exam
• Seminar with a research paper
• Research, writing, or skills assignments
• Courses with multiple assessment types
• Permitted-uses table
• Disclosure language
• Design options
• Complete assignment packages

The short version

Say what students may do. Say it at the assignment level when rules differ. Design the work so your policy is credible.

A blanket syllabus sentence is rarely enough. Students need to know whether AI may be used for studying, brainstorming, research, outlining, drafting, revising, cite-checking, grammar, simulations, practice questions, and uploading course materials. They also need to know when disclosure is required.

A good course AI policy does four things:

1. Defines permitted, permitted-with-disclosure, and prohibited uses.
2. Separates study use from submitted work and exams.
3. Makes students responsible for accuracy, sources, confidentiality, and judgment.
4. Gives a concrete disclosure form when disclosure is required.

Do not build the policy around AI detection. Detector performance varies by tool, model, genre, language, threshold, and mixed authorship; do not treat a score as standalone proof of misconduct. Better design choices include aligned process evidence, source checks, critique of AI output, and assignment-specific disclosure.

Start with the learning goal

Before choosing policy language, decide what the assignment is meant to measure.

If the goal is independent recall, first-pass legal analysis, exam performance, or fluency without assistance, prohibit AI for that task. If the goal is research strategy, revision judgment, professional communication, or critique, AI may be useful if the assignment requires students to supervise and verify it. If the goal is AI fluency itself, make AI part of the task and grade the student’s judgment, not the polish of the machine output.

The key move is to regulate the work, not the technology in the abstract. “No AI” is sometimes right. “Use AI, but show your work” is sometimes right. “Use AI because evaluating it is the assignment” is sometimes right.

Policy models

The publications behind this guide use slightly different taxonomies, but they point toward the same practical menu.

Davis’s guide usefully frames three common choices: allow AI for research but not writing, provide a menu of permitted uses, or permit broad use while requiring students to account for it. Munsterman’s guide adds a teaching-design frame: GenAI-prohibited, GenAI-resistant, GenAI-transparent, and GenAI-integrated assignments. For syllabus purposes, those approaches can be translated into the following working categories.

You can use more than one model in the same course. A seminar might prohibit AI for a short in-class diagnostic, allow AI-assisted brainstorming for a paper topic, require disclosure for draft feedback, and require AI critique in one class exercise.

Course policy or assignment policy?

Use a course-level policy for stable defaults. Use assignment-level language whenever the rule changes.

A course-level policy should cover:

• Whether AI is permitted for studying or class preparation.
• Whether students may upload course materials, recordings, slides, or notes.
• The default rule for submitted work.
• The default rule for exams and quizzes.
• Whether disclosure is required.
• What students should do when they are unsure.

Assignment-level instructions should cover:

• Which AI uses are permitted for that task.
• Whether students may use AI before or after creating their own first draft.
• Whether students must save prompts, outputs, chat logs, revision history, or a process memo.
• What form of disclosure is required.
• How AI-related process work will affect grading.

Common questions students will ask

A clear policy should answer these questions without requiring a separate conversation:

1. May I use AI to study for class?
2. May I upload the syllabus, slides, readings, class recordings, or my notes to an AI tool?
3. May I use AI to brainstorm paper topics or research paths?
4. May I use AI to find sources, summarize sources, or check citations?
5. May I use AI to outline, draft, revise, edit, translate, or polish submitted work?
6. Must I disclose AI use even if no AI-generated wording appears in the final submission?
7. Which tools are allowed, and are any tools prohibited?
8. What happens if I am unsure whether a proposed use is permitted?

Sample syllabus language

The blocks below are starting points. Keep the language that does real work; replace bracketed terms; add assignment-specific rules where needed.

1. Traditional lecture course with a final exam

For a doctrinal lecture course where the final exam is the main assessment, separate study use from exam use. The syllabus can permit AI as a study aid while preserving a strict exam rule.

Artificial Intelligence Policy. This course has no graded written assignments other than the final exam. You may use generative AI tools as study aids, subject to the limits below. For example, you may ask an AI tool to generate practice hypotheticals, quiz you on rule elements, help you compare outlines, or explain a doctrine after you have read the assigned materials.

You may not use AI as a substitute for the assigned reading, rely on AI summaries instead of your own case analysis, or treat AI output as authority. AI tools can be incomplete, misleading, or wrong. You are responsible for checking any AI output against the assigned materials and your own notes.

Do not upload nonpublic course materials, recordings, student work, exam questions, model answers, or grading materials to an AI tool unless I have expressly authorized that use and the tool has been approved for that type of data.

The final exam AI policy will be stated in the exam instructions. Unless those instructions expressly permit AI use, you may not use generative AI tools, chatbots, AI-assisted search, or AI drafting tools during the exam.

Optional short version:

You may use AI tools to study for this course, but not during the final exam unless the exam instructions expressly allow it. AI study use must supplement, not replace, your own reading, outlining, and analysis. You remain responsible for the accuracy of anything you learn from an AI tool.

Optional exam instruction:

Exam AI Rule. You may not use generative AI, chatbots, AI-assisted search, AI drafting tools, or similar systems during this exam. The exam assesses your own understanding under the stated exam conditions. Use of AI during the exam is prohibited unless these instructions expressly say otherwise.

If you plan to allow AI on an exam, write that rule separately and with precision. Say which tools are permitted, whether internet access is permitted, whether students may upload the exam question, whether students must disclose AI use in the answer, and whether AI output may be quoted, paraphrased, or used only as a reference.

2. Seminar with a research paper

A seminar paper needs a process policy. Students need to know what AI may do at each stage: topic development, source identification, thesis formation, outline, draft, revision, and final submission.

Artificial Intelligence Policy. You may use generative AI as a limited research and writing-process assistant in this seminar. You remain responsible for the thesis, research judgment, organization, legal analysis, citations, and final text of your work.

Unless an assignment says otherwise, you may use AI to brainstorm possible topics, identify counterarguments, test whether an outline is clear, ask for feedback on a draft you have already written, or generate questions you should consider as you revise. You may not submit unreviewed AI-generated text, use AI-generated citations without checking them, rely on AI summaries instead of reading the relevant sources, or allow AI to replace your own research and analysis.

For each major paper milestone, I may ask you to submit process materials such as a topic memo, preliminary source list, thesis paragraph, outline, partial draft, research trail, or revision memo. These materials help show how your project developed and give us better material for conferences and feedback.

If you use AI in connection with any submitted work, including a topic memo, presentation, outline, draft, or final paper, include an AI-use statement with the submission. The statement should identify the tool, the stage of work, the purpose, whether any AI-generated language appears in the submission, and how you checked sources, quotations, factual claims, and legal authorities.

Optional research-but-not-writing rule:

You may use AI to generate research leads, identify possible counterarguments, and ask background questions. You may not use AI to draft, rewrite, or polish the submitted paper. Any source, case, article, quotation, statistic, or factual claim surfaced by AI must be independently located, read, and verified before you rely on it.

Optional process-checkpoint language:

For the paper, you will submit [topic memo / source list / thesis paragraph / outline / partial draft / conference memo / final paper]. At each stage, identify any AI assistance you used. I may ask you to discuss your process in conference. The goal is not to punish responsible AI use; the goal is to make your research and writing choices visible.

The attribution-and-disclosure model shared by Professor Cathie Struve is a strong version of this approach. The Project’s AI Use Policy Templates reproduce her syllabus block and provide sample AI-use statements.

3. Research, writing, or skills assignments

For writing and skills courses, the AI rule should track the learning goal. If the assignment assesses first-pass legal analysis, restrict drafting. If the assignment assesses revision, client communication, negotiation strategy, or professional judgment, AI can be part of the exercise.

If the goal is independent first-pass analysis

Artificial Intelligence Policy for This Assignment. This assignment assesses your own legal analysis and writing. You may use ordinary spelling, grammar, and citation-formatting tools, but you may not use generative AI to draft, revise, organize, polish, translate, or rewrite the submitted work. You may use AI only for general background questions that do not involve the facts, law, or substance of this assignment.

If you are unsure whether a tool or use is covered by this rule, ask before using it. Do not assume that a tool is permitted because it is built into a word processor, research platform, or search engine.

If the goal is revision or professional judgment

Artificial Intelligence Policy for This Assignment. You must prepare your own first draft before using AI. After you have a complete draft, you may use AI to critique clarity, organization, tone, concision, counterarguments, or audience fit. You may not ask AI to write the assignment for you, and you may not paste AI-generated language into your submission without revising, checking, and disclosing it.

With your final submission, include a short reflection explaining what AI feedback you requested, which suggestions you accepted, which suggestions you rejected, and why. Your grade will turn on your judgment, revision choices, and final work product.

If AI is part of the exercise

Artificial Intelligence Policy for This Assignment. AI use is part of this assignment. You will generate, receive, or evaluate AI output and then assess it using the legal authorities, facts, client goals, and professional standards relevant to the problem. Your grade will depend on your critique, corrections, verification, and explanation, not on whether the AI output sounds polished.

You must identify any hallucinated citations, unsupported factual claims, missing authorities, weak reasoning, confidentiality concerns, or professional-responsibility issues. You are responsible for the final analysis.

Example integrated assignments:

• Give students an AI-generated answer and ask them to grade it against a rubric.
• Ask students to identify hallucinated cases or unsupported factual claims.
• Have students compare their own first draft with AI feedback and explain what they changed.
• Ask students to prompt an AI tool for a client letter, then revise it for accuracy, tone, and professional judgment.
• Have students build a research trail showing which AI-generated leads were useful, which were false, and which required human source checking.

Complete assignment packages

The short examples above establish policy direction. For a ready-to-adapt sequence, student instructions, disclosure, deliverables, rubric, class-size variants, accessibility guidance, and evidence limitations, use the package that matches the capability you want to observe:

• Independent First Pass + AI-Supported Revision — preserve evidence of independent analysis, then assess selective revision judgment.
• Audit and Correct AI Legal Work — assess verification, diagnosis, correction, and professional judgment.
• Staged Research or Writing Project — use meaningful checkpoints and stage-specific AI rules for substantial work.
• AI-Permitted Work + Complementary Evidence — assess assisted performance and, only when necessary, personal explanation or application through different evidence.

These are design patterns, not mandatory policies. Select the least burdensome structure that supports the inference you need to make about learning.

4. Courses with multiple assessment types

Many courses need layered rules. A professor might prohibit AI on quizzes, permit AI study aids between classes, require disclosure on practice problems, and integrate AI into one class session on legal research or drafting.

Course AI Policy. Different assignments in this course have different AI rules. The default rule is that you may use AI for studying, brainstorming, and practice, but you may not use AI to draft submitted work unless the assignment instructions expressly permit it. When an assignment has a different rule, the assignment instructions control.

You are responsible for reading the AI rule attached to each assignment. If the rule is unclear, ask before using AI.

You can make the layered rule easier to read with a permitted-uses table:

Davis’s menu approach works especially well here. The point is not to anticipate every tool. The point is to name the recurring activities students actually ask about.

General disclosure language

When disclosure is required, use a simple form students can complete without guessing what counts. Disclosure is not the same as citation. Disclosure tells the professor how AI affected the student’s process. Citation tells the reader where an idea, quotation, source, or authority comes from.

AI-use statement. I used [tool name and version, if known] on [date] for [purpose] at [stage of work]. Specifically, I used it to [brainstorm topics / generate practice questions / critique my outline / review a draft / identify possible counterarguments / check clarity / other]. [No AI-generated language appears in this submission / AI-generated language appears only where quoted or cited / I revised all AI-assisted wording before submission]. I checked all legal authorities, quotations, factual claims, and citations. I take responsibility for the final submission.

For longer projects, ask students to answer five questions:

1. What tool did you use?
2. What did you ask it to do?
3. What output affected the submitted work?
4. How did you verify authorities, quotations, factual claims, and citations?
5. What responsibility do you take for the final work?

Optional certification language:

I certify that this submission complies with the AI policy for this course and assignment. Any AI assistance I used is disclosed above. I understand that fabricated citations, unverified authorities, undisclosed AI-generated text, or use of AI contrary to the assignment instructions may raise academic-integrity concerns.

Privacy and course materials

AI policy is also data policy. Students should not upload material to AI systems unless the course permits it and the tool is appropriate for that material.

Recommended syllabus language:

Do not upload nonpublic course materials, class recordings, transcripts, slides, exams, model answers, grading materials, student work, client information, confidential facts, or personal information about another person to an AI tool unless I expressly authorize that use. If an assignment requires AI use, I will specify what material may be uploaded and which tool or tool category is appropriate.

For public cases, statutes, regulations, articles, and other generally available legal materials, the privacy concern is usually lower. Accuracy still matters. Students must verify any AI-generated account of those materials against the actual source.

Design options that make the policy work

A syllabus policy works best when the assignment design supports it.

Use process evidence. Ask for an outline, revision history, research trail, source list, reflection memo, or short oral explanation. Process evidence makes learning visible and gives you a better basis for feedback.

Make AI an object of critique. Ask students to evaluate an AI answer, identify hallucinated authorities, improve a weak prompt, compare AI output to a model answer, or explain why an AI-generated analysis is persuasive but wrong.

Use local or course-specific facts for relevance. Assignments tied to class discussion, assigned materials, specific hypotheticals, or current course themes can improve alignment. Do not assume specificity prevents outsourcing or establishes authorship.

Require source verification. If AI supplies a case, quotation, statistic, factual claim, or scholarly argument, students should trace it to a real source before using it. AI output is a lead, not authority.

Use a structured complement when needed. A conversation, written explanation, or new application may provide complementary evidence of the student’s reasoning. It is not authorship proof. Oral components require transparent criteria, practice where appropriate, accommodations, and equitable alternatives.

Use low-stakes practice. Let students use AI for practice questions, self-quizzing, critique, or revision exercises before a high-stakes assessment. This teaches supervision rather than avoidance.

Complete prohibition option

Some faculty will choose a restrictive rule for a course or assignment. If so, define the scope carefully. A broad prohibition can be hard to administer because AI features appear inside ordinary research, writing, and productivity tools.

Artificial Intelligence Policy. Students may not use generative AI tools, chatbots, AI-assisted drafting tools, or similar systems to prepare, draft, revise, edit, translate, or complete any assignment, paper, project, quiz, or examination submitted for this course, unless I expressly authorize that use in writing.

This prohibition includes using AI to generate ideas, outline an answer, draft text, revise text, summarize sources for use in submitted work, create citations, or solve course problems. Ordinary spelling and grammar checking is permitted unless an assignment says otherwise. AI-assisted legal research tools are [prohibited / permitted only for source identification / permitted subject to disclosure].

If you are unsure whether a tool or use is permitted, ask before using it.

Structural supports for a restrictive rule:

• Use in-class writing, oral explanation, or timed components where appropriate.
• Require staged drafts or process evidence for longer work.
• Avoid assignments that can be completed by a generic prompt with little connection to class materials.
• Tell students whether embedded tools in Word, Google Docs, Westlaw, Lexis, Grammarly, or similar platforms are covered.

Enforcement: conversation before accusation

If work appears inconsistent with the policy, start with a conversation. Ask the student to explain their process, sources, revisions, and use of any tools. Where appropriate, ask for revision history, research notes, prompts, outputs, or a short oral explanation.

If concerns remain after an ordinary inquiry, follow the current academic-integrity process and consult the appropriate Law School office. This guide does not define grounds or procedures for escalation. A detector score alone should not be the basis for an accusation.

Possible syllabus language:

If I have questions about whether submitted work complied with the AI policy, I may ask you to meet with me, explain your process, provide process materials, or complete a short follow-up conversation about your work. This is not an accusation. It is a way to understand authorship, learning, and compliance with the assignment rules.

Faculty checklist

Before the semester starts:

• Add a course-level AI policy to the syllabus.
• Decide whether the exam rule is settled now or will be supplied with the exam instructions.
• Identify assignments that need their own AI instructions.
• Decide whether students may upload course materials to AI tools.
• Choose a disclosure format if AI use is permitted.
• Decide what process evidence, if any, you will collect.
• Tell students what to do when they are unsure.

Before each major assignment:

• Restate the AI rule in the assignment instructions.
• Say whether AI may be used before the student’s own first draft.
• Say whether AI may be used for revision, editing, source identification, or citation checking.
• Say whether prompts, outputs, chat logs, or an AI-use statement are required.
• Explain how AI-related work will affect grading.

Sources and related Project materials

This guide aligns with the Fall 2026 teaching guidance on the Penn Carey Law Faculty Pedagogy Resources portal. It draws especially on:

• Kirsten K. Davis, Guide to Syllabus and Assignment Generative AI Policies and Advice (source publication; verify the current edition and citation before adding it to a syllabus or bibliography).
• Korin Munsterman, GenAI in Legal Education: A Practical Guide for Professors and Students (CALI eLangdell Press, 2026).
• Teaching materials shared by Professor Cathie Struve on attribution and disclosure for writing-intensive seminars.

Related Project materials:

• AI Use Policy Templates — Professor Struve’s attribution and disclosure model, with sample AI-use statements.
• Best Practices for AI in Legal Education — broader teaching design and student-use guidance.
• Prompt Guide & Scenarios — prompting patterns for law school tasks.

Acknowledgment

With thanks to the AI Law Lab alumni who contributed to the original guide:

• Meghana Bhimarao ‘25 — AI Law Lab & CTIC Fellow
• Lakshmi Prakash ‘25 — AI Law Lab & CTIC Fellow

Status

Maintained for the Penn Carey Law community. The sample syllabus language above is intended for direct adaptation by faculty.