top of page

Perplexity AI Lawsuit: RAG Systems and the Fight Over Copyright and Trademark

Writer's picture: Lewis D. SorokinLewis D. Sorokin


Dow Jones & NYP Holdings have filed a lawsuit against Perplexity AI, alleging massive copyright infringement and trademark dilution. The complaint, filed in the U.S. District Court for the Southern District of New York, claims that Perplexity’s AI answer engine illegally copies vast amounts of news content from The Wall Street Journal and New York Post to generate responses for users without authorization.


The lawsuit focuses on Perplexity’s “retrieval-augmented generation” (RAG) system, which the plaintiffs argue creates a substitute for their original content. Perplexity encourages users to “skip the links” to the original articles, thereby diverting critical revenue away from news publishers by reducing ad impressions and subscription opportunities. The plaintiffs further allege that Perplexity’s outputs often include verbatim reproductions or paraphrases of their articles, and sometimes even generate false information, leading to trademark dilution and damage to their brand. Rather than building its own foundation model, Perplexity chooses to forego the complexity of pre-training and instead combines its RAG system with a combination of foundation models from OpenAI, Claude, and their own fine-tuned version of Meta’s open-source LLaMA. To this end, they take the stance that their web crawler’s approach to robots.txt files is no different from any other crawlers built for indexing the web for search engines.


Dow Jones and NYP Holdings are also pursuing claims of trademark dilution, arguing that Perplexity’s misuse of their well-known trademarks, such as The Wall Street Journal and New York Post, in connection with AI-generated content—including fabricated or false information. This tarnishes their brand reputation by associating their trademarks with unreliable content. The plaintiffs assert that Perplexity's outputs often attribute hallucinated or incorrect content to their publications, which confuses users and devalues the credibility and trust associated with these iconic brands.


Trademark dilution occurs when the distinctiveness or value of a famous mark is weakened, and in this case, Dow Jones and NYP Holdings claim that the association of their trademarks with unreliable or inaccurate AI-generated content erodes consumer trust and harms their standing in the news industry. This adds another layer of complexity to the case, as it raises questions not only about copyright infringement but also about the integrity of well-established brands in the face of emerging AI technologies.


While this case shares some similarities with the ongoing New York Times v. OpenAI lawsuit—which involves claims of unauthorized use of copyrighted content in training language models—it is distinct in that it highlights the direct market impact of Perplexity’s model, which not only reproduces content but also competes with news publishers by replacing the need for users to visit the original sites. Notably, less than a week earlier, the New York Times sent a cease-and-desist letter to Perplexity regarding its use of their content.


As the intersection of AI and intellectual property continues to develop, this case will be important to watch. It raises new questions about the role that RAG plays in the increasingly messy web of AI and IP rights, and emphasizes the harm to brands caused by AI-generated hallucinations—false content attributed to trusted publications. Given the stakes involved, I anticipate a hard-fought legal battle ahead.


As AI continues to evolve, it’s crucial for IP owners and technologists alike to advocate for their respective interests in this growing industry. How do we protect intellectual property while fostering innovation in AI? I’d love to hear your thoughts—what’s the path forward for balancing these competing interests?

bottom of page