By: Landon Coles (2025 Summer Associate), Joshua A. Bloomgarden, Hillary Hughes
Artificial intelligence (“AI”) continues to grow as a fixture of daily life, but not everyone is embracing it (according to a recent Pew Research Center survey, about half of U.S. adults (51%) say they are more concerned than excited about the increased use of AI), and even fewer have an understanding of how to leverage this technology while respecting creators’ intellectual property rights. Through a patchwork of government regulations and ongoing litigation, we are gradually beginning to see contours emerge around AI policy, albeit with plenty of grey areas persisting.
Government Regulation
Comprehensive federal regulation, specifically regarding copyright infringement, continues to remain elusive. Legislation has largely been aimed at addressing specific harms caused by AI technology, such as deepfake technology to spread misinformation on elections and candidates, generate nonconsensual pornography or to mimic music artists’ voices without permission.
President Donald J. Trump has taken an expansive approach to AI, signing an Executive Order in January 2025 to revoke “existing AI policies and directives that act as barriers to American AI innovation,” though the application of that Executive Order is unclear. In May 2025, President Trump signed into law the bipartisan Take It Down Act which makes it illegal to “knowingly publish” or threaten to publish intimate images without a person’s consent, including AI-created “deepfakes.”
In response to the void of federal regulation, states have enacted a patchwork of legislation with over 30 states enacting laws addressing deepfakes as of May 2025.
Litigation Related to AI Copyright Infringement
On the copyright front, leading technology companies such Meta, ChatGPT-maker OpenAI and Microsoft have been embroiled in controversy in recent years related to the practice of training AI language models on copyrighted material. These companies have argued they are immune from liability under the “fair use” doctrine, pursuant to which courts apply a four-factor test on the allegedly infringing uses analyzing: (1) the purpose and character of the use, (2) the nature of the copyrighted work, (3) the amount and substantiality of the portion used and (4) the effect of the use on the potential market.
Bartz v. Anthropic PBC
In Bartz v. Anthropic PBC, the United States District Court for the Northern District of California ruled in favor of Anthropic, finding that use of copyrighted books to train large language models is a fair use due to its “highly transformative” nature. Bartz v. PBC, 2025 U.S. Dist. LEXIS 136877, *56. There, the court distinguished Anthropic’s conversion of purchased print books to digital format for a central library (which the court found to be a fair use), as opposed to Anthropic’s use of pirated copies of books (which the court found not to be a fair use). Anthropic has since reached a landmark $1.5 billion settlement (pending court approval) with the authors and copyright holders of the pirated works.
Kadrey v. Meta Platforms, Inc.
In a later case, the United States District Court for the Northern District of California, ruled in favor of Meta, finding that the company’s use of the plaintiffs' copyrighted books to train its large language model Llama, was highly transformative and favored fair use under the first fair use factor. Kadrey v. Meta Platforms, Inc., 2025 U.S. Dist. LEXIS 121064, *1 The court did limit the scope of the ruling, stating the case is “not a class action,” and “does not stand for the proposition that Meta's use of copyrighted materials to train its language models is lawful.”
Thomson Reuters Enterprise Centre GMBH v. Ross Intelligence Inc.
Earlier this year, Thomson Reuters sued Ross Intelligence, alleging that Ross violated Thomson Reuters’ copyright in instances where Ross used Thomson Reuters’ Westlaw headnotes to train Ross’ new AI legal-research search engine. The United States District Court for Delaware granted partial summary judgement for Thomson Reuters, affirming that Ross improperly used 2,243 Westlaw headnotes and that Ross’s use harmed the market for Thomson Reuters’s headnotes and derivative products. Thomson Reuters Enter. Ctr. GmbH v. Ross Intel. Inc., 765 F. Supp. 3d 382, 390 (D. Del. 2025)
Pending Cases
-
- Earlier this year, several major publishers, including Conde Nast, The Atlantic, Politico and Vox sued AI startup, Cohere, alleging the improper use of more than 4,000 copyrighted works to train its large language model and unauthorized publishing of large portions of or entire articles for users while bypassing visits to the publishers’ websites.
- In October 2024, Concord Music Group alongside several music publishers filed a copyright lawsuit against AI developer, Anthropic PBC, over its use of copyrighted song lyrics to train its AI model, Claude. In January 2025, the court entered a judgement permitting Anthropic to continue training its AI model while the litigation proceeded. As of July 2025, the case remains active, as the court will have to resolve the action, which seeks to enjoin Anthropic’s alleged use of unauthorized copies of the publishers’ song lyrics to train AI models.
- A group of authors, including Sarah Silverman, filed suit over the unauthorized use of their novels by Anthropic to train its artificial intelligence chatbot, ChatGPT. Most of the claims, including claims for negligence, unjust enrichment and vicarious copyright infringement, have since been dismissed. The primary claim for direct copyright infringement remains pending.
- Getty Images (US) alleged that Stability AI unlawfully used over 12 million Getty Images to train its Stable Diffusion model without consent. The case, currently before the UK High Court, saw Getty recently drop its direct copyright claims, choosing instead to focus on claims related to trademark, passing off and secondary infringement. A verdict on this case is expected later this year.
Please note that these cases are just a small survey of the many ongoing disputes between AI developers and copyrights holders, and the outcome of particular cases may vary based on the facts and technology at issue. As navigating intellectual property law in the age of AI continues to be a difficult task, Foster Garvey is actively monitoring these developments and we encourage you to contact us if you are in need of guidance.
