Generative AI Meets Generative Litigation: News Corp Continues Its Battle Against Perplexity AI

In their amended complaint, filed on December 11, the plaintiffs have asserted copyright infringement, false designation of origin, and trademark dilution claims against Perplexity. The plaintiffs acknowledge that Perplexity seemingly responded to the plaintiffs’ initial demand letter by offering to share information concerning Perplexity’s revenue sharing model, but that the model fails to adequately compensate copyright holders and protect content. Dow Jones & Co., Inc. & NYP Holdings, Inc., v. Perplexity AI, Inc., No. 24-cv-7984-KPF (S.D.N.Y. Filed Dec. 11, 2024).

The lawsuit demands a jury trial and specifically alleges that Perplexity engaged in “massive illegal copying” by impermissibly scraping copyrighted content from the news outlets, allowing users to bypass or “Skip the Links” to the publishers’ websites and instead rely exclusively on Perplexity for news and analysis. The plaintiffs, owners of The Wall Street Journal, New York Post, Financial News, and Barron’s, claim the generative AI (GenAI) system scours the internet to gather information from their authoritative sources and then compiles the content into an easy-to-understand answer in response to a user’s question or prompt. According to the plaintiff, this adversely impacts the news outlets’ advertising and subscription profits, thereby leading to lost revenue and licensing opportunities within the evolving AI environment.

In addition to their copyright claims based on Perplexity’s “inputs” and “outputs,” the plaintiffs have also asserted trademark-related claims. The false designation of origin and dilution claims allege Perplexity’s GenAI system produces “hallucinations” or fabricated statements that are incorrectly attributed to the plaintiffs’ publications and use the plaintiffs’ trademarks, including “The Wall Street Journal” and the “New York Post,” without authorization.

The plaintiffs are seeking an injunction to halt the alleged scraping of their copyrighted material, the destruction of Perplexity’s indices and databases that contain the plaintiffs’ copyrighted works, and the monetary damages, among others.

This lawsuit differs from certain other pending IP lawsuits against GenAI companies in that it involves a “retrieval-augmented generation” (RAG) index. This AI architecture interacts with large language models (LLMs) to optimize LLM performance and increase their accuracy and reliability. RAG indices reference material outside of the model’s training data to augment the model’s performance and allow it to consider more recent or timely content when crafting responses to a user’s query. While arguably efficient for end-users, the plaintiffs argue journalists, editors, and staff who create the underlying content ultimately suffer.

This case highlights the escalating tensions over the use of copyrighted content in GenAI models. Its outcome could have far-reaching implications for the AI industry, potentially setting precedents regarding fair use, copyright limitations for generating AI outputs, and the necessity to update IP laws in the age of AI.