AI and Copyright: Courts Establish Framework for Fair Use in Artificial Intelligence Training

The rapid development and widespread adoption of artificial intelligence (AI) have created a complex legal landscape in the United States, particularly regarding copyright law. As AI training relies heavily on copyrighted materials, courts have been tasked with addressing the intersection of copyright, AI training, and fair use. Recent summary judgment decisions by U.S. Federal District Court judges in the Northern District of California have established a framework for how federal courts will address these issues. In this alert, we will break down these pivotal decisions, highlight other key cases and regulatory developments, and outline the implications for clients across industries.

Key Takeaways:

Judge William Alsup ruled in Bartz et al. v. Anthropic PBC that Anthropic's use of lawfully acquired, digitized books for AI training constituted fair use, emphasizing the "spectacularly" transformative nature of this process.
The court analogized AI training to a person internalizing knowledge from books, noting that such learning does not require ongoing royalties or licenses.
Judge Vince Chhabria ruled in Richard Kadrey et al. v. Meta Platforms, Inc. that Meta's use of copyrighted books to train its LLaMA model qualified as fair use because the works were used solely as raw material to teach language patterns, not to reproduce or compete with the originals.
The court emphasized that the plaintiffs failed to show that LLaMA's outputs substituted for or harmed the market for the original works.
The Delaware court found in Thomson Reuters v. ROSS Intelligence that ROSS's use of Westlaw's proprietary headnotes to train an AI legal research tool was not fair use, citing that ROSS's use was not transformative and directly competed with Westlaw's core market.
The U.S. Supreme Court held in Google v. Oracle that Google's copying of 11,500 lines of code from Oracle's Java SE program for use in Android was fair use, emphasizing the importance of "transformative" use in fair use analysis.
The Copyright Office's reports reiterate that fair use is a fact-specific inquiry, requiring careful consideration of purpose, nature, amount, and market effect.
Courts will ultimately define the boundaries of AI training under fair use, making legal guidance essential for clients.

Statistics:

20th Century legal principles governing copyright law are being applied to AI, but courts are struggling to keep pace.
7 million pirated books used by Anthropic were found to be not protected by fair use and could give rise to liability at trial.
11,500 lines of code were copied by Google from Oracle's Java SE program for use in Android, which was found to be fair use.
2025: U.S. Copyright Office reports and Thomson Reuters v. ROSS Intelligence decision.
2021: U.S. Supreme Court ruling in Google v. Oracle.

Sources:

Bartz et al. v. Anthropic PBC (June 23, 2025)
Richard Kadrey et al. v. Meta Platforms, Inc. (June 25, 2025)
Thomson Reuters v. ROSS Intelligence (D. Del. 2025)
Google v. Oracle (U.S. Supreme Court 2021)