Case id: 3:24-cv-05417
Parties: Plaintiffs: Andrea Bartz, Charles Graeber, and Kirk Wallace Johnson on behalf of themselves and all other similarly situated; Defendant: Anthropic PBC.
Court: District Court, N.D. California
Decision: Summary judgment on fair use; case settled before trial on piracy claims.
Decision date: 23 June 2025 (ruling)
Settlement date: 27 August 2025 (filed 5 September 2025)
Summary
In a significant ruling on June 23, 2025, U.S. federal Judge William Alsup delivered what initially appeared to be a major victory for AI companies. The court determined that the use of legally purchased books to train an AI model constitutes "fair use" under U.S. copyright law. Judge Alsup characterised this use as "quintessentially transformative," drawing a compelling analogy between teaching an AI how to write and a human reading a book to learn how to write (judgment). The court reasoned that since the AI was merely learning from the books, not copying or republishing them, it did not violate copyright laws. This decision was initially heralded as crucial for the growth and development of AI models.
Despite the initial fair use victory, the Bartz v. Anthropic case took a complex and potentially costly turn. Judge Alsup explicitly stated that the fair use defence likely does not apply to the millions of pirated books Anthropic used for training its Claude AI model. This critical element of the litigation was allowed to proceed and was subsequently upgraded to a class action lawsuit. This decision exposed Anthropic to potential liability for damages to the authors of an estimated seven million illegitimately sourced books. Under U.S. law, copyright owners can claim statutory damages of up to $150,000 per infringement (see 17 U.S.C. § 504 Remedies for infringement: Damages and profits), which could have led to what industry observers described as potentially "hundreds of billions of dollars" or even up to $1 trillion in damages if found liable for piracy (according to Wired).
Historic Settlement
On August 27, 2025, just months before the scheduled December trial, Anthropic reached a settlement with the class of authors. The settlement came after Judge Alsup's class certification order created what Anthropic characterised as "inordinate pressure" and a potential "death knell" situation for the company (according to Forbes).
On 5 September 2025, the Parties filed a Motion within which Anthropic has agreed to pay to the Plaintiffs at least US$1.5bn plus interest. Also part of the settlement terms is Anthropic's agreement to destroy the original files of the works it obtained from from online pirated sites/torrent downloads.
The District Court of California is expected to approve the settlement on 8 September 2025. The case represents one of the first major settlements in the wave of AI copyright litigation, potentially setting important precedents for future disputes between AI companies and content creators.
Legal Implications and Industry Impact
These developments within the Bartz v. Anthropic case demonstrate a crucial legal nuance: the process of AI learning can be deemed transformative and thus fair use, but the legality of data acquisition is a separate and overriding factor that can negate the fair use defence and lead to immense liability. The "how" of training is distinct from the "from where." This implies that AI companies cannot solely rely on the "transformative" nature of their models as a blanket defence. They must implement rigorous data provenance and due diligence practices to ensure their training datasets are legally sourced.
The settlement "sends a strong signal to AI companies that they can't pirate works of authorship instead of engaging legally," according to Maria A. Pallante, president and CEO of the Association of American Publishers (Bloomberg Law). This creates a strong incentive for proactive licensing agreements with copyright holders, especially for large-scale data ingestion.
The court's willingness to certify the class action despite Anthropic's objections demonstrates a judicial readiness to adapt procedural mechanisms to facilitate large-scale copyright enforcement against AI companies, especially when evidence points to massive, systemic illicit data acquisition. This ruling suggests that courts may prioritise the aggregation of claims from numerous impacted creators over the convenience or logistical challenges faced by AI defendants, potentially paving the way for more widespread class actions against AI companies, particularly where large-scale, potentially infringing datasets are used.
Ongoing Challenges
Anthropic isn't completely clear of copyright claims, as lawsuits brought by music publishers and Reddit Inc. remain pending in the same California district court. In Concord Music Group, Inc. et al. v. Anthropic PBC, No. 3:23-cv-01092 (M.D. Tenn. filed Oct. 18, 2023), transferred to No. 5:24-cv-03811 (N.D. Cal.), publishers including Universal Music Corp. and Concord Music Group Inc. sued Anthropic alleging the AI company engaged in wholesale copying of protected lyrics to train its signature large language model, Claude. Additionally, in Reddit, Inc. v. Anthropic PBC, No. 3:25-cv-05643 (N.D. Cal. filed June 4, 2025), Reddit sued Anthropic for allegedly using Reddit's platform and data without proper authorisation and in violation of Reddit's user agreement.
View case docket: https://www.courtlistener.com/docket/69058235/bartz-v-anthropic-pbc/?page=2
Summary judgment: https://storage.courtlistener.com/recap/gov.uscourts.cand.434709/gov.uscourts.cand.434709.231.0_4.pdf
Settlement filing: https://storage.courtlistener.com/recap/gov.uscourts.cand.434709/gov.uscourts.cand.434709.363.0.pdf
Disclaimer: The above is intended for information purposes only and does not constitute legal advice. Please refer to the terms and conditions page for more information.
Sources:
- Bartz v. Anthropic PBC, No. 3:24-cv-05417 (N.D. Cal.)
- Federal judge rules in AI company Anthropic's favor in landmark copyright infringement lawsuit brought by authors, NPR, 25 June, 2025, https://www.npr.org/2025/06/25/nx-s1-5445242/federal-rules-in-ai-companys-favor-in-landmark-copyright-infringement-lawsuit-authors-bartz-graeber-wallace-johnson-anthropic
- 17 U.S.C. § 504 Remedies for infringement: Damages and profits
- Anthropic Settles Major AI Copyright Suit Brought by Authors, Bloomberg Law, 1 August, 2025, https://news.bloomberglaw.com/ip-law/anthropic-settles-major-ai-copyright-suit-brought-by-authors
- Claude Now Joins OpenAI in Getting Sued for Copyright Infringement, 30 August 2025, https://www.forbes.com/sites/solrashidi/2025/08/30/claude-now-joins-openai-in-getting-sued-for-copyright-infringement/
- Concord Music Group, Inc. et al. v. Anthropic PBC, No. 3:23-cv-01092 (M.D. Tenn. filed Oct. 18, 2023), transferred to No. 5:24-cv-03811 (N.D. Cal.)
- Reddit, Inc. v. Anthropic PBC, No. 3:25-cv-05643 (N.D. Cal. filed June 4, 2025)