The plethora of AI-generated images floating on the internet has many photographers perplexed about the state of the medium. While some continue to support AI-generated images as the “evolution” of photography, there are many who resist it for ethical reasons or find these pictures to be “soulless.” However, as the world continues to stand divided, there has been a great development in this sphere. For the first time, an AI-generated image has managed to get a copyright protection in the United States.
The lead image is a screenshot from Invoke’s video.
In a report by Artnet, this landmark decision was officially taken on January 30, 2025. The AI-generated photograph, titled A Single Piece of American Cheese, became the first AI-generated image to get protection under the law. How did it happen? After many grueling back and forth between Kent Keirsey and the U.S. Copyright Office.
Keirsey is the founder and CEO of Invoke, a generative AI platform for professional studios to create visual media. According to reports, Keirsey used Invoke’s inpainting feature to create an AI-generated image, which allowed him to make multiple edits to the original AI work. As a result, the image turned out to be somewhat different from what the AI originally created. Then, in 2024, the company filed an application to register the piece. However, these rights were for the selection, coordination, and arrangement of components of the image, but not the individual AI-generated segments.
As one would expect, the application was denied for the lack of “human authorship necessary to support a copyright claim.” However, after seeing the video created by Keirsey showing the step-by-step process of the changes or edits he made, the Copyright Office changed its mind. In the end, the copyright was given on the grounds of “a sufficient amount of human original authorship” in the initially proposed draft. This means that the office realized that the edits made by Keirsey are worth the protection, even if the material is AI-generation.
This is pretty huge as it means anyone who has made changes to the AI-generated artwork can still get protection for their human authorship and inputs. However, previous attempts to get the same protection failed on the basis that the creations were text prompts and did not have sufficient human creativity to get a copyright patent. This test was also done not only to help AI artists get protection but also to prove that Invoke’s “Provenance Records,” a tool that helps track edits, is useful to people applying for a similar patent and, thus, helping their authorship.
However, we tend to forget that for any AI-generated images to exist, human input is needed on what that thing looks like. For instance, if you want to create a photograph of a landscape with the sun setting down, then the tool will have to swift through massive data to give you examples. As a result, the work, no matter what the judge says, is created from existing images on the World Wide Web. And so, when you do make edits, that human input is simply offering a variation to someone’s already existing work. So, won’t that, to some extent, be considered plagiarism?
This also then brings to question what happens to artworks (I don’t like to call them photographs because they aren’t that) that are created by pooling in works of other artists? For instance, an artwork that combines Ansel Adam and Edward Weston‘s style? And if the photograph is then further edited with my input, would that still be copyrightable?
This is exactly why photographers have to be really careful of where they post their work or who can get their hands on it. However, if anyone comes up to you to debate how images were created without a camera, then please remind them that those photographs were not a result of someone else’s input but your own. And so, that work is rightly yours. The same can’t be said about AI content. This is just the start of making artificial intelligence more mainstream. However, its repercussions will be felt later on.