AI Plagiarism: Is Using AI Cheating? Where Universities and Style Guides Draw the Line

AI plagiarism is the everyday name for submitting text a chatbot produced as if you wrote it. Most universities do treat that as misconduct, but the name hides two separate problems. One is credit: presenting words or ideas that are not yours. The other is permission: using help the assignment did not allow, even if you admit it. This page quotes two university policies that define the line, sets plagiarism and unauthorized assistance side by side, shows what APA and MLA say about citing AI, and ends with a routine for using AI inside the rules.

Is AI plagiarism? What two university policies say

Two questions decide it: was the help allowed, and did you say so?

Princeton's list of academic integrity violations handles generative AI in its own entry, next to plagiarism. It says generative AI "is not a source" in the plagiarism sense "because its output is not created by a person," so where an instructor permits it, students "must disclose its use rather than cite or acknowledge the use." It then lists what counts as inappropriate use on submitted work: directly copying the output, representing output generated by or derived from AI as your own "(including rewording or paraphrasing output from generative AI and representing it as their own)," exceeding the instructor's limits, and failing to disclose use, including use for rewording, paraphrasing or editing.

Stanford's generative AI policy guidance, adopted October 15, 2026, sets a default for courses without a written AI policy. Students may use AI to practice and support learning, for example self-quizzing or asking the AI to check comprehension, and the guidance offers a rule of thumb: if a human tutor would be allowed for the task, that kind of AI use is likely allowed too. Surface-level editing of spelling, grammar and punctuation is also permitted. Students may not use AI "to create deliverables" such as essays, reports or presentations, and any use beyond the permitted list must be documented and disclosed.

Both policies leave the final say with the course: Stanford's guidance says instructors "may establish different and more specific expectations." The same prompt can be fine in one class and a violation in the next.

Plagiarism and unauthorized assistance are different problems

Plagiarism is a credit problem: you present someone else's words or ideas as yours. Unauthorized assistance is a permission problem: you used help the rules did not allow, whether or not you credited it. AI cases can be one, the other or both, which is why "I cited it" is not always a defense and "I wrote every word" is not always enough.

What APA and MLA say about citing AI

Style guides do not decide what is allowed, but they say how to give credit when AI is allowed. APA Style's 2025 guidance on citing generative AI says it is still not a good idea to submit AI-generated work as though you did it yourself, and that for students doing so would be academic misconduct. It says AI cannot be an author, because it cannot consent to or take responsibility for authorship, so the author element of the reference is the company that made the tool. For a shared chat, the template is the company, the date of the chat, the chat title in italics with a description such as [Generative AI chat], the tool or model name, and the chat's URL. When you used AI to edit or refine your own writing, APA says you should disclose that use and cite the tool generally rather than every chat.

The MLA's updated guidance on citing generative AI also advises against treating the tool as an author. Its template describes what was generated in the Title of Source element (often the prompt), names the tool as the container, the model in the Version element and the company as publisher, then gives the date the content was generated and a stable, shareable URL where the tool offers one. MLA adds that AI tools can invent sources, so cite the sources an AI links to, not the AI response.

For a step-by-step version, see how to cite a chatbot answer in your reference list.

A worked example: from chatbot answer to permitted use

Suppose a student asks a chatbot why a 1930s farm crisis spread and gets this answer (an example written for this page): "The crisis spread because drought, overplowing and falling crop prices reinforced one another across the region." Three ways to use it:

• Plagiarism: the student puts that sentence, or a reworded version of it, into the essay with no mention of the chatbot. Under Princeton's wording, both the copy and the rewording count as representing AI output as their own.
• Permitted, disclosed use: the course allows AI for brainstorming. The student uses the answer as a list of three causes to research, reads sources on each, writes the paragraph from those sources, cites the sources, and adds a note: "I used a chatbot to brainstorm possible causes; all claims and wording are from my own reading."
• Quoting AI as evidence: in an essay about how chatbots explain history, the student quotes the sentence in quotation marks and cites the chat in the course's style, following the APA or MLA template above.

How to use AI and stay within the rules

1. Find the written policy

   Check the syllabus, then the assignment sheet, then the school's default. Stanford's guidance says instructors are expected to state AI policies at least in the syllabus.

2. Match the use to the policy

   List what you plan to do: brainstorm, explain a concept, quiz yourself, check grammar, draft. Keep only the uses the policy allows, and ask about any it does not mention.

3. Keep a log as you go

   Save the prompts and shared chat links for anything you use. A one-line note per session is enough to write an accurate disclosure later.

4. Write the submitted text yourself

   Turn any AI help into your own reading and your own sentences. If you catch yourself rewording a chatbot paragraph, stop: that is the case Princeton names.

5. Verify every fact and source

   Open each source an AI mentions and confirm it exists and says what was claimed. MLA warns that AI tools can invent sources or summarize them wrongly.

6. Disclose and cite

   Disclose permitted use in the form your course asks for, and cite any AI output you quote. A guide to writing an AI disclosure has short templates.

Pitfalls

• Thinking rewording makes it yours. Princeton's policy names rewording and paraphrasing AI output as misrepresentation.
• Citing sources the AI gave you without opening them. Invented or mismatched references are a separate integrity problem, and Stanford's guidance says they may prompt a review.
• Relying on a detector to judge your own work. Detectors estimate; policies decide. What an AI detection score means explains why a low score proves nothing about permission.
• Losing your drafts. If your work is questioned, saved versions and notes are your best evidence. How to show you wrote your own work lists what to keep.

Expert tips

• Use the tutor test. Stanford's rule of thumb works well beyond Stanford: if you would be allowed to ask a human tutor for this help, ask whether the AI version is allowed too.
• Check each tool's AI features. Grammar checkers and word processors now include generative buttons; whether Grammarly counts as AI shows how one tool splits.

AI plagiarism is real, but it is two problems with one name. Ask whether the help was allowed, then whether you said you used it. Read your course policy, keep AI to the uses it permits, write and verify the submitted text yourself, and disclose and cite in the style your course uses. Done that way, AI is help, not cheating. More on spotting and editing AI writing is on the AI writing hub.