The US Copyright Office (USCO) is seeking public input on generative AI and ownership of its outputs. As this technology captures the legal community's attention, the USCO has begun gathering comments on critical issues related to AI and copyright.
Key topics include:
- The legality of companies training AI models on copyrighted content.
- Copyright eligibility of AI-generated materials and liability for potential infringements.
- Management of machine-produced outputs that resemble works created by human artists.
The USCO acknowledges the growing debate surrounding the implications of generative AI for creative industries, stating, “The widespread use of generative AI systems by millions of Americans has raised substantial questions for the copyright system.”
A primary concern is determining the level of human authorship required to register a copyright for AI-generated content. The agency notes an increase in attempts to designate AI as an author or co-author. “The essential query is whether the work is fundamentally a product of human creativity with AI as a tool, or if the core elements of authorship were autonomously conceived and executed by the machine,” they explain.
Recent legal cases provide some insight into these boundaries. In February, the USCO ruled that the human-created text and layout of a partially AI-generated graphic novel were eligible for copyright, while the AI-generated images were not. Conversely, a federal judge denied a copyright application for entirely AI-generated art, which involved no human input beyond the initial prompt, stating, “Copyright has never extended so far as to safeguard works produced by new technologies operating without any guiding human touch.”
Furthermore, the USCO is interested in addressing the rising number of copyright infringement claims against AI companies. Notably, Sarah Silverman is among those suing OpenAI and Meta for allegedly using her copyrighted work, including her memoir The Bedwetter, to train their AI systems. OpenAI is also facing a class-action lawsuit related to its use of web-scraped data for training its chatbot.