Anthropic has agreed to a significant $1.5 billion settlement for copyright infringement with over 300,000 authors. This marks one of the largest settlements of its kind. While some authors appreciate the outcome, others express mixed feelings.
The Settlement’s Background
Charles Graeber, among the main plaintiffs and a best-selling author, states he feels proud of challenging a major company like Anthropic. He views the effort as a testament to authors and publishers uniting to hold Anthropic accountable. However, he hesitates to claim the lawsuit as an outright success.
Despite gaining financial compensation, with Graeber receiving around $3,100 per book used by Anthropic, the process proved costly in terms of time and missed opportunities. Graeber notes the irony of feeling financially poorer despite the settlement.
Implications for Authors
Although the overall settlement appears substantial, the distribution among so many authors, combined with legal fees, dilutes individual payouts. Over $100 million went to attorneys involved in the case.
“Training AI on books is fair use under copyright law.” – Aparna Sridhar, Deputy General Counsel for Anthropic
Aparna Sridhar from Anthropic remarks the judgment doesn’t deem training AI on books illegal, provided there’s compensation. The resolution frames this under the fair use doctrine.
Challenges for Publishers and Authors
New York Times bestselling author Andrea Bartz argues against the fair use ruling, suggesting it potentially undermines human authors by allowing algorithms to create competitive texts.
Similar cases emerged, such as the lawsuit involving Meta. It reflected on the fair use issue, as authors failed to prove negative market effects caused by Meta’s use of their works.
The Global Copyright Picture
Intellectual property lawyer Dale Nelson highlights international complications for copyright enforcement, particularly with Chinese AI companies like DeepSeek. Incidents point to increasing complexity in protecting intellectual property globally.
Bartz, looking at international solutions, advocates for regulation and treaties to address these challenges collectively, suggesting treaties akin to nuclear non-proliferation agreements.
Future Steps and Licensing
Bartz expresses optimism about ongoing and future copyright infringement cases in the U.S. courts, emphasizing the need for authors to protect their rights amid growing AI capabilities.
Licensing offers a promising route. Umair Kazi of the Authors Guild suggests licensing ensures legal training and places limitations on how AI outputs use copyrighted works.
Some deals, like Perplexity AI’s license agreements with media entities, illustrate licensing potential. Yet, the trend hasn’t gained widespread traction due to ongoing litigation among major AI companies.