Need research help? Citation help? Want to consult a librarian? Make an appointment using the button below!
When using images in your work, it's essential to adhere to copyright laws and uphold academic integrity.
For A.I.-generated visual content, citation practices may vary among style guides and depend on specific requirements set by A.I. tool developers. While MLA and Chicago have provided guidance on citing A.I.-generated content, more comprehensive instructions from other style guide providers are anticipated soon. Until then, we recommend referring to the general rules of each style guide for citing images.
Please be aware, that currently several lawsuits for copyright infringement are pending against the companies behind some of the A.I. image creation tools on the market. Depending on the respective outcome, that might have an effect on advise on how to use A.I. created visual art.
Current Status
In the commentary section of their blogpost from April 7, 2023, on how to cite text generated by ChatGPT and similar tools, APA has replied to a question about how to cite visual content generated by or with the assistance of an A.I. tool.
Interim Guidelines
If you have used image- or video-generating tools, you can follow the guidelines in the post about A.I. generated text
A short summary:
Whether to include the images or videos as figures, appendices, or supplemental materials would be a decision to discuss with your instructor or editor. There may also be copyright issues to consider (see, e.g., these articles about a recent U.S. Copyright Office decision).
Additional tip from UVic Libraries (subject to change)
Depending on the specific image generation tool you use, and its specific featueres, we advise to study sections 10.12–10.14 of the APA Publication Manual, 7th edition, specifically the guidelines on how to cite an clip art or stock images that requires an attribution. Clip art/stock images might be the visual content form featured in the APA Style guide that is closest to A.I. generated images, and said A.I.-based images do need referencing as well.
Current status
Chicago Manual of Style (CMOS) Online has released guidelines on how to reference visual A.I. generated art. According to CMOS Online's guidelines, it is crucial to credit the source when including an A.I.-created image in your project (see also CMOS 3.29 – CMOS 3.37). CMOS Online is referring to an article by OpenAI on how to cite an image created using their tool DALL·E 2, which is suggesting wording like "This image was created with the assistance of DALL·E 2" or "This image was generated with the assistance of AI".
Example
The credit for that image in Chicago style might read as follows (with the prompt used to generate the image in quotation marks):
Chicago Style has based this guideline on how OpenAI has suggested work created with their application DALL-E 2 should be credited. Please keep in mind that other companies providing image A.I. generation tools may have different preferences on how visual art generated with the help of their technology should be cited.
Information coming soon.
Current Status
MLA has released guidelines on how to cite content generated by A.I.
General Advice
In accordance with those guidelines it is crucial to caption an A.I.-created image as outlined in Section 1.7 of the MLA Handbook when incorporating it in your project.
Template
MLA recommends providing a description based on the prompt, followed by the A.I. tool used, its version, its provider, the creation date, and – depending on the availability of a publishing feature for the images generated with the chosen tool – the URL of the published image. If publishing such an image, making it accessible with an individual URL, is not possible with the tool of your choice, include the general URL to that tool instead.
Example
The full credit for this image according to MLA style might read as follows:
Note: The same elements (prompt, A.I. tool used, version, creation date, individual image or general tool URL) can be used for a works-cited-list entry, if you choose that over including the full citation in the caption (see MLA Handbook, sec. 1.7).
General: Retrievability of images
In accordance with MLA's approach in their official guidelines for citing A.I.-generated visual content, we explain here how to publish such content, if possible, and make it retrievable.
How-to make A.I. generated images retrievable
As of April 4, 2023, some of the A.I. tools for image generation that are available to the wider public offer features to publish images generated with them, providing a platform and individual URLs to retrieve them. The image generation applications that allow for publication of content created with them are:
Tipp: Archiving published images
Since currently there is no certainty about the long-term preservation of any A.I.-generated content, you should consider preserving a snapshot of your published images by feeding it them into a web archiving service such as the Internet Archive's Wayback Machine.
Example: https://web.archive.org/web/20230404193313/https://labs.openai.com/s/7DTSjEDro0rRUG1y6jWLvTZM
Pending lawsuits (as of August 15, 2023)
More detailed information coming soon.
DALLE-E 2 (OpenAI)
DALL-E 2 is an A.I. application for image generation provided by OpenAI. It was released to the wider public in September 2022. It offers a free and a paid-for plan (on a token-basis) and lets you create and edit images of various kinds (from photos to paintings to rendered images) using a text-based prompt.
In order to use and reference images created with the help of DALLE-2, it is possible to publish these images and create a unique URL for them. The provided URL after publishing such an image does not link directly to the image itself, but to a landing page, that contains the visual, but also additional information on who the human creator is as well as the text prompt, that has been used to create the image.
Example: https://labs.openai.com/s/jrbn6Dl8aRTDvS4Kobny33n2
When using such an image, for example by embedding it on a website or including it in a manuscript, this prompt can be used as the alt text to meet accesssibility standards, as it ususally describes what can be seen in the image.
If you are the human creator of such an A.I. image you might want to save a copy of the image file and maybe even preserve a snapshot of the published image by dropping it in a web-archiving service like the Wayback Machine, since there is no certainty about the long-time preservation of such images or the landing pages created for them.
DreamUp (by DeviantArt Inc.)
DreamUp is a prompt-based image-generating tool that is offering a free as well as several paid-for plans. According to DeviantArt's terms of service,
In order to use and reference images generated with the help of DreamUp, it is possible to publish these images which creates a unique URL for them. The provided URL does not link directly to the published image, but to a landing page that contains the visual, but also additional information on the human creator as well as the text prompt that has been used to generate the image.
Example
https://www.deviantart.com/malschewski/art/DreamUp-Creation-954791163
Midjourney (by Midjourney)
Midjourney is a prompt-based image-generating tool. It is only accessible from within Discord, and there are free and paid-for plans.
The usage rights for your own creations with Midjourney depend on the plan you are using. According to the Midjourney terms of service (as of February 10, 2023) a user does not own the assets they create if they are not a paying customer. Instead, Midjourney grants a license to the content under the Creative Commons Noncommercial 4.0 Attribution International License.
Stable Diffusion (by Stability AI)
Stable Diffusion is an A.I. model for generating images provided by Stability AI. Stable Diffusion has been released under an open source license.
Pending lawsuits (as of March 22, 2023)
How should I credit DALL·E in my work? (OpenAI, 2023)
DALL-E: Content policy (OpenAI, September 19, 2022)
Midjourney: Terms of service (Midjourney, February 10, 2023)
AI Art: Machine Visions and Warped Dreams by 
Ask us