Comedian and actor Sarah Silverman joined a class of authors and publishers filing suit against the creators of OpenAI for infringing their copyrighted work.
Silverman claims generative artificial intelligence models such as ChatGPT are trained to produce unlawful derivative versions of her 2010 book, โThe Bedwetter,โ when the models are prompted to summarize the work. Silverman and the other class-authors claim they never granted permission for their copyrighted books to be used as training material for the large language models behind ChatGPT and other AI models.
LLMs are algorithms trained by culling data from text sources across the internet to learn the context, grammar, and language patterns requisite to generate human-like responses to users. According to the complaint against OpenAI, these algorithms are designed to pull copyrighted text directly from the authorsโ booksโas well as from unauthorized online โshadow librariesโ such as Library Genesisโand then feed that text into the LLMs.
Shadow libraries enable online users to torrent e-books, similar to how the infamous Pirate Bay network allowed users to download movies and TV shows. The writers claim that much of the content scraped from these libraries for OpenAIโs training datasets was copied, fed into, and remixed by its LLMs to generate user responses without the requisite consent, credit, or compensation of the original authors.
The writers seek statutory damages and recovery of lost profits from OpenAI for violations under state and federal law, including direct and vicarious copyright infringement, unfair competition, unjust enrichment, and negligence.
The case likely will turn on a โfair useโ defense. Fair use is a doctrine in the US that allows limited use of copyrighted content without the authorization of the copyright owner. The defense typically turns on key factors including the purpose of the use and whether the use threatens the success of the copyright owner by competing with their work.
In this case, OpenAI probably will argue that its LLM transforms the original work by summarizing or rephrasing the language in its user responsesโas opposed to merely repeating itโto qualify for a fair-use exemption. OpenAI might run into issues making this argument, however, if the class-plaintiffs can show that the tech giant is using its AI model and embedded copyright-infringing content to drive profits for the company.
In Andy Warhol Foundation for the Visual Arts v. Goldsmith, the Supreme Court found that the fair use defense was unavailable to defendants given that the apparent intent of their unauthorized use of the copyrighted work was for commercial gain. Mitch Glazier, chairman of the Recording Industry Association of America, a music advocacy organization, responded that โclaims of โtransformative useโ cannot undermine the basic rights given to all creators under the Copyright Act.โ
If AI companies sell access to their AI models after training them with copyrighted works, the Warhol decision may have rendered nil any argument that the companies sufficiently transform the original works to qualify for the fair use exemption. Still, it is uncertain how the courts will rule on a case-by-case basis due to a lack of consensus on the issue.
Perhaps more important than offering writers a possibility to collect damages for the alleged infringement of their work, this lawsuitโamong othersโhighlights the broader impact of AI on creative industries.
In May, several key players in the entertainment industry presented to a hearing before the House Judiciary Subcommittee on Courts, Intellectual Property, and the Internet to discuss the intersection of AI and copyright law.
Music executives and other industry leaders argued that creators need Congress to pass legislation that bars tech companies from the unauthorized collection and copying of their works to train their AI models.
Ashley Irwin, president of the Society of Composers and Lyricists, said at the hearing, โThe rapid introduction of generative AI systems is seen as an existential threat to the livelihood and continuance of our creative professions unless immediate steps are taken on legal interpretive and economic fronts to address these emerging issues โฆ Itโs essential to prioritize policies and regulations to safeguard the intellectual property and copyright of creators and preserve the diverse and dynamic US cultural landscape.โ
Irwin further argued that AI companies should be required to obtain creatorsโ consent before using their work to train their LLMs, provide due credit to the authors in any subsequent new work, and further compensate them at fair market value for any such derivative work.
Creators should hope that the courtsโ treatment of these cases will provide guidance on important questions regarding the ethical obligations of AI developers and whether there is need for stricter regulation to protect the rights of artists and authors in the digital age.
The case is Silverman v. OpenAI Inc., N.D. Cal., No. 3:23-cv-03416, filed 7/7/23.
This article does not necessarily reflect the opinion of Bloomberg Industry Group, Inc., the publisher of Bloomberg Law and Bloomberg Tax, or its owners.
Author Information
Thomas W. Brooke is an intellectual property partner at Holland & Knight.
Danielle N. Garno is a transactional and litigation partner at Holland & Knight and serves as a co-chair of the firmโs entertainment law team.
Sadie Mlika, a summer associate at Holland & Knight, contributed to this article.
Write for Us: Author Guidelines
