The U.S. Copyright Office has released Part 2 of its Report on Copyright and Artificial Intelligence, addressing the copyrightability of AI-generated works and reaffirming that human authorship remains essential for copyright protection in the United States. The report, a continuation of the Office’s AI initiative launched in early 2023, clarifies the level of human involvement required for AI-assisted works to qualify for copyright and examines how other countries approach similar issues.
The Office emphasizes that copyright law in the U.S. requires human authorship, citing the Copyright Clause in the Constitution and various legal precedents. Courts have consistently ruled that non-human entities cannot hold copyrights, a position the Office upholds. The Supreme Court has previously stated that an “author” is the person who translates an idea into a fixed, tangible expression entitled to copyright protection. Based on this principle, the Office asserts that AI-generated works alone cannot receive copyright protection, but works with sufficient human creativity may qualify.
The report outlines three scenarios in which AI-generated works may be eligible for copyright protection. First, AI tools used as an assistive mechanism, where the final creative expression is fundamentally human-authored, should not affect copyright eligibility. Second, when human-authored content is input into an AI system and remains perceptible in the output, copyright protection extends only to the human-created elements. Third, when AI-generated material is arranged or modified in a sufficiently creative way by a human, that specific human-authored contribution can be protected.
However, the Office firmly concludes that prompts alone do not constitute authorship, as AI systems produce unpredictable variations even when given identical inputs. The report notes that AI functions as a “black box,” with users and developers often unable to predict the exact outputs. As such, merely crafting a prompt is not enough to warrant copyright protection for the resulting AI-generated work.
The report also compares global approaches to AI-generated copyright. The European Union allows copyright protection only if significant human input is involved. The United Kingdom, under a pre-existing statute, grants copyright to computer-generated works where no human author exists, though this is currently under review. Japan evaluates copyright eligibility based on factors like user input, the number of generation attempts, and post-processing edits. China, in contrast, grants copyright to the individual using AI to create a work.
No new Acts Required
Despite calls for new protections for AI-generated materials, the Office does not see the need for legislative changes. It argues that existing U.S. copyright laws are flexible enough to accommodate AI advancements while ensuring human creativity remains protected. The report expresses concern that excessive legal protections for AI-generated works could diminish incentives for human authors, potentially stifling creative output.
Legal professionals are advised to consider the Office’s stance when assessing AI-related copyright matters. The key takeaways include the necessity of human authorship for copyright protection, the case-by-case evaluation of AI-assisted works, and the exclusion of AI prompts as a basis for copyright claims. Additionally, while AI-assisted creations may be protected under specific conditions, new legal frameworks are not currently needed.
The Copyright Office will continue to monitor technological and legal developments in AI and copyright law. The upcoming Part 3 of the report will address legal concerns related to training AI models on copyrighted works, licensing considerations, and the allocation of liability in AI-generated content.