# Silverman v. OpenAI

Silverman v. OpenAI was a 2023 copyright infringement lawsuit by authors Sarah Silverman, Richard Kadrey, and Christopher Golden against OpenAI, alleging unauthorized use of their books in training ChatGPT. The case was partially dismissed in 2024, with claims for direct infringement dismissed but unfair competition claims allowed to proceed.

Silverman v. OpenAI was a federal lawsuit filed in the United States District Court for the Northern District of California in July 2023. The plaintiffs, authors Sarah Silverman, Richard Kadrey, and Christopher Golden, alleged that OpenAI, the developer of the [large-language-model](https://www.wikiprompt.org/wiki/large-language-model) ChatGPT, infringed their copyrights by using their books as training data without permission. The case became one of the first high-profile legal challenges to the practice of training generative [artificial-intelligence](https://www.wikiprompt.org/wiki/artificial-intelligence) systems on copyrighted text, raising questions about fair use, the transparency of training datasets, and the liability of AI companies for outputs that resemble protected works.

The lawsuit was filed on July 7, 2023, and later consolidated with similar actions. The plaintiffs sought statutory damages, injunctive relief, and a declaration that OpenAI's use of their works constituted copyright infringement. They also alleged violations of the Digital Millennium Copyright Act (DMCA) and unjust enrichment. The case drew widespread attention because it tested whether the 'fair use' doctrine, which permits limited use of copyrighted material without permission, extends to the massive-scale ingestion of books for machine-learning purposes.

## Background and Legal Context

The rise of [generative-ai](https://www.wikiprompt.org/wiki/generative-ai) systems, particularly [large language models](https://www.wikiprompt.org/wiki/large-language-model) like ChatGPT, relies on training datasets that often include millions of books, articles, and web pages. OpenAI's training process involved scraping publicly available text, including copyrighted works, to teach the model to predict and generate human-like text. The plaintiffs argued that this use was not transformative but rather a form of unauthorized copying that deprived authors of licensing revenue and control over their creations.

The case was part of a broader wave of litigation against AI companies. Other authors, including John Grisham, George R.R. Martin, and Jodi Picoult, filed similar suits against OpenAI and other developers. The outcome of Silverman v. OpenAI was seen as a bellwether for how courts would interpret fair use in the context of AI training.

## Plaintiffs' Claims

Silverman, Kadrey, and Golden alleged that OpenAI had copied their books in full to create the training dataset, which was then used to train ChatGPT. They argued that this copying was not protected by fair use because it was commercial, non-transformative, and harmed the market for their works. They also claimed that ChatGPT could generate summaries or excerpts of their books, demonstrating that the model had memorized substantial portions of the texts.

The plaintiffs further alleged that OpenAI violated the DMCA by removing copyright management information from the books, such as title pages and copyright notices, during the digitization process. They sought damages for both direct and vicarious infringement, as well as for unjust enrichment.

## OpenAI's Defense

OpenAI moved to dismiss the lawsuit, arguing that the use of copyrighted works in training was fair use. The company contended that training data is used to create statistical patterns, not to reproduce expressive content, and that the resulting model does not compete with the original works. OpenAI also argued that the plaintiffs lacked standing because they had not registered their copyrights with the U.S. Copyright Office, a prerequisite for filing an infringement suit.

In its motion to dismiss, OpenAI emphasized that the plaintiffs had not alleged that ChatGPT produced infringing copies of their books, only that the training process involved copying. The company argued that such intermediate copying is transformative and has been upheld in prior cases involving search engines and text-mining.

## Court Rulings

On February 12, 2024, Judge Araceli Martínez-Olguín of the Northern District of California granted in part and denied in part OpenAI's motion to dismiss. The court dismissed the plaintiffs' claims for direct copyright infringement, holding that they had not plausibly alleged that ChatGPT outputs infringed their works. The court reasoned that the plaintiffs had not shown that the model generated reproductions or derivative works of their books, only that it could produce summaries or excerpts, which the court found insufficient.

However, the court allowed the plaintiffs' claim for unfair competition under California law to proceed, as well as a claim for negligent interference with prospective economic advantage. The court also granted the plaintiffs leave to amend their complaint to address the deficiencies in their copyright claims. The DMCA claims were dismissed with prejudice, as the court found that the plaintiffs had not adequately alleged that OpenAI removed copyright management information.

The ruling was seen as a mixed outcome. While the direct infringement claims were dismissed, the case continued on other grounds, and the plaintiffs were given an opportunity to replead. The decision also highlighted the difficulty of proving infringement when the alleged copying occurs in the training process rather than in the model's outputs.

## Subsequent Developments

Following the dismissal, the plaintiffs filed an amended complaint in March 2024, adding more specific allegations about ChatGPT's ability to reproduce copyrighted text. They also sought to certify a class of authors whose works were used in training. OpenAI again moved to dismiss, and the court held a hearing in October 2024. As of early 2025, the case remained pending, with no final ruling on the merits.

The case was part of a larger legal landscape. In 2024, the U.S. Copyright Office initiated a study on the copyright implications of AI, and Congress held hearings on the issue. Other courts were also grappling with similar questions, including a case involving the New York Times and OpenAI, which was filed in December 2023.

## Impact on AI Industry

Silverman v. OpenAI had significant implications for the [artificial-intelligence](https://www.wikiprompt.org/wiki/artificial-intelligence) industry. The case prompted AI companies to reconsider their data sourcing practices. Some companies, including [OpenAI](https://www.wikiprompt.org/wiki/openai), began offering opt-out mechanisms for copyright holders and licensing agreements with publishers. The case also influenced discussions about the need for transparency in training datasets and the potential for a statutory licensing scheme.

The litigation highlighted the tension between innovation and intellectual property rights. While AI developers argued that broad access to text is essential for creating useful models, authors and publishers contended that they deserve compensation and control. The outcome of the case, and others like it, could shape the future of [machine-learning](https://www.wikiprompt.org/wiki/machine-learning) and the commercial viability of [generative-ai](https://www.wikiprompt.org/wiki/generative-ai) systems.

## Legal and Ethical Considerations

The case raised several legal and ethical questions. One key issue was whether the 'fair use' doctrine, which has traditionally been applied to limited copying for purposes such as criticism, comment, or research, should extend to the wholesale ingestion of copyrighted works for AI training. Courts have previously upheld text-mining as fair use in cases like Authors Guild v. Google, but those cases involved search indexes, not generative models that can produce expressive content.

Another issue was the role of [neural networks](https://www.wikiprompt.org/wiki/neural-network) in creating derivative works. The plaintiffs argued that ChatGPT's ability to generate text in the style of an author, or to produce summaries that capture the essence of a book, could constitute derivative works. OpenAI countered that the model's outputs are not substantially similar to the original works and that the model does not store copies of the texts.

Ethically, the case underscored the need for consent and compensation when using creators' works. It also raised concerns about the concentration of power in AI companies, which can access vast amounts of data without meaningful oversight. The case contributed to a broader debate about the social contract between technology developers and content creators.

## Conclusion

Silverman v. OpenAI remains a landmark case in the evolving field of AI law. Its rulings have provided early guidance on how courts may treat copyright claims against AI companies, but many questions remain unresolved. The case is likely to be appealed, and its final outcome could have far-reaching consequences for the [company](https://www.wikiprompt.org/wiki/openai) and the broader [deep learning](https://www.wikiprompt.org/wiki/deep-learning) community. As of early 2025, the litigation was ongoing, and the legal landscape for AI and copyright continued to develop.

---
Source: https://www.wikiprompt.org/wiki/silverman-v-openai
License: CC BY-SA 4.0 (https://creativecommons.org/licenses/by-sa/4.0/)
Last updated: 2026-09-14T04:11:38.889665+00:00