A number can change the shape of a copyright argument. At an October 21 hearing in San Jose, attorneys for music publishers told a federal judge that Anthropic’s Claude had received nearly 120,000 requests to generate synthetic lyrics during a six-month period. Anthropic answered that the publishers had not identified specific market harm. No ruling was issued.
The dispute is not a referendum on every use of artificial intelligence. It is a closely watched test of whether using copyrighted lyrics to train a generative system can qualify as fair use, and of how courts should evaluate a system that may also produce language competing with human-written work.
Why the output evidence matters
Fair use disputes often turn on purpose, transformation and market effect. Anthropic argues that training Claude on lyrics helped the system learn language and produce new content. The publishers, including Concord, Universal Music Group and ABKCO, argue that the process copied protected works and enabled competing synthetic lyrics.
The reported prompt count does not prove infringement by itself. A request can produce many kinds of results, and a court must evaluate evidence rather than headlines. But the figure shows why the case is moving beyond the abstract question of whether a model “learns” from culture. The judge is also being asked to consider what people ask the system to make and whether those outputs affect an existing or reasonably developing market.
That distinction matters to visual artists. Training litigation is frequently discussed as if the only relevant act occurs before a model is released. The hearing highlights a second layer: what the product does in public, how often users seek substitutes, what safeguards operate, and whether rights holders can document displacement or close imitation.
A record artists can learn from
For creators, the practical lesson is evidence. Save dated examples when a system appears to reproduce protected expression. Record the prompt, output, model name, version, date and any public sharing or commercial offer. Preserve the original file and its creation history. If a marketplace, client or platform replaces commissioned work with generated material, keep the communication that explains the decision.
None of those steps guarantees a legal claim, and artists should seek qualified advice for a specific dispute. They do, however, turn a general concern into a factual record. Courts distinguish between a theoretical risk and demonstrable market effect. The music publishers’ strategy shows the value—and the limits—of quantifying how a system is used.
The ruling is still ahead
U.S. District Judge Eumi Lee is considering cross-motions in Concord Music Group v. Anthropic. Other Northern California judges have reached different conclusions in earlier generative-AI training cases, so this decision may add an important piece without ending the wider debate.
Artists should resist premature certainty from either side. A decision for Anthropic would not make every dataset or output lawful. A decision for the publishers would not automatically resolve every visual-art case. The works, acquisition methods, product behavior, licenses and evidence of market harm differ.
What the hearing makes clear is that the next phase of AI copyright litigation will be argued not only through theories of innovation, but through records of actual use. For creators who want their rights taken seriously, documentation is becoming part of studio practice.
ATIM approaches artificial intelligence with curiosity, independence, and balance. We explore its extraordinary potential to empower artists while examining the questions it raises around authorship, copyright, consent, creative identity, transparency, and human expression.
We are neither advocates for AI nor opponents of it. Our focus is how the technology is used—and what that means for artists.
Above all, our commitment is to artists. We look at every development through the lens of what can support, protect, and empower the creative community—identifying opportunities, examining potential risks, and providing artists with the information they need to make informed decisions about their work, their rights, and their future.