The head of the US Copyright Office, Shira Perlmutter, has reportedly been dismissed just one day after the agency released a report suggesting that the use of copyrighted material to train artificial intelligence models may fall outside the bounds of fair use, according to the US local media.
The finding came in a draft of the agency’s third report on copyright and AI, published on 9 May. The first two parts of the Copyright Office’s multi-phase study focused on digital replicas and whether generative AI outputs can be copyrighted. Part Three examines whether using copyrighted content to train AI systems requires the consent or compensation of copyright holders.
The draft report concluded that while certain uses, such as training models for research or analysis, may not violate fair use, commercial deployments that generate expressive outputs competing with original works likely do. Particularly, the report warns that training AI models on vast amounts of copyrighted content accessed without permission crosses legal boundaries, especially when those outputs are sold in existing creative markets.
Legal experts say the report could have major implications for several high-profile lawsuits. Major AI companies, including OpenAI, Meta, Google, and Microsoft, are currently facing litigation over copyright violations. Many have defended their practices under fair use provisions.
“This is very bad news for the AI companies in litigation,” Blake E. Reid, a professor of technology law stated to The Register, who called the findings “a straight-ticket loss” for tech firms relying on fair use defenses.
The Copyright Office emphasized that the conclusions of the draft are unlikely to change in the final version, which is expected to be released soon.
The timing of Perlmutter’s removal has sparked speculation. Representative Joe Morelle (D-NY) suggested the firing may have been politically motivated, noting it came less than 24 hours after Perlmutter “refused to rubber-stamp Elon Musk’s efforts to mine troves of copyrighted works to train AI models.” Musk has previously expressed support for dismantling intellectual property laws and has indicated plans to train his own AI system, Grok, on content from X (formerly Twitter).
The Copyright Office falls under the jurisdiction of the Library of Congress. Its leadership also recently saw a shake-up: the Librarian of Congress was dismissed amid controversy over diversity and inclusion policies and content decisions involving children’s materials.
Whether Perlmutter’s removal is linked to political efforts to reshape federal cultural institutions or specifically to AI-related copyright policy remains unclear.