Anthropic Defends AI Training as Fair Use in Copyright Lawsuit Over Song Lyrics

Anthropic Faces Critical Fair Use Arguments in AI Copyright Case

Anthropicis defending its use of copyrighted song lyrics to train its AI chatbot Claude in a landmark copyright lawsuit brought by Universal Music Group, Concord, and ABKCO. US District Judge Eumi Lee heard arguments on October 21 in San Jose, California, but did not issue an immediate ruling on the fair use defense that could determine the fate of dozens of similar AI training copyright cases1

1

2

2

. The music publishers filed the lawsuit in 2023, alleging that Anthropic infringed copyrights on hundreds of songs by artists including Beyoncé, the Rolling Stones, and the Beach Boys when using their lyrics to train Claude to respond to human prompts1

1

.

Competing Arguments Over Transformative Use and Market Harm

Anthropicargued through its lawyer Sonal Mehta of WilmerHale that its AI training makes transformative use of song lyrics to help Claude understand human language and enable progress in science, business, and education. The company contends that allowing copyright holders to veto such transformative technology would contradict the Supreme Court's mandate that copyright's primary objective is serving the public, not rewarding authors1

1

. Mehta emphasized that AI has not reduced demand for lyrics and that the music publishers could not identify specific instances of market harm from Claude or other large language models. She noted that Universal Music Group itself has described AI as "overwhelmingly a net positive" for the music industry2

2

.

The music publishers countered through attorney Matt Oppenheim of Oppenheim + Zebrak that Anthropic misuses lyrics to generate competing content that harms human songwriters. Oppenheim revealed that Claude received nearly 120,000 requests to generate "synthetic" lyrics over a six-month period, arguing that large language models "could overwhelm the market with AI-generated content very easily"2

2

. The publishers assert that Anthropic's use is not fair use because it enables Claude to generate new lyrics that dilute the market for their works, with AI-generated songs already saturating streaming platforms and climbing Billboard charts1

1

.

Split Judicial Opinions Create Uncertainty for AI Companies

The legal landscape around training AI on song lyrics and whether it constitutes fair use remains uncertain, with two Northern California judges reaching opposite conclusions in similar cases in 2025. Now-retired US District Judge William Alsup sided with Anthropic, ruling that its AI training made fair use of books. Just two days later, US District Judge Vince Chhabria ruled for Meta Platforms in a similar lawsuit but warned that AI training "in many circumstances" would not be fair use1

1

2

2

. This split creates significant uncertainty for tech companies facing copyright infringement allegations.

Judge Lee pressed both sides on the fair use question and challenged the publishers' assertion that Claude's ability to compose lyrics threatens their market position2

2

. How judges like Lee rule on the fair use defense will likely determine the outcome of pending copyright lawsuits brought by authors, news outlets, visual artists, and others against tech giants including Anthropic, OpenAI, and Meta Platforms over their use of copyrighted material for AI training without content owners' permission1

1

.

Global Implications and India's Proposed Licensing Framework

The case carries implications beyond US borders, with India proposing its own approach to AI training and copyright. On December 8, 2025, India's Department for Promotion of Industry and Internal Trade proposed a mandatory blanket licensing framework dubbed "one nation, one licence, one payment" for AI training. Under this proposal, lawfully accessed works cannot be withheld by rights holders, who would instead receive royalties determined by a government-appointed committee when AI systems are commercialized3

3

. This contrasts with the US fair use doctrine, as India provides specific fair-dealing exceptions under Section 52 of the Copyright Act, 1957, rather than a general fair use defense3

3

.

Anthropicalso faces additional copyright lawsuits involving song lyrics brought by Sony Music Publishing and Round Hill Music, while Indian courts have yet to resolve the legality of using copyrighted works to train AI models3

3

. The outcome of this case will establish critical legal precedent for how intellectual property rights intersect with AI development, potentially reshaping licensing frameworks and determining whether AI companies must obtain permission and pay for training data or can rely on transformative use arguments to access copyrighted material freely.