Introduction
The U.S. Copyright Office has released part two of its Report on Copyright and Artificial Intelligence. The report, which is being published in three parts, addresses the legal and policy issues related to artificial intelligence (AI) and copyright. Part one of the report, which was published on July 31, 2024, focused on digital replicas (i.e., deepfakes). Part two addresses the copyrightability of works created using generative AI, focusing on the necessary human contribution for such works to qualify for copyright protection. Part three of the report, which is anticipated later this year, will address the training of AI models on copyrighted works, licensing considerations and allocation of any liability.
The report is a culmination of extensive research and public engagement. Following an August 2023 notice of inquiry, the Copyright Office received over 10,000 comments from a diverse range of stakeholders, including authors, composers, performers, artists, publishers, producers, lawyers, academics, technology companies and public interest organizations. This input, along with additional research and information from other agencies, forms the basis of the report’s discussion and recommendations.
Copyrightability and the human authorship requirement
Human authorship requirement
The report reaffirms the foundational principle that copyright protection in the United States requires human authorship. This principle is rooted in the Copyright Clause of the U.S. Constitution and has been consistently upheld by courts, including as recently as 2023. See Thaler v. Perlmutter, 687 F. Supp. 3d 140, 146 (D.D.C. 2023) (holding that the “key” to copyright protection is “human involvement in, and ultimate creative control over, the work at issue”). The report emphasizes that purely AI-generated material, without sufficient human control over the expressive elements, is not eligible for copyright protection.
Assistive uses of AI
The report distinguishes between using AI as a tool to assist human creativity and using AI as a stand-in for human creativity. It concludes that assistive uses of AI, where the technology enhances human expression, do not affect the availability of copyright protection for the resulting work. For example, if a motion picture company uses an AI tool for tasks like color correction, detail sharpening, or de-blurring in visual effects, that use is considered assistive and does not undermine the copyrightability of the final work.
Prompts and human contribution
The report provides an in-depth analysis of the role of prompts in generating AI outputs. It concludes that, given current technology, prompts alone do not provide sufficient human control to make users the authors of the AI-generated output. Prompts are seen as instructions that convey unprotectable ideas rather than controlling the expressive elements of the output. The report notes that while highly detailed prompts could contain the user's desired expressive elements, they do not control how the AI system processes them in generating the output. The report also addresses the possibility that in the future a different conclusion may be called for should further advances in technology provide users with sufficient control over the expressive elements of an AI-generated output.
Expressive inputs
The report acknowledges that human-authored inputs, such as text, images, audio or video, that are reflected in the AI-generated output contribute more than just an intellectual conception. For instance, the report concludes that if a human author inputs their own copyrightable work into an AI system and the output retains the expressive elements of the original work, the human author can claim copyright in those elements. This type of contribution is seen as analogous to creating a derivative work, where copyright protection extends to the material contributed by the human author.
Modifying or arranging AI-generated content
The report highlights that human authorship can be added to the final product by selecting, coordinating and arranging AI-generated material in a creative way. The report concludes that such actions would provide copyright protection for the output as a whole, although not for the AI-generated material alone. The report cites examples of comic books that combine human-authored text with AI-generated images, where the selection and arrangement of these elements are protectable as a compilation. The report also cites the use of tools offered by popular AI platforms that allow users to select, edit and adapt AI-generated content. The use of these tools, unlike the use of prompts alone, can enable the user to control the selection and placement of individual creative elements. Though the report cautions that whether such modifications rise to the minimum standard of originality, and are therefore eligible for copyright, will depend on a case-by-case determination.
International approaches
The report also examines how other countries are addressing the copyrightability of AI-generated material. It finds that most countries agree on the necessity of human authorship for copyright protection. For example, South Korea and the majority of European Union member states have emphasized that only works with significant or creative human input are eligible for copyright. Additionally, some countries, like Japan, consider factors including subsequent human additions on a case-by-case basis to determine eligibility for copyright. The report also notes that while some jurisdictions, like the United Kingdom, Hong Kong, India and New Zealand, have statutes that protect computer-generated works, there is still uncertainty about how these laws will apply to AI-generated content.
Arguments for legal change
The report evaluates various arguments for and against extending copyright protection to AI-generated material. Proponents argue that such protection would encourage the creation of more works, furthering cultural and knowledge progress. However, most commenters and the Copyright Office itself maintain that the Copyright Clause requires human authorship and encourages progress with legal and economic incentives. AI systems, as inanimate objects, do not need such legal and economic incentives to create more works. The report also notes that, as evidenced by the rapid development of AI systems, the developers already have sufficient incentives under existing law. The report also expresses concern that providing legal protection to AI-generated content could discourage human authorship and lead to a proliferation of low-quality works.
Empowering creators with disabilities
The report acknowledges the potential of AI to empower individuals with disabilities by assisting in the creative process. It emphasizes that copyright protection remains available where AI functions as an assistive tool, allowing human authors to express their creativity. For example, AI tools that help individuals with physical or cognitive disabilities to create works would not undermine the copyrightability of the resulting human-authored work.
Conclusion
The report emphasizes that copyright law has long adapted to new technologies and can continue to do so on a case-by-case basis. While the use of AI as an assistive tool does not affect the availability of copyright protection, purely AI-generated material without sufficient human control is not eligible for copyright. The report also highlights the importance of ongoing monitoring of technological and legal developments to evaluate the need for any changes in the future.
Client Alert 2025-035
