Images generated by text prompts to current generative AI models, such as Midjourney or Stable Diffusion, are not copyrighted in the United States. According to the US Copyright Office (USCO), such prompts are analogous to a buyer giving directions to a commissioned artist. “They identify what the prompter wishes to depict, but the machine determines how those instructions are implemented in its output,” the USCO wrote in a new guidance document published in the Federal Register.
“When an AI technology receives only a human prompt and produces complex written, visual, or musical works in response, the ‘traditional elements of authorship’ are determined and executed by the technology — not the human user,” according to the office.
It stated that the level of human creativity involved in a work is an important factor in determining whether or not it will grant copyright protection. It implied that current AI models are incapable of producing copyrightable work. “Based on the Office’s understanding of the currently available generative AI technologies, users do not have ultimate creative control over how such systems interpret prompts and generate material,” the USCO stated. “It is well established in the Office’s opinion that copyright can protect only material that is the product of human creativity.” In one famous case, the office ruled that selfies taken by a monkey could not be copyrighted.
When it comes to works containing material generated by an AI, the USCO considers whether the model’s contributions to the work are the result of human intervention “If they are a “mechanical reproduction” (i.e., generated in response to text prompts) or if they represent the author’s “own mental conception.
“The USCO “will not register works produced by a machine or mere mechanical process that operates randomly or automatically without any creative input or intervention from a human author,” according to current rules.
The office, on the other hand, has left the door open to granting copyright protections for work with AI-generated elements. “The answer will depend on the circumstances, particularly how the AI tool operates and how it was used to create the final work,” the document stated. “This is a case-by-case investigation. If a work’s traditional authorship elements were produced by a machine, the work lacks human authorship and will not be registered by the Office.”
The USCO ruled last month that images generated by Midjourney and used in a graphic novel were not copyrightable. However, it stated that the text and layout of Kris Kashtanova’s Zarya of the Dawn could be protected by copyright. According to the office, there is too much “distance” between Kashtanova’s inputs and Midjourney’s output for the images to be copyrightable. According to Kashtanova’s lawyers, the office “applied the wrong legal standard” by focusing on the output rather than the input.
Meanwhile, in response to requests from Congress and the public, the USCO has launched an initiative to further investigate copyright law and policy issues related to AI. In April and May, it will host several panel discussions on the topics. The office intends to solicit public comments on a wide range of copyright issues relating to the use of AI later this year.