Ethics and academic integrity

Privacy, copyright, disclosure and USC policy. About 30 min, beginner level.

You'll be able to

The syllabus wins

There's no single rule for AI at a university. One instructor may encourage it, another may ban it, and the same course can treat two assignments differently. The university's academic integrity policy sets the frame; your course and assignment instructions set the specifics.

So before you use AI on graded work:

Using AI in a way your instructor hasn't allowed can be treated as academic misconduct, even if you'd have used it the same way elsewhere.

If you're an instructor, the same applies in reverse: say what's allowed for each assignment, specifically. Researchers have another layer, since many journals and funders publish their own AI rules, and journals commonly say an AI tool can't be listed as an author.

Privacy: what you put in

Anything you type or upload goes to the tool's provider. Depending on the tool, your account type and your settings, it may be stored, reviewed or used to improve models.

Never paste other people's private information into a tool that isn't approved for it:

For university work, use the university's tools. USC provides ChatGPT Edu, which keeps chats inside a university workspace; sign in with your USC account. Even there, check which kinds of data it's approved for before using anything sensitive.

Recording people raises the same issue: get consent from everyone before you record or transcribe a conversation.

Disclosure: credit AI like a source

When your course allows AI, disclose it the way you'd cite a source. A good disclosure answers four questions in a sentence or two:

“I used ChatGPT” is too vague to be useful. “I used ChatGPT to suggest an outline and to check grammar in my final draft; the argument and sources are mine, and I rejected one suggested section that misread the reading” tells your instructor exactly what happened.

Major citation styles, including APA, MLA and Chicago, have published guidance on citing generative AI. Follow the one your course uses. Cite the original sources you actually read, not the AI that pointed you to them.

You're accountable for everything you submit, including any error the AI introduced. The AI Fluency Framework calls this Diligence: being transparent about AI's role and taking responsibility for the result.

Copyright is unsettled

Copyright and AI is an area where the law is still developing, so be careful about confident claims either way.

What's reasonably clear: in the US, the Copyright Office has said that copyright protects works of human authorship. Material generated entirely by AI, without meaningful human creative contribution, may not be protected. Work where a person selects, arranges or substantially modifies AI output is more complicated and is judged case by case.

What's still open: several lawsuits are testing whether training AI models on copyrighted books, images and articles is lawful. The answers may differ by country and by case.

Practical habits while it's unsettled:

Fairness, bias and access

Language and image models learn from huge amounts of human-made data, and they reflect its patterns, including stereotypes. Ask for an image of a “scientist” or a story about a “nurse” and notice who appears. Bias can also be subtler: whose examples, dialects or perspectives are treated as normal.

This matters most when AI output affects people. Don't use AI to make judgments about individuals, like screening applicants or evaluating a classmate, without a person reviewing the decision. Check examples and images in your own work for who's represented.

Fairness includes access, too. Paid tools can give some students an advantage; university-provided tools help level that.

And AI has real accessibility benefits. Captions and transcripts, reading support, text-to-speech, and help drafting image descriptions can make learning easier for many people. These are legitimate uses. As with anything else, check the output: captions mishear technical terms, and generated descriptions can miss what matters in an image.

From a vague note to a real disclosure

A student used AI on a history essay. The syllabus says: “You may use AI for brainstorming and proofreading. Don't use it to write any part of the essay. Disclose any use.”

First attempt: “I used AI.”

This doesn't say which tool, for what, or how it was checked. The instructor can't tell whether the use was allowed.

Revised: “I used ChatGPT Edu to brainstorm three possible angles on the reading and to proofread my final draft for grammar. I chose the second angle and wrote the essay myself; I checked each grammar suggestion and rejected two that changed my meaning.”

Checking it against the policy:

The student also noticed that while brainstorming, they'd pasted in a classmate's discussion post to “get more ideas”. They'd had no permission to share it, so next time they'll summarize the idea in their own words and credit the classmate instead.

Try it

A two-sentence disclosure for a recent assignment that names the tool, what you used it for and how you checked it, with a note confirming it fits the course's AI policy, or the question you'll ask your instructor if it doesn't.