The defendant sells the web to AI companies
Most publisher copyright cases name a model maker. These two name a supplier. Brave runs a search index that the complaints, quoting Brave itself, put at more than 30 billion web pages. Through its Search API, companies pay for endpoints with names like LLM Context and Answers, which return extracted page text and source metadata ready to feed a model, for training or for grounding answers to user questions. Brave's own marketing, quoted in the Times complaint, describes the most common uses as "training and grounding several of the Web's biggest large language models", and a 2023 Brave post cited in the filing says Anthropic used the API to train Claude 2.
The customer list is not a list of defendants, and it is not proof that every named company trained a model on these publishers' articles. The complaints identify Anthropic, Perplexity, Mistral, xAI and Cohere as users or customers of Brave's API, but do not establish which endpoints each company used or whether each received the specific works at issue. Training and retrieval are also different uses: an API can supply data for model training or retrieve documents to ground an answer at the moment a user asks a question.
The publishers' argument is that the index was built without permission and without a crawler-specific opt-out. Both complaints allege Brave has never disclosed a separate identity for the crawler that built its index, making it impossible for site owners to block Brave independently through robots.txt. Brave's published documentation says its crawler follows Googlebot's restrictions and that publishers can use a noindex directive to remove pages from Brave Search. The dispute is therefore not whether any exclusion mechanism exists, but whether publishers can identify and block Brave's access without also restricting Googlebot.
The tests are the strongest part of the filing
Copyright suits against AI companies usually argue about training, where the copying is hard to show. These filings lead with output a judge can read. The Times says it gave Ask Brave the URL of a paywalled article, was offered the full article, and received more than 400 words word for word.
The second complaint describes additional examples involving the three newspapers, including Ask Brave reproducing more than 1,200 words from a Chicago Tribune investigation. Separately, the Times complaint says an LLM Context query returned 1,291 of a 1,295-word article verbatim, while the second complaint describes an Orlando Sentinel article returned in full through the API. These are plaintiffs' own tests, not independent reproductions.
Those are the plaintiffs' tests, run by the plaintiffs, and Brave has not answered them in court or in the press coverage read for this piece. But they change the shape of the dispute. A search index that returns links and short excerpts is one legal argument. A paid feed that can return an article almost in full, on request, is a different one, and it is aimed at the product AI companies actually buy.
The browser is in the case too
The suits do not stop at the index. Both target a Brave browser setting described in the filings as a "Paywall blocker", which the complaints say copies and displays paywalled articles without a login. The Times quotes Brave describing the feature as a way to "Bypass Paywalls". The complaints also allege that Brave blocks publishers' advertising and monetizes its own advertising business, depriving publishers of revenue. Their screenshots show publisher advertising and subscription prompts disappearing when the same pages are opened in Brave. They do not, however, show Brave advertisements appearing in those exact spaces. The plaintiffs' separate unjust-enrichment claim argues that Brave benefits financially from the displacement of publishers' advertising inventory. The Times complaint alleges that Brave made more than $100 million by copying and reselling publishers' work. That is the plaintiffs' characterization, not an independently established figure for revenue from the index or API. The underlying revenue estimate cited in the reporting concerns Brave's overall business, and the available material does not break it down by product or establish how much was attributable to the publishers' content.
There is an important distinction between the two browser claims. The Paywall blocker allegation concerns access to subscription-protected articles. The separate advertising claim concerns Brave blocking publishers' ads and benefiting from its own advertising business. The complaints argue that the two practices damage publishers in different ways, and the unjust-enrichment count is specifically about the advertising revenue.
Context matters here. Brave is already litigating with News Corp, having asked a California court to rule its indexing fair use before News Corp countersued in July. This is not a defendant caught by surprise. It is a business model its owner has been defending, now challenged on the index, the API and the browser at the same time.
The Times complaint describes two warning letters to Brave CEO Brendan Eich, dated August 11, 2025 and April 13, 2026. The second group of publishers does not describe equivalent prior notices in its complaint. The distinction may matter to the Times' allegation that Brave knowingly continued the conduct after receiving notice.
What the case decides for everyone else
Anthropic, Perplexity, Mistral, xAI and Cohere are named in the complaints as buyers of Brave's index access. None of them is a defendant, and only Perplexity responded in the coverage read for this piece, dismissing the suit. The lawyer for the second group of publishers stated the theory plainly: AI companies that buy content from a supplier are still exposed if the supplier had no right to sell it.
A ruling against Brave could put a price on that assumption. The consequences would likely extend beyond one search provider, prompting AI companies to examine the provenance, licensing representations and indemnities attached to third-party search and grounding data. The extent of that effect would depend on what the court actually decides.
Sources
- [1] The New York Times Company v. Brave Software, Inc., complaint, U.S. District Court for the Southern District of New York, 26-cv-9549, filed November 2, 2026, via CourtListener (primary source, read first-hand in full before publication; the supplier and "fence" framing and the customer list, the index of "over 30 billion webpages" with millions of new pages indexed each day, the undisclosed crawler, the "Paywall blocker" setting Brave promotes as allowing users to "Bypass Paywalls", the 1,291-of-1,295-words LLM Context test, the letters of August 11, 2025 and April 13, 2026 with the $150,000-per-infringement figure, the "more than $100 million" allegation, and the filing's own statement that third-party reports put Brave's total 2025 revenue above $100 million)Read source
- [2] PPC Land, "New York Times and three papers sue Brave over AI answers and ad blocking", November 3, 2026 (detailed breakdown of both complaints, read first-hand; the second complaint's tests — more than 1,200 words of a Chicago Tribune investigation via Ask Brave and an Orlando Sentinel article returned in full via LLM Context — Brave's published crawler explanation, the advertising comparisons showing publisher material disappearing with no Brave ad in its place, the remedies sought, and the fact that neither complaint states a damages total)Read source
- [3] Daily News, LP, Chicago Tribune Company, LLC and Orlando Sentinel Communications Company, LLC v. Brave Software, Inc., complaint, S.D.N.Y., 26-cv-9554, filed November 2, 2026, via CourtListener (primary source; its document page could not be opened at the publication check, so this piece reports its contents from PPC Land's complaint-level breakdown above and Unite.AI's summary below)Read source
- [4] Unite.AI, "Brave Faces Two Publisher Lawsuits Over AI Use of News Content", November 2026 (complaint summary read first-hand; both filings, the index and crawler allegations, the reproduction tests, the paywall and advertising allegations, the letters, the five counts and the remedies sought)Read source
- [5] TechCrunch, "New York Times and other outlets sue web browser Brave over alleged content scraping", November 2, 2026 (read first-hand; both suits filed the same day in Manhattan federal court, the named customers, the unidentified crawler, the paywall feature and the ad-blocking practice)Read source
- [6] Madrobot, "New York Times sues Brave for selling its articles to AI firms", November 2, 2026 (read first-hand; both case numbers, the "supplier, dealer, and distributor" characterization as quoted from the complaints, the publishers' lawyer statement via The Next Web, the Times and Perplexity statements via The Verge, and the News Corp litigation context)Read source