The US Copyright Office released its long-awaited report on generative AI training, sparking controversy over copyright infringement. The report’s findings have significant implications for major tech companies like OpenAI and Google.
In a bold departure from the narrative relied upon by Big Tech, the Copyright Office rejected the broadest fair use defense advanced by AI companies – that training models on copyrighted content is inherently transformative because it doesn’t express the same ideas. Instead, the office suggests that a larger training set makes the process less likely to be transformative.
The report also explores a new theory of market harm from AI, known as “market dilution,” which warns that floodgates of AI-generated content could dilute the market for human-authored works. This theory mirrors real-world fears from creators who worry about being displaced by plagiarism and sheer volume of potential category bloat.
Moreover, the report puts to rest a long-simmering debate around opt-out systems for copyright holders, shutting down a key proposal favored by tech giants. Instead, the office suggests exploring voluntary collective licensing mechanisms, which could allow publishers, authors, and other rights holders to be compensated for training use of their works.
The timing of the report’s release is particularly noteworthy, coming just days after President Trump fired Librarian of Congress Carla Hayden and Register of Copyrights Shira Perlmutter. This raises questions about whether the Copyright Office was expecting its own leadership to be next.
As policymakers and developers grapple with the implications of this report, one thing is clear: can the law catch up with the pace of AI?
Source: https://www.forbes.com/sites/torconstantino/2025/05/29/us-copyright-office-shocks-big-tech-with-ai-fair-use-rebuke