AI art generators like Midjourney have made headlines, even winning awards. However, the US Copyright Office (USCO) recently issued guidance clarifying that AI-generated content is not protected under US copyright law. This report addresses how copyright applies to AI-generated video, images, and text, concluding that existing copyright rules largely suffice.
The core issue is authorship. The Supreme Court’s 1989 ruling in Community for Creative Non-Violence v. Reid established that copyright belongs to the “person” who translates an idea into tangible form. The USCO argues that the unpredictable nature of AI output means users lack sufficient control over the generative process, even with complex prompts. The report states, “the issue is the degree of human control, rather than the predictability of the outcome.” Essentially, accepting the AI’s interpretation doesn’t constitute authorship.
This doesn’t mean all AI-involved creations are unprotected. The 2024 film “Here,” featuring AI-de-aged actors, is copyrighted because the AI served as a tool, not the creator. Similarly, the USCO explains that films using AI for special effects or background art are copyrightable, even if those elements alone are not.
Artists modifying their existing work with AI also have some protection. While the AI-generated portions aren’t copyrightable, the overall piece, reflecting “perceptible human expression,” is.
This isn’t a new dilemma. Since the 1960s, the USCO has grappled with authorship in the digital age, questioning whether content is human-authored or “written” by computers. Then-Register of Copyrights Abraham Kaminstein highlighted the key question: is the computer an assisting instrument or the conceiver and executor of the work?
The USCO acknowledges that its guidance may evolve as AI technology develops. The report suggests that future AI systems might allow users enough control to make the system’s contribution “rote or mechanical,” but current AI prompts don’t meet that threshold.
