A comic book creator, Kristina Kashtanova, submitted a copyright application to the U.S. Copyright Office for her comic book, which included AI-generated images.  She used an artificial intelligence platform, Midjourney, to create the images by providing Midjourney with descriptive text of what the illustrations should look like.

The Copyright Office initially allowed the registration and protection of the full comic book, including the AI-generated images. In a letter issued this week, the Office cancelled part of the registration due its discovery that the registration was based on "inaccurate and incomplete information." Kashtanova did not explain that the illustrations were AI-generated when she submitted the application. 

Kashtanova argued that she authored every part of the work since she directed the platform on how to create the illustration. She claimed the illustrations were merely "AI-assisted." However, the Office concluded that using an AI-generated platform to create the illustrations was "not the same as that of a human artist, writer or photographer," which is how copyright rights arise. The Office doubled down that Kashtanova did not create the illustrations; the artificial intelligence program did.

The Office then replaced the original registration with a more limited one. The new registration covers the text in the comic book and the "selection and arrangement" of the images and text in the comic book. The images in the comic book were expressly excluded.  The registration states that the exclusion applies to "artwork generated by artificial intelligence." 

The selection and arrangement of the images is (and has always been) a valid authorship contribution to a work. It appears that with this case, the Copyright Office has clarified its approach to AI-generated content and the parameters of copyright protection - noting that while the arrangement of the AI-generated content is protectable, the content itself is not.