The AI Policy Patchwork
We analyzed 768 college syllabi. Here's what professors are actually telling students about AI right now.
There is still no shared rulebook.
That's the reality students walk into every semester. One class treats AI like a source you have to cite. The next treats it like plagiarism by default. A third mentions it just enough to make you nervous without telling you what's actually allowed.
We pulled 768 processed syllabi from the DormWay corpus — uploaded by 248 students across 208 named universities between October 2025 and April 2026 — and looked at the actual policy text. Not press releases. Not university statements. The language professors put on page one of the syllabus.
Here's what we found.
The big number
46% of syllabi explicitly mention AI. The other 54% still say nothing at all.
Among the 353 that do mention AI, the breakdown looks like this:
The center of gravity isn't enthusiasm. It's control — either a default ban, or tightly bounded use with disclosure. Open-ended encouragement is essentially nonexistent.
The middle ground is conditional
The most common posture isn't "use it freely." Almost half the AI-mentioning syllabi are restrictive. The main alternative isn't open-ended permission either — it's conditional use: if you use AI, say so clearly and be ready to account for it.
A typical example:
"If you choose to utilize AI programs to generate content, you must clearly acknowledge the use of AI generated material."
— SSS 1001, Louisiana State University
That kind of rule is more useful than vague integrity boilerplate. Students know there's a line, and they know disclosure is part of it. But it's still highly course-specific. Disclosure usually comes bundled with warnings that AI should be limited to brainstorming, revision support, or narrowly defined assignment stages.
The restrictive side
Nearly half of AI-mentioning syllabi either ban AI outright or treat it as disallowed unless the instructor specifically carves out an exception.
Some are blunt:
"You may not submit any coursework that was generated by AI or assisted by AI."
— BCH4033L, University of North Florida
Others spell out a broader default ban:
"I specifically forbid the use of ChatGPT or any other generative artificial intelligence (AI) tools at all stages of the work process, including brainstorming."
— CHE 1121 0B1, University of Texas at San Antonio
These aren't edge cases. They're a major share of the current rule set.
Ambiguity hasn't disappeared
Some syllabi mention AI without ever resolving the student's actual question: what, exactly, am I allowed to do?
The clearest current example is almost an admission of uncertainty:
"There's too much to say to put a complete and coherent policy on AI here."
— WRTG 10600, Ithaca College
That's honest. It's also exactly the kind of sentence that forces students to guess where brainstorming ends and misconduct begins.
The same-campus problem
The patchwork is most obvious inside universities, not between them.
26 of the 38 universities where we have at least three AI-mentioning syllabi show more than one stance. Same campus, same semester, completely different rules.
Michigan is a clean example:
That's the real patchwork. Not one campus policy, but a stack of course-level negotiations.
What this means
The most striking finding isn't that professors are ignoring AI. Many aren't. It's that they're converging on different answers to the same question:
- cite it
- use it only for limited steps
- don't use it unless I say so
- don't use it at all
For a student juggling four or five classes, that isn't a usable norm. It's a compliance exercise that resets every time they open a new syllabus — and it's exactly the kind of cognitive overhead that makes the difference between a B and a missed assignment.
We built syllabus parsing into DormWay because students shouldn't have to be policy detectives. When you upload your syllabus, Ace surfaces the AI policy alongside the late policy, the grading breakdown, and the exam dates. Same place, every class. So when the rule for class A is "cite it" and the rule for class B is "don't touch it," you don't have to remember which is which.
The patchwork isn't going away. The least we can do is make it legible.
Sources & Methodology
- DormWay Syllabus Processing
We pulled 768 processed syllabus documents from production. Starting from `braingains_documents`, we filtered to `document_type='syllabus'` and successful processing states (`completed` or `indexed`), then joined each document to its parsed output in `service_data` using the shared workflow ID. Where a document had multiple processed rows, we kept the latest parse. The corpus covers uploads created between October 23, 2025 and April 27, 2026, from 248 users across 208 named universities. AI-policy detection ran on extracted syllabus policy text only, using the `policies` payload from each parse — we deliberately excluded any model-generated advice fields so that recommendations from our own pipeline wouldn't count as syllabus language. A syllabus was marked AI-mentioning if its policy text referenced ChatGPT, generative AI, artificial intelligence, AI tools, AI assistants, Copilot, or large language models. From there we grouped into four buckets: disclosure/citation required, restrictive/default-ban, ambiguous/reference-only, and cleanly open-ended encouragement. Restrictive counts include instructor-permission-only language, since the default rule for the student is still "no" unless the syllabus explicitly opens a door.
About Riley
Founder & CEO
Riley made DormWay to solve his own problems, and in the process is solving all college students'. A fourth-year at U-M with 100K+ followers across platforms, Riley taught himself to code while building DormWay.