The Times Sues Brave, Calling It an AI ‘Fence’

NEW YORK, Nov. 2 — TheThe New York Times sued Brave Software in federal court in Manhattan on Monday, accusing the maker of the Brave browser and search engine of copying its journalism into a search index and selling it to AI companies. The Daily News, the Chicago Tribune and the Orlando Sentinel filed a suit in largely the same words that day.

Both complaints name five AI companies among those Brave supplies: Anthropic, Perplexity, Mistral, xAI and Cohere. None is a defendant. “In this way, Brave acts both like a burglar that misappropriates the content of publishers, and an AI ‘fence’ that sells stolen content to third parties,” the Times’s complaint says.

Anthropic, whose Claude models are used to write this paper, is named first. For it, the complaints cite a 2023 Brave post, which they quote: Anthropic “used Brave’s Search API to train their latest model, Claude 2, on high-quality data to help optimize Claude’s outputs for retrieval augmented generation (RAG) queries.” For Cohere, Mistral and Perplexity they cite Brave’s own web pages; none of the passages they quote mentions xAI. Anthropic’s amended prospectus, filed with the S.E.C. on Monday, does not mention Brave.

The Times has also sued OpenAI and Perplexity, TechCrunch noted. This suit goes after a supplier: a search index that, by the complaints’ account, AI companies pay to use both to train their models and to ground their answers in current web pages. A 2023 Brave guide, quoted in the complaints, says the most common uses of its Search API “are in training and grounding several of the Web’s biggest large language models (LLM).” The complaints sue Brave alone, but describe its customers as “the knowing receivers of that stolen content, which they have used to build their trillion dollar businesses.”BraveFounded in 2015 by Brendan Eich and Brian Bondy, Brave is closely held and based in San Francisco. Its browser had close to 130 million monthly users on Sept. 30, the Times’s complaint says, citing Brave; it adds that third-party reports put Brave’s 2025 revenue past $100 million. Brave has raised about $250 million, TechCrunch reported, citing private-market trackers.

In this way, Brave acts both like a burglar that misappropriates the content of publishers, and an AI ‘fence’ that sells stolen content to third parties.

The Claims

The Times’s five counts allege direct, contributory and vicarious copyright infringement; circumvention of its paywall, and trafficking in a tool to do it, under the Digital Millennium Copyright Act; and unjust enrichment under New York law. The last concerns the Times’s allegation that Brave’s browser blocks its advertising and puts Brave’s own ads in its place; that count, the complaint says, “does not depend on any copying.” Brave says on its website that its ads “do not replace the ads that the Brave Browser blocks on web pages.” The Times seeks statutory and actual damages, restitution of profits and a permanent injunction, and asks for a jury trial. It names no figure.

The paywall count rests on Section 1201 of the act, under which a judge in the same court refused on Oct. 30 to dismiss YouTube creators’ suit against Runway. The Brave browser has a setting called “Paywall blocker,” which, the complaint says, Brave promotes as allowing users to “Bypass Paywalls.” One Reddit user, quoted in the complaint, wrote of switching the filter on, heading to the Times and finding that “OOH LA LA the paywall had gone.”

The index, as the complaints quote Brave’s own description, holds “over 30 billion webpages, with millions of new pages indexed each day.” Brave has never disclosed the name of the crawler that builds it, they say, so sites cannot block it. Brave says on its help pages that its crawler will not crawl a page that is closed to Googlebot, Google’s crawler, and that publishers can have pages delisted by marking them noindex. The Times’s complaint gives examples of Ask Brave, the company’s answer engine, reproducing paywalled Times articles at length on request: in one, about 87 percent of an article.Ask BraveThe AI answer feature of Brave Search, launched in 2025 as an updated and renamed version of a 2023 feature called Summarizer, according to the complaint. The Brave API’s “Answers” endpoint, it says, is the same service. A query to the API’s “LLM Context” endpoint for the title of a Times article about a park full of robots returned, the complaint says, “1,291 out of 1,295 words in the article, verbatim.” The other papers’ complaint says Ask Brave produced over 1,200 words of a Chicago Tribune article on local tax breaks for data centers near O’Hare.

The Times says it twice told Brave to stop, in letters to Brendan Eich, Brave’s chief executive, on Aug. 11, 2025, and April 13, 2026, and that Brave ignored both. The second warned of “enhanced statutory damages up to $150,000 per infringement.” The Times says it owns over 10 million registered, copyrighted works. The other papers’ complaint describes no such letters.

Brave’s Side

Brave did not immediately respond to a request for comment, TechCrunch reported on Monday, and the paper found no statement on its blog by early Tuesday. It has made its case before. In its own suit against News Corp, Brave has argued that indexing the publisher’s content to make it searchable, and giving users snippets and “high-level summaries” of it, is fair use, Reuters reported in July.The Other SuitBrave first sued News Corp, publisher of The Wall Street Journal and The New York Post, in federal court in San Francisco in March 2025, after a cease-and-desist letter; CourtListener shows that case closed that June. Brave sued again in May 2026, and News Corp countersued in July, accusing it of “flagrant theft,” Reuters reported.

Brave’s first complaint, filed in March 2025, set out the argument at greater length. “Decades of legal precedent and practice confirm that it is not copyright infringement to index website content to maintain a search engine,” it said. It said Brave “never accesses or indexes content that a provider places only behind a paywall, or for which a provider requires login and password information,” and that the “alternate snippets for AI” its API then gave chatbots were limited to 300 characters each, up to five per website. Monday’s complaints concern other products, Ask Brave and the LLM Context endpoint, and do not say what limits apply to them.

As of early Tuesday, the dockets in both New York cases showed requests for summonses and no date yet for Brave’s answer. The other three papers also filed a statement of relatedness on Monday; its text was not available on CourtListener.