A student hands in a poster. The layout is theirs, the typography is theirs, and the background image came from an image generator. Is that fine? Is it cheating? Should they have told you? And if your course doesn't say anything about it, whose fault is the confusion?
Most of the friction around AI in design classes comes from that last question. Students aren't sure what's allowed, teachers aren't sure what to ask for, and everyone makes up the rules as they go. This article covers the ethical ground worth teaching, and then gets practical: how to write a short AI policy for your course so nobody has to guess.
Why this matters more in design
Design students are going to make a lot of the images, interfaces and messages other people see every day. The habits they build now around disclosure, bias and privacy will follow them into studios and product teams. Teaching them to use AI tools is the easy part. Teaching them to use AI tools with judgment is the part that sticks.
Five principles worth teaching
1. Transparency
Students should say when they used AI. Disclosure isn't a confession. It's the same professional habit as crediting a photographer or naming the typeface. Clients and employers increasingly want to know how work was made, and a designer who can explain their process clearly is more trustworthy than one who can't.
2. Attribution
Disclosure is more useful when it's specific. "I used AI" says very little. "I generated three background textures with an image model, chose one, and retouched it; the layout, type and copy are mine" tells a reviewer exactly what to evaluate.
3. Privacy
Anything you type into a cloud AI service leaves your computer and is processed on someone else's servers, under their terms. That's fine for a lot of work. It isn't fine for student names, photos, grades, contact details or anything else that identifies a person.
Two practical options: anonymize before you paste (replace names with "Student A", remove identifying details), or use a local model that runs entirely on your own machine, where nothing is sent anywhere. For teachers who handle a lot of student information, local tools are worth learning for this reason alone.
4. Fair access
Not every student can pay for a subscription, and not every student has a powerful computer. If an assignment quietly assumes a paid tool, it grades students partly on their budget. Design assignments so they can be completed with free options, and say which ones are acceptable. Free tiers of cloud tools and free local apps both count.
5. Bias
Models learn from existing data, and existing data carries existing biases. Image generators, for example, have been widely documented to default to narrow representations when prompts don't specify otherwise: "a CEO" or "a nurse" can come back looking very similar every time, and people of some backgrounds can be rendered less accurately or in stereotyped roles.
This is teachable, and design students are well placed to notice it. Some useful habits:
- Be deliberate in prompts. Specify age, background, context and setting instead of accepting the default.
- Review every output critically. Ask who is shown, how, and who is missing.
- Compare tools. Different models fail in different ways.
- Keep notes. When a tool produces something biased, write it down and bring it to class. Real examples teach better than warnings.
AI and assessment
AI can save teachers real time around assessment. It's good at drafting quiz questions, turning rough notes into rubric language, suggesting general feedback phrasing, and spotting common patterns across many submissions when you describe them.
It should not replace a person's judgment of creative work. A model can't see the sketches that didn't make it, the conversation where a student changed direction, or the risk they took that didn't quite land. Grades are decisions about people, and those belong to the teacher. And student work should only go into an AI tool with the same privacy care as anything else: anonymized, or processed locally.
A simple routine for teachers
Before using a tool with students: check whether it sends data off your computer, skim its privacy terms, and decide what information you will not put into it. Tell students which tools you use and why.
During use: encourage students to question outputs rather than accept them, give the process as much weight as the result, and respect students who prefer not to use AI for a given project.
After: review what came out for bias and errors, note what worked and what didn't, and share it with colleagues. Your policy should get better each term.
Questions to ask about AI-assisted work
When a project involves AI, these five questions cover most of what matters:
- Did the student disclose how AI was used?
- Can they explain their full process, including why they kept or rejected outputs?
- Does the AI-generated material carry bias the student didn't catch?
- Where did the student's own judgment show up: selection, editing, composition, concept?
- Does the project show that they learned what the assignment was meant to teach?
If the answers are good, the tool is mostly beside the point.
Writing your course AI policy
A policy doesn't need to be long. It needs to be clear enough that a student can answer "am I allowed to do this?" without emailing you. Before writing it, spend ten minutes on your context:
- What kinds of projects do students make in this course?
- Which AI tools are relevant (text, image, code, audio)?
- What are the risks here: overreliance, plagiarism, bias, privacy?
- Where could AI genuinely help students learn?
Then cover five things: what's allowed, what's not, how to disclose, how AI-assisted work is assessed, and what free options students have.
It's fine to draft the policy with an AI tool. It's a good demonstration of the process you're asking of students. But edit it heavily. Generated policies tend to be generic and overly restrictive. Check the tone (clear, not threatening), the specifics (does it fit your actual assignments?) and fairness (does it work for a student with no paid tools?).
A short template
Adapt this to your course. Replace the bracketed parts, and cut anything that doesn't apply.
AI USE POLICY: [Course name]
Why this policy exists
AI tools are part of professional design practice. In this course you
may use them to support your learning, as long as you stay the author
of your decisions and are open about how you worked.
You may use AI to:
- Research, brainstorm and explore directions early in a project
- Generate [assets such as textures, placeholder copy, mockup images]
that you then select, edit and integrate
- Get feedback on drafts of your writing
- [Add course-specific uses]
You may not use AI to:
- Produce [the core skill the assignment assesses, e.g. the final
layout, the logo concept, the written analysis]
- Submit generated work as if you made it yourself
- Upload other people's personal information or classmates' work
to any AI service without their permission
How to disclose
Add a short "AI use" note to every submission that used AI. Include:
- Which tool(s) you used
- What you used them for
- What you changed, selected or rejected
If you did not use AI, write "No AI used."
How your work will be assessed
Your grade is based on your process and decisions as well as the result.
Be ready to explain any part of your project. Undisclosed AI use is
treated as an academic integrity issue; disclosed use is not penalized
when it follows this policy.
Access
Every assignment can be completed with free tools. Acceptable free
options include [list free cloud tiers and/or local tools]. If you are
unsure whether a tool is allowed, ask before using it.
This policy may be updated during the term. Changes will be announced.
A disclosure example
Policies make more sense with an example. Here's what a good disclosure note looks like for the poster at the top of this article:
AI use: I generated about 20 background images with an image generator
using prompts about "abstract rain on glass". I chose one, cropped it
and adjusted color and contrast in Photoshop to match my palette.
The concept, layout, typography and all text are my own. I tried using
AI for headline ideas but rejected all of them.
Short, specific, and easy to grade.
One thing to watch for
Terms of service for AI tools change, and they differ a lot between providers, including on whether generated images can be used commercially and whether your inputs are used for training. Don't write specific legal claims about a tool into your policy. Point students to the current terms, and review them yourself at the start of each term.
Try this
Before your next course starts, write the one-page policy above and add a disclosure note to one of your own recent projects. If you can't describe your own AI use clearly in a paragraph, your students won't be able to either, and that's the place to start.
Further reading
- UNESCO, Guidance for Generative AI in Education and Research (2023)
- Neil Selwyn, Should Robots Replace Teachers? AI and the Future of Education (Polity, 2019)