The Chicago Tribune on Thursday filed a lawsuit against the Perplexity AI search engine, alleging copyright infringement in New York federal court.
According to the lawsuit, The Tribune claims its lawyers reached out to Perplexity in mid-October to inquire whether the AI search engine was utilizing its content. Perplexity’s legal team reportedly responded that they do not train models on The Tribune’s work, although the AI “may receive non-verbatim factual summaries.”
However, The Tribune’s lawyers contend that Perplexity is, in fact, delivering their content verbatim.
The newspaper’s legal team is focusing on Perplexity’s use of retrieval-augmented generation (RAG) as a core issue. RAG, a technique intended to mitigate “hallucinations” by restricting the language model to reliable, verified data sources, is allegedly being fueled by the newspaper’s content, which The Tribune claims was scraped without authorization.
Furthermore, The Tribune alleges that Perplexity’s Comet browser is circumventing the paper’s paywall to provide users with extensive summaries of the articles.
The Tribune is one of 17 news publications from MediaNews Group and Tribune Publishing that sued OpenAI and Microsoft over model training material: eight sued in April and another nine sued in November. Those suits are ongoing.
Creators have filed numerous lawsuits against model makers for using their work in model training. It remains to be seen how the courts will address the legal liabilities of Retrieval-Augmented Generation (RAG) systems.
Perplexity has not yet responded to the Chicago Tribune’s article about its lawsuit. The company is facing several other lawsuits, including one filed by Reddit in October and another from Dow Jones. Last month, while Amazon did not file a lawsuit, it did send a cease-and-desist letter regarding AI-driven shopping in its browser.