What did the New York Times sue Brave over?
The New York Times Company sued Brave Software on 2 November 2026 in the Southern District of New York. The 72-page complaint accuses the browser maker of copying Times journalism, using it in its own AI products, and selling access to it for AI companies. The case number is 1:26-cv-09549.
Hours later, the publishers of the New York Daily News, the Chicago Tribune and the Orlando Sentinel filed a second, similar complaint in the same court. That case is 1:26-cv-09554. Both ask for injunctions and damages, and both demand a jury trial.
The Times describes Brave as "a burglar that misappropriates the content of publishers" and an "AI fence" that sells stolen material onward. The wording is aggressive, but the facts behind it are ordinary product documentation. Brave runs a search index, licenses API access to it, and ships a browser with ad blocking and a paywall filter list.
The New York complaint brings five counts. Three are copyright claims: direct, contributory and vicarious infringement. The fourth alleges circumvention under the Digital Millennium Copyright Act. The fifth is unjust enrichment under New York law, focused on advertising rather than copying.
The copyright count lists more than 10 million registered Times works in an exhibit. The three Alden-owned papers say their complaint covers more than two million registered articles, according to The Next Web. Neither filing puts a dollar figure on the claim.
The second complaint was filed by Steven Lieberman, an attorney whose clients include MediaNews Group and Tribune Publishing, both owned by Alden Global Capital. He said AI companies are "desperate to avoid getting caught red-handed stealing news" and accused Brave of laundering "pirated reporting for the AI industry".
What is the Brave Index, and who buys it?
Brave built its own web index rather than renting results from Google or Bing. The company says the index covers "over 30 billion pages", with millions of new pages added each day. It also says Brave Search handles more than 50 million queries daily, and that about a third of them trigger an AI-written summary.
That index is the commercial product at the centre of both lawsuits. Brave sells query access through the Brave Search API, including endpoints for news and for what it calls LLM Context. The API page markets the company as having zero dependence on Google or Bing and being trusted by "some of the world's top AI labs".
Brave's own About page names four customers: Cohere, Mistral, Perplexity and You.com. The Times complaint names five different customers, adding Anthropic and xAI, and says they received paid access to real-time content from the Brave Index. You.com appears on Brave's site but not in the complaint.
None of those AI companies is a defendant. That is a deliberate choice, and it shapes the legal question. The publishers are not asking a court to rule on whether training a model on news is fair use. They are asking whether an intermediary may copy articles, package them for AI customers, and sell the results.
The Times does not argue that indexing is unlawful in itself. Its complaint says the paper lets search engines index its pages so results can link back to its site. It says search engines have historically returned links and limited snippets rather than using full articles to keep users on their own platforms. The claim is that Brave went beyond that bargain.
Brave has described this business in its own words for years. A November 2023 blog post on the Leo assistant said Anthropic "used Brave's Search API to train their latest model, Claude 2" to improve retrieval-augmented answers. The Times complaint quotes that line directly.
The scale is substantial but not huge by platform standards. Brave's transparency feed reports 128.3 million monthly and 54.6 million daily browser users as of 30 September 2026. The complaint, citing third-party reports, says Brave's 2025 revenue passed $100 million, up from about $52 million in 2024. TechCrunch reports that Brave has raised roughly $250 million since 2015.
Does Brave's browser really bypass paywalls?
The complaints point to a browser setting called Paywall blocker. They say that, once activated, it fetches a paywalled article and displays the text without a login. The Times quotes a Reddit user who switched it on and then read the paper "without interruption and without needing to set up an account".
Brave's public filter catalogue supports the existence of the feature. It lists an entry titled "Paywall blocker" with the description "Bypass Paywalls Clean Filters". The list is drawn from the Bypass Paywalls Clean project and sits among 61 other filter lists in the catalogue.
That catalogue detail matters for the defence. A filter list is something a user enables, not a feature that necessarily ships switched on. The complaints describe a setting that acts "when activated", and they do not say whether new installations start with it enabled.
Brave does not maintain the underlying rules. Its catalogue entry credits a third-party project, Bypass Paywalls Clean, hosted on the code platform GitFlic, as the source of the filters. That could matter if the case turns on who designed the bypass and what it was designed to do.
The paywall feature is the basis of the DMCA count. That claim is separate from copyright. It alleges that Brave circumvented technological measures controlling access to the articles, and that the bypass tool has little commercial purpose beyond circumvention.
The crawler is the other half of the access argument. Brave's help page says its crawler "does not advertise a differentiated user agent". Brave explains that a distinct agent would invite blocking by sites that allow only Google to crawl them.
The same page says robots.txt is not used to prevent indexing. Brave tells site owners to use a noindex directive instead, and says it will not crawl a page that Googlebot cannot crawl. For publishers, that means the usual opt-out file does not stop Brave. The Times says Brave has taken steps that prevent it from blocking access.
Why sue Brave instead of Anthropic or Perplexity?
Suing an index seller is a different strategy from suing a model developer. A training case turns on fair use, market harm and the purpose of the copying. An index case turns on something more concrete: whether Brave made copies, whether it distributed them, and whether it charged for them.
It also concentrates the dispute on one defendant. The complaints assert that five AI companies bought access. If that theory survives, every customer of the API becomes either a licensee or a potential co-defendant, without the publishers having to litigate five separate fair-use defences at once.
Publishers have already gone after the labs directly. The Times sued OpenAI and Microsoft in late 2023 and has also sued Perplexity. The new filings add the supply layer beneath those products, which is where the publishers say the copying actually happens.
The complaint also builds a willfulness case from correspondence. Times counsel Carl Mazurek wrote to Eich on 11 August 2025 asking Brave to stop. Outside counsel followed on 13 April 2026 with a letter warning of enhanced statutory damages. The complaint says Brave ignored both letters.
The advertising claim widens the case beyond data. The Times says Brave blocks its ads by default, including prompts to subscribe, then sells its own ads to the same audience. One exhibit shows Brave stripping a Times banner that offered "$1/week for your first six months". Brave Rewards shares some ad revenue with users. The complaint calls the arrangement unjust enrichment because Brave profits from readership the paper paid to build.
Perplexity, named as a customer but not a defendant, is not treating the suit as an existential threat. Its chief communications officer, Jesse Dwyer, told The Verge: "The New York Times is overpaying its attorneys and underpaying its journalists." That quote captures how the AI side views the filing.
What does Brave say?
Brave has not responded to the substance of either complaint. TechCrunch and The Verge both reported that the company did not reply to requests for comment before publication.
Its existing documentation sets out the outline of a defence. Brave argues that it runs an independent index, that it follows the same crawl rules as Googlebot, that publishers can delist pages with noindex, and that it offers a copyright takedown channel. Each of those points answers a specific allegation.
The company has also fought this ground before. In March 2025, Brave sued News Corp and asked a court to declare that its indexing was fair use. That case was voluntarily dismissed and then refiled by Brave in May 2026. News Corp answered with counterclaims in July 2026, accusing Brave of scraping and selling its journalism.
A ruling in Brave's favour would reach well beyond one browser. It would legitimise a market in which search indexes are licensed to AI developers as a substitute for crawling. A ruling against Brave would make every search API a potential liability, and would push AI companies back toward direct deals with publishers.
Brave's privacy pitch sits awkwardly beside the allegations. The company markets itself as an alternative to ad-funded platforms, and its browser blocks trackers and ads by default. The same design removes publisher ads and subscription prompts, which is why the Times treats ad substitution as part of the harm rather than a side issue.
The company's corporate profile is part of the story. Brave was founded in 2015 by Eich and Brian Bondy and markets itself on privacy. The Times complaint cites industry observers who value the company at about $980 million, and notes that it discloses little financial detail because it is closely held.
What are the weak points in the publishers' case?
Nothing here has been tested in court. Brave has not filed an answer, and no judge has ruled on whether a search index is a copy for these purposes. The complaints are allegations, drafted to survive a motion to dismiss rather than to prove the case.
The most vivid examples are also the most controlled. The Times says its API test returned 1,291 of 1,295 words from one article, and that Ask Brave produced more than 400 words and nearly 700 words in two exchanges. Those are results the plaintiffs elicited with prompts they wrote.
The filings do not show how often ordinary users ask for full articles. They also do not identify which endpoints each named AI company uses, or show that a specific Times article reached a specific lab. The link between the named customers and the alleged copying rests largely on Brave's marketing claims.
Fair use remains the central uncertainty. Search engines have long relied on indexing and snippet display, and Brave's earlier declaratory action shows that it intends to make that argument. A court would also have to decide whether a paywall is an effective access control and whether a filter list circumvents it, rather than merely ignoring it.
Brave's public position also supplies a rebuttal. If its crawler follows the same rules as Googlebot and honours noindex, the company can argue that publishers keep a working opt-out. The Times would then have to show that its chosen controls were effective and that Brave defeated them. That is a higher bar than showing that copying happened.
Damages are open-ended. The Times' April 2026 letter, quoted in the complaint, warns of statutory damages up to $150,000 per work for willful infringement. Across millions of registered works, that ceiling would be enormous, but courts have discretion, and maximum awards are rare.
What happens next?
The immediate step is Brave's response. A motion to dismiss is likely, and it will probably lean on fair use, on the limits of the DMCA's anti-circumvention provisions, and on the argument that the complaints lump separate products together.
Watch whether the publishers try to add the AI companies as defendants. Adding them would turn a supply dispute into a training dispute, and would force the fair-use question into the open much sooner.
Watch the News Corp case as well. It covers similar conduct, has been running longer, and may produce the first ruling on whether Brave's indexing is lawful. Any decision there will shape how the New York cases are argued.
Watch Brave's crawler policy too. If the company publishes a distinct user agent or a publisher opt-out that works without noindex, the claim that sites cannot block it becomes harder to sustain. That would be a product change with legal consequences.
One near-term signal is commercial rather than legal. If AI developers start demanding indemnities or direct publisher licences for search grounding, the market may move before any court rules. Brave's API customers would then face a choice between a cheaper index and a cleaner chain of title.
No hearing date has been announced. Federal cases of this size typically take months to reach a ruling on a motion to dismiss, and longer to reach trial. For now, the practical effect is uncertainty for every company that buys search-index access rather than crawling the web itself.