The US District Court for the Northern District of California granted summary judgment in favor of an artificial intelligence (AI) company
Aktualności|
Getting your Trinity Audio player ready...
|
Andrea Bartz, et al. v. Anthropic PBC, Case No. 24-CV-05417-WHA (N.D. Cal. June 23, 2025) (Alsup, J.)
alerts-practices-aiml-district-court-issues-ai-fair-use-decision
17 U.S. Code § 107 – Limitations on exclusive rights: Fair use
Notwithstanding the provisions of sections 106 and 106A, the fair use of a copyrighted work, including such use by reproduction in copies or phonorecords or by any other means specified by that section, for purposes such as criticism, comment, news reporting, teaching (including multiple copies for classroom use), scholarship, or research, is not an infringement of copyright. In determining whether the use made of a work in any particular case is a fair use the factors to be considered shall include—
(1)
the purpose and character of the use, including whether such use is of a commercial nature or is for nonprofit educational purposes;(2)
the nature of the copyrighted work;(3)
the amount and substantiality of the portion used in relation to the copyrighted work as a whole; and(4)
the effect of the use upon the potential market for or value of the copyrighted work.The fact that a work is unpublished shall not itself bar a finding of fair use if such finding is made upon consideration of all the above factors.(Pub. L. 94–553, title I, § 101, Oct. 19, 1976, 90 Stat. 2546; Pub. L. 101–650, title VI, § 607, Dec. 1, 1990, 104 Stat. 5132; Pub. L. 102–492, Oct. 24, 1992, 106 Stat. 3145.)Source:
,,The US District Court for the Northern District of California granted summary judgment in favor of an artificial intelligence (AI) company, finding that its use of lawfully acquired copyrighted materials for training and its digitization of acquired print works fell within the bounds of fair use. However, the district court explicitly rejected the AI company’s attempt to invoke fair use as a defense to rely on pirated copies of copyrighted works as lawful training data.
Anthropic, an AI company, acquired more than seven million copyrighted books without authorization by downloading them from pirate websites. It also lawfully purchased print books, removed their bindings, scanned each page, and stored them in digitized, searchable files. The goal was twofold:
- To create a central digital library intended, in Anthropic’s words, to contain “all the books in the world” and to be preserved indefinitely.
- To use this library to train the large language models (LLMs) that power Anthropic’s AI assistant, Claude.
Each work selected for training the LLM was copied through four main stages:
- Each selected book was copied from the library to create a working copy for training.
- Each book was “cleaned” by removing low-value or repetitive content (e.g., footers).
- Cleaned books were converted into “tokenized” versions by being simplified and split into short character sequences, then translated into numerical tokens using Anthropic’s custom dictionary. These tokens were repeatedly used in training, allowing the model to discover statistical relationships across massive text data.
- Each fully trained LLM itself retained “compressed” copies of the books.
Once the LLM was trained, it did not output any of the books through Claude to the public. The company placed particular value on books with well-curated facts, structured analyses, and compelling narratives (i.e., works that reflected well-written creative expressions) because Claude’s users expected clear, accurate, and well-written responses to their questions.
Andrea Bartz, along with two other authors whose books were copied from pirated and purchased sources and used to train Claude, sued Anthropic for copyright infringement. In response, Anthropic filed an early motion for summary judgment on fair use only under Section 107 of the Copyright Act.
To assess the applicability of the fair use defense, the court separated and analyzed Anthropic’s actions across three distinct categories of use.
Transformative training (fair use)
The authors challenged only the inputs used to train the LLMs, not their outputs. The district court found that Anthropic’s use of copyrighted books to train its LLMs was a transformative use, comparable to how humans read and learn from texts and produce new, original writing. While the authors claimed that the LLMs memorized their creative expression, there was no evidence that Claude released infringing material to the public. The court concluded that using the works as training inputs – not for direct replication, but to enable the generation of new content – favored a finding of fair use.
Format-shifting copies (fair use)
The authors challenged Anthropic’s conversion of the copyrighted works from print to digital format, although they did not allege that Anthropic distributed any of the digital copies outside the company. The district court found that Anthropic had lawfully purchased the print editions and acquired the right to retain and use them for all ordinary purposes. Each print copy was digitized to save space and enable search functionality, and the original was destroyed after conversion. The court concluded that the print-to-digital format change was transformative under fair use.
Liability for piracy (not fair use)
The district court agreed with the authors that Anthropic’s downloading and retention of more than seven million pirated books – without payment – was not a fair use, regardless of whether the books were ultimately used to train its AI models. Even after Anthropic decided not to train its LLMs on those pirated copies, it kept them as part of a central research library, a use the court found inherently infringing and non-transformative. The court rejected Anthropic’s argument that its long-term goal of a transformative use (training LLMs) could retroactively justify the initial infringement, emphasizing that each act of copying must be judged by its own objective use. The court explained that “such piracy of otherwise available copies is inherently, irredeemably infringing even if the pirated copies are immediately used for the transformative use and immediately discarded.”
Anthropic now faces a jury trial limited to damages for its pirated copies.
Practice note: This is the first federal court decision analyzing the defense of fair use of copyrighted material to train generative AI. Two days after this decision issued, another Northern District of California judge ruled in Kadrey et al. v. Meta Platforms Inc. et al., Case No. 3:23-cv-03417, and concluded that the AI technology at issue in his case was transformative. However, the basis for his ruling in favor of Meta on the question of fair use was not transformation, but the plaintiffs’ failure “to present meaningful evidence that Meta’s use of their works to create [a generative AI engine] impacted the market” for the books.
”
,,United States District Court ruled that training LLMs on copyrighted books constitutes fair use. A number of authors had filed suit against Anthropic for training its models on their books without permission. Just as we allow people to read books and learn from them to become better writers, but not to regurgitate copyrighted text verbatim, the judge concluded that it is fair use for AI models to do so as well.
Indeed, Judge Alsup wrote that the authors’ lawsuit is “no different than it would be if they complained that training schoolchildren to write well would result in an explosion of competing works.” While it remains to be seen whether the decision will be appealed, this ruling is reasonable and will be good for AI progress.
AI has massive momentum, but a few things could put progress at risk:
- Regulatory capture that stifles innovation, including especially open source, in the false name of “AI safety”
- Loss of access to cutting-edge semiconductor chips (the most likely cause would be war breaking out in Taiwan)
- Regulations that severely impede access to data for training AI systems
Access to high-quality data is important. Even though the mass media tends to talk about the importance of building large data centers and scaling up models, when I speak with friends at companies that train foundation models, many describe a very large amount of their daily challenges as data preparation. Specifically, a significant fraction of their day-to-day work follows the usual Data Centric AI practices of identifying high-quality data (books are one important source), cleaning data (the ruling describes Anthropic taking steps like removing book pages’ headers, footers, and page numbers), carrying out error analyses to figure out what types of data to acquire more of, and inventing new ways to generate synthetic data.
I am glad that a major risk to data access just decreased. Appropriately, the ruling further said that Anthropic’s conversion of books from paper format to digital — a step that’s needed to enable training — also was fair use. However, in a loss for Anthropic, the judge indicated that, while training on data that was acquired legitimately is fine, using pirated materials (such as texts downloaded from pirate websites) is not fair use. Thus, Anthropic still may be liable on this point. Other LLM providers, too, will now likely have to revisit their practices if they use datasets that may contain pirated works.
Overall, the ruling is positive for AI progress. Perhaps the biggest benefit is that it reduces ambiguity with respect to AI training and copyright and (if it stands up to appeals) makes the roadmap for compliance clearer. This decision indicates it is okay to train on legitimately acquired data to build models that generate transformational outputs, and to convert printed books to digital format for this purpose. However, downloading from pirate sites (as well as permanently building a “general purpose” library of texts, stored indefinitely for purposes to be determined, without permission from the relevant copyright holders) are not considered fair use.
I am very sympathetic with the many writers who are worried about their livelihoods being affected by AI. I don‘t know the right solution for that. Society is better off with free access to more data; but if a subset of people is significantly negatively affected, I hope we can figure out an arrangement that compensates them fairly. ,,
Sources:
https://www.govinfo.gov/app/details/USCOURTS-cand-3_24-cv-05417/summary
https://natlawreview.com/article/fair-use-or-foul-play-ai-fair-use-copyright-line