The New York Times sues Brave for selling its articles to AI companies

Most people know Brave as the privacy browser that blocks ads. According to two lawsuits filed in Manhattan on November 2, 2026, Brave has quietly become something else: a wholesale supplier of news articles to the AI industry. The New York Times calls the company "a burglar that misappropriates the content of publishers, and an AI 'fence' that sells stolen content to third parties."

The Times complaint was filed in the US District Court for the Southern District of New York as case 26-cv-9549. On the same day, the New York Daily News, the Chicago Tribune and the Orlando Sentinel filed a nearly identical complaint as case 26-cv-9554, using the same law firm, Rothwell, Figg, Ernst & Manbeck. Both ask for a jury trial, damages and a permanent injunction.

What makes these cases different from the earlier wave of AI copyright suits is the target. Brave is not a frontier lab. It does not sell a famous chatbot. The publishers argue that it sits one step earlier in the chain, collecting the web into an index and then renting that index to the companies that build the models. If that theory holds, the legal risk of AI training and AI search could move from the labs to the data suppliers behind them, and back again.

What the publishers say Brave did

The complaints describe a business that changed direction around 2023. Brave launched Brave Search as an independent search engine with its own index instead of renting results from Google or Bing. According to Brave's own marketing quoted in the filing, that index now holds "over 30 billion webpages, with millions of new pages indexed each day."

The Times says Brave never told website owners the name of the crawler that built this index. That detail matters more than it sounds. Publishers control crawlers mainly through the robots.txt file, where they can allow Googlebot and block, say, GPTBot. If a crawler has no published name, there is nothing to block. The complaint says Brave "crawled and copied content from the over 30 billion webpages in the Brave Index without identifying itself or providing website owners with any mechanism to block its crawler."

According to the filing, Brave then found that search alone was not paying off and began to monetize the index in two ways. The first is Ask Brave, an AI answer feature inside Brave Search that sends pages from the index to a large language model and writes a direct answer. The second is the Brave Search API, which other companies can pay to query. Brave markets parts of it as "AI Grounding" and offers endpoints called News Search and LLM Context. The Times quotes Brave's documentation saying the LLM Context endpoint "returns extracted page chunks and source metadata in a shape and length that satisfies the needs of an LLM without need for further scraping."

The third product is the browser itself. The complaints focus on a filter setting called "Paywall blocker", which Brave describes as a way to "Bypass Paywalls", and on Brave's ad blocking, which removes publishers' ads and lets Brave show its own ads through the Brave Ads and Brave Rewards programs.

Who the alleged buyers are

The most explosive part of the filings is the list of customers. The Times names Anthropic, Perplexity, Mistral, xAI and Cohere as prominent AI companies that received content through Brave. Much of the evidence comes from Brave's own words. Its API page says the service is "trusted by the world's top AI labs", its About page lists "Cohere, Mistral, Perplexity, and You.com", and a 2026 blog post about a Snowflake integration says the API serves "most of the top 10 AI LLMs with real time Web search data."

The complaint also cites Brave's November 2023 blog post about its Leo assistant, which stated that "Anthropic used Brave's Search API to train their latest model, Claude 2." The filing adds that Meta trained Llama 3 to call the Brave API for questions about recent events.

None of those AI companies is a defendant. But the language is clearly aimed at them as well. The Times says the third parties are "serving as the knowing receivers of that stolen content, which they have used to build their trillion dollar businesses." Steven Lieberman of Rothwell Figg, who signed both complaints, told The Verge: "Buying stolen goods secondhand doesn't make them legitimate."

Perplexity answered fast and sharply. Its spokesperson Jesse Dwyer said the Times "is overpaying its attorneys and underpaying its journalists." Brave had not responded publicly as of November 3, according to PPC Land.

The evidence: whole articles on request

Both complaints are built around screenshots. In one Times example, a user gave Ask Brave nothing but the link to a paywalled article. Ask Brave summarized it and then offered: "would you like me to pull up the full article." After a simple "yes please", it printed more than 400 words of the story verbatim. In another, about a Times investigation into Elon Musk's midterm spending, Ask Brave replied "Here's the full article" and produced nearly 700 words. A third example shows Ask Brave delivering about 87 percent of a paywalled article while adding that this was only "the full publicly accessible portion."

The API examples are even starker. A query to the LLM Context endpoint with the headline of a Times article about a robot park returned 1,291 of the article's 1,295 words. In other words, the paid API delivered practically the entire piece. A News Search query returned more than a third of another Times story verbatim.

The Tribune papers show the same pattern. Ask Brave reproduced more than 1,200 words of a Chicago Tribune article on tax breaks for data centers near O'Hare, more than 1,000 words of another Tribune story, and nearly 600 words of a Daily News article on Central Park horse carriages. The API returned nearly 350 words of a Daily News piece and more than 250 words of a Tribune article.

These examples are allegations, chosen by the plaintiffs to make their case, and Brave has not yet answered in court. Still, the pattern is the core of the argument. A search engine that shows a link and a snippet sends readers to the publisher. An answer engine that prints the whole article removes the reason to visit. The complaints cite a study finding that AI bots send on average 95.7 percent less referral traffic than traditional Google search.

Five counts, one theory

Each complaint has five counts. The first three are copyright claims: direct infringement for building the index and serving the text, contributory infringement for helping users and AI customers copy it, and vicarious infringement for profiting from that copying while controlling the tools.

Count four is under the Digital Millennium Copyright Act, Section 1201, which forbids circumventing technical measures that control access to a work, and forbids offering tools designed for that purpose. Here the publishers point to the Paywall blocker setting. They also cite posts by Brave employees that explain how to remove the Times "paywall nag" and a Reddit user who wrote that after turning the filter on, the Times paywall "had gone."

Count five is unjust enrichment under New York law, and it does not depend on copying at all. It targets Brave's ad model: the browser blocks the ads that pay for the article and shows Brave's own ads instead, so the money, the papers argue, moves from the newsroom to the browser.

The Times also says it warned Brave twice. On August 11, 2025, its intellectual property counsel wrote to Brave chief executive Brendan Eich asking the company to stop accessing and using Times content. A second letter on April 13, 2026, warned that ignoring the first one showed "reckless disregard" and made the infringement willful, which can raise statutory damages to $150,000 per work. According to the complaint, Brave did not respond to either letter. With more than 10 million registered works, the theoretical exposure is enormous, even though real awards are usually far lower.

How big Brave really is

Brave is private and publishes few numbers, so the complaint relies on Brave's transparency page and third party estimates. It says the browser had close to 130 million monthly users and more than 50 million daily users as of September 30, 2026. Brave Search answers more than 50 million queries a day, and about a third of them lead to an AI generated answer. The API handles billions of calls each week. Revenue reportedly passed $100 million in 2025, up from about $52 million in 2024, and industry observers value the company at about $980 million.

That is small next to OpenAI or Google, which is exactly why the case is interesting. Brave has raised only around $250 million in total, according to private market trackers cited by TechCrunch. A long copyright fight with statutory damages on the table is a much bigger threat to a company of that size than to a lab with tens of billions in funding.

Brave's likely defense

Brave has not filed an answer yet, but its past positions hint at the arguments. When a newspaper trade group attacked its ad replacement plans in 2016, Eich dismissed the claims as "false assertions" and pointed out that Brave planned to share up to 70 percent of ad revenue with publishers. Expect a similar line on the ad and paywall counts: the browser runs on the user's own device, and users choose what it blocks. Brave will likely also say that a search index is a classic fair use, as courts have long held for search engines, and that what its customers do with API results is their responsibility.

The weak point for Brave is the verbatim output. Fair use for search has always rested on showing short snippets and links. Printing 1,291 of 1,295 words in response to a paid API call, or offering to "pull up the full article" from behind a paywall, looks less like search and more like distribution. How the court treats that line will matter far beyond Brave.

What it means for you

If you use the Brave browser, nothing changes today. A lawsuit is not an injunction, and the browser, Brave Search and Leo keep working. If the publishers win, the first things at risk are the Paywall blocker filter and the way Ask Brave answers requests for full articles.

If you build AI products on top of the Brave Search API, this case deserves a careful read. The publishers are openly testing whether buying content through an intermediary protects the buyer. The Times also has suits running against OpenAI and Microsoft and against Perplexity, while it has licensed its content to others, such as Amazon. The message to the industry is clear: license the news, or expect to be asked where it came from.

And if you run a website, the complaint highlights a practical gap. You can only block crawlers you can name. Brave's index was built, the papers say, by a bot that never identified itself. That will add pressure on all search and AI companies to publish crawler names and respect opt outs.

The bigger picture

The first round of AI copyright cases targeted the labs that trained the models. This round goes after the plumbing: the search indexes and grounding APIs that feed live news into chatbots and agents. As more AI answers rely on real time retrieval rather than training data, whoever owns the index becomes the gatekeeper, and now also the defendant. The outcome in Manhattan will help decide whether AI search pays the newsrooms it reads from or keeps running on content that nobody licensed.

Sources: The New York Times Company v. Brave Software, complaint, S.D.N.Y. 26-cv-9549, Chicago Tribune, Daily News and Orlando Sentinel v. Brave Software, complaint, S.D.N.Y. 26-cv-9554, The Verge, TechCrunch, PPC Land.