How to Cite AI-Generated Images (Midjourney, DALL-E, ChatGPT, Firefly)

With text, the question is how to cite something you quoted. With an image, the generated thing is the exhibit itself, sitting in your paper, slides or poster. So the credit goes where the image goes: in the caption. A reference-list entry is secondary, and some styles skip it.

This page covers images you generated yourself. An AI-generated image you found online, made by someone else, is cited like any other online image. For artworks in general, see how to cite art.

A note on the examples. An image you generated is yours, so we cannot show a "real" one. Each section gives the format as a template with [bracketed] slots, followed by the published example from the style authority, or from a university guide where the authority's text is paywalled.

Which company do I name?

Image generators are easy to misattribute because they are embedded in other products.

The rule is the same as for chatbots: cite the product you used and the company responsible for it, not the model you believe is underneath.

Is there a link?

An image made inside a chat can be shared through the conversation's share link, and MLA's DALL-E example uses exactly that. If there is no public link, give the tool's general address and keep the original file with its metadata and your prompt.

MLA 9

MLA's August 2025 guidance gives a caption format: the prompt as the title, then tool, version, company, date and location.

MLA's own example:

MLA says you can use the same information for a works-cited entry if you choose to create one instead of putting the full citation in the caption.

APA 7

APA figures have a number, a title and a note. APA's 2025 guidance says that when you use AI to create tables, figures or other visuals, you indicate that in the table or figure note, and cite the tool in general. A layout to adapt:

This follows APA's general template for AI tools (company as author, tool in italics, description in brackets, URL). APA's examples use [Large language model] for chat tools and say to use the description that best fits; "AI image generator" is our wording. APA's own illustration of describing an image prompt in the text begins: I provided the following prompt to Gemini (Google, 2025). If the image was made in a chat you can share, APA's specific-chat format works too. Some instructors want only the figure note and no reference, so ask.

Chicago

University guides to the 18th edition say AI-generated images are credited in the caption (McMaster cites section 14.111; Monash cites 14.112 for chatbots and section 3.38 for AI-generated images). The credit follows the pattern for generated text:

McMaster University Library's example:

As with text, library guides say there is normally no bibliography entry.

Harvard (Cite Them Right)

Robert Gordon University Library reproduces the Cite Them Right template for artwork an AI made at your request: name of AI, year, description in italics, the medium in square brackets, who it was created for, and the day and month.

The example RGU gives is: Open AI Dall-E (2023) Infinite pile of books [Digital art]. Created for Malcolm Reed, 14 June. In text: ([Company and tool], [year]).

IEEE, Vancouver and AMA

None has a reference format for generated images. What they have are disclosure rules. IEEE's policy names figures and images among the AI-generated content that must be disclosed in the acknowledgments, with the system identified. For medical journals, ICMJE requires AI use to be described in the manuscript, and says authors should be able to assert there is no plagiarism in text and images produced by AI; individual journals often go further for images. State in the caption that the image is AI-generated, name the tool and company, and follow the journal's instructions.

What journals and universities allow

Images are where publisher policies are strictest, because a generated image can be mistaken for evidence. Elsevier's journal policy lets authors use AI tools to support certain explanatory images, such as flow charts, timelines and schematic illustrations, requires the tool, version and use to be disclosed in each caption, and says AI tools must not be used to create or alter images that represent primary observed or experimental data. The Science journals' policy announcement on generative AI says AI-generated images and other multimedia are not permitted without explicit permission from the editors. Nature Portfolio's policy sorts AI use by risk level and lists creating photorealistic images (deepfakes) as not permitted. Policies like these are rewritten often, so read the current page for your target journal, not a summary of it.

For coursework, a reasonable baseline is a caption that says the image is AI-generated and gives the tool and prompt, and no use of generated images where the assignment asks for your own visual work. A generated image of a real person, place or event should be labelled unmistakably, and is often better left out.

A disclosure sentence

A pattern to adapt, with your own details in the brackets:

Only say what is true of your own work.

Sources

All pages were checked on 21 September 2026.

• MLA Style Center, How do I cite generative AI in MLA style? (Updated and Revised) (13 August 2025)
• American Psychological Association, Citing generative AI in APA Style: Part 1, reference formats (9 September 2025)
• The Chicago Manual of Style, 18th ed., is paywalled; we followed McMaster University Library, Columbia College Library and Monash University Library
• Cite Them Right templates as reproduced by Robert Gordon University Library
• IEEE Author Center, Submission and peer review policies: AI-generated content
• ICMJE, Recommendations: artificial intelligence, use by authors
• Elsevier, Generative AI policies for journals
• Science, Change to policy on the use of generative AI and large language models
• Nature Portfolio, Artificial intelligence (AI) editorial policy
• U.S. Copyright Office, Copyright and Artificial Intelligence, Part 2: Copyrightability (29 January 2025)