Robots Don’t Make Art

One of the central tensions of the past 30 years when it comes to art and culture has been the need to balance the unbridled creativity afforded us by the internet with the ability of our legal institutions to make sense of who owns it. Since the artificial intelligence revolution began in earnest last summer with the launch of OpenAI’s Dall-E 2, the guiding wisdom has been that these early months of the generative AI age would play out something like the Napster era of the social web: A lot of copyright infringement would happen, eventually companies would get a handle on what is and isn’t fair use, and then finally we’d figure out passable business models for what’s allowed.

Then a few weeks ago, the U.S. Copyright Office ruled that a series of images created by the generative-AI tool Midjourney could not be protected by copyright. The ruling centered on a comic book created by Kris Kashtanova, a machine learning educator from New York, called “Zarya of the Dawn.” According to the Copyright Office, Kashtanova, who wrote the text, is the author of the comic, but not the illustrator, even though they wrote the AI prompts that produced the images within it.