The copyright fight over artificial intelligence has previously pointed at the companies building the models, and the Times has sued OpenAI and Perplexity over similar claims. On 2 November 2026, a group of publishers aimed somewhere else — at the company sitting between the open web and the AI companies, supplying the raw material.
The New York Times sued Brave Software, the company behind the Brave browser, in federal court in Manhattan. On the same day, the publishers of the Daily News, the Chicago Tribune and the Orlando Sentinel filed a separate, similar lawsuit in the Southern District of New York. Both suits allege that Brave copied publishers’ copyrighted work into a vast search index and then sold AI companies access to it. Neither suit has been decided by a court, and the claims in them remain allegations. Brave did not immediately respond to requests for comment when the suits were first reported.
What makes these cases different is not that scraping is alleged. It is who is accused of doing the supplying.
The middleman theory
The Times’ suit characterises Brave as a supplier, dealer and distributor of copyrighted content for AI companies. That framing matters because it reframes the browser company not as a passive piece of software, but as a wholesaler in a supply chain: one end gathers the web, the other end sells access to what was gathered.
According to the complaints, Brave built an index of more than 30 billion web pages. The publishers allege the index was assembled with a crawler whose name Brave has not disclosed, and that site owners have no way to block that specific crawler. A publisher that wants to keep its work out, the argument runs, cannot simply close a door it cannot find.
The complaints then describe what happened next. AI companies including Anthropic, Perplexity, Mistral, xAI and Cohere allegedly paid Brave for real-time access to content from the Brave Index, using it to train models, to ground answers to user queries, or both. That access came through the Brave Search API.
Those AI companies are not defendants in these suits.
What the publishers say they can prove
The Times’ filing identifies more than 10 million registered copyrighted works, spanning the Times itself, The Athletic, Cooking, Games and Wirecutter. The other plaintiffs’ complaint says their registered newspaper issues contain more than 2 million articles. Each complaint brings five counts: direct, contributory and vicarious copyright infringement, circumvention of technological access controls under the Digital Millennium Copyright Act, and unjust enrichment under New York law.
The filings also include reproduction tests. The Times says its tests produced extensive article text rather than only links or short snippets. In one example, Ask Brave returned more than 400 words verbatim after being given a paywalled article’s URL and then asked for the full text. A separate query to the API’s LLM Context endpoint returned 1,291 of 1,295 words from an article verbatim, according to the filing.
The suits go beyond the index. Both target a browser setting called Paywall blocker, alleging that the configured browser copies and displays paywalled articles without a login or subscription. The complaints also allege that Brave removes publisher advertising, including subscription prompts, and substitutes advertising sold through Brave Ads.
The publishers also say they warned Brave first. The Times describes two unanswered letters to Brave chief executive Brendan Eich: one on 11 August 2025 demanding an end to unauthorised access and use, and another on 13 April 2026 repeating the demand. The April 2026 letter argued that the alleged wilful infringement warranted enhanced statutory damages of up to $150,000 per infringement. A Times spokesperson has said the company believes Brave ignored repeated requests to stop its infringing conduct.
Why this case reaches further than the last ones
The Times has sued OpenAI and Perplexity over similar claims before. Suing a model builder is, by now, a familiar move. Suing the index that feeds many model builders at once is a broader claim, and that breadth is what gives this case its stakes.
If a court accepts the publishers’ distributor theory, the liability question moves upstream. It would no longer be only about what a chatbot does with text, but about whether collecting the web at scale, packaging it into an index, and selling real-time access to it is itself an infringing business — even when the buyer, not the seller, trains the model or answers the user. That is a question with consequences far beyond one browser company, because search indexes, data vendors and API providers all sit in that same middle layer.
Brave did not immediately respond to requests for comment when the suits were first reported, but the strongest argument against the publishers’ thesis is one of lawful tools and where liability sits: that indexing, ad-blocking and paywall handling may be defended as lawful user tools, and that any liability for training on content rests with the AI companies rather than the index provider. On that argument, a court could reject distributor liability even if some scraping occurred.
If a court rejects the distributor theory, that would not by itself resolve the other counts or the browser allegations, which would remain to be decided, and the broader question of how the middle layer is treated would stay open.
Either way, the case tests a distinction the AI era has blurred. There is a difference between a search engine that points readers to a publisher’s site and a service that, as the publishers allege, extracts the article, hands the text to a model, and lets the reader never visit the site at all. The complaints argue Brave crossed that line in several places at once: in the index, in the API, in Ask Brave’s answers, and in the browser itself.
What to watch
The first thing to watch is Brave’s own answer. So far it has not publicly responded to the suits in the initial reporting, and its defence — on the crawler, on the index, on the paywall feature and on the advertising claims — will shape whether this becomes a quick settlement, a long doctrinal fight, or something in between.
The second is whether the AI companies named in the complaints change how they source content while the case runs. They are not defendants here, but their data pipelines are now described, in detail, in a federal complaint. Whether that description leads any of them to re-examine how they source content is an open question.
The third is the remedy. The Times seeks an injunction and damages, and both complaints demand jury trials for eligible claims. What scope an injunction might have, and what it would mean for Brave’s API customers, remains an open question.
Brave was founded in 2015 and has raised roughly $250 million, according to private-market trackers cited in the initial reporting. It is described in the initial reporting as a privacy-focused browser. The publishers’ suits tell a different story about the same company: not a shield for readers, but a supplier for machines. A court will now decide which story the evidence supports. Until then, the most consequential defendant in the AI copyright wars may not be the company that built the chatbot at all. It may be the company that fed it.
Editor’s note: The complaints quote Brave’s earlier figure of more than 30 billion pages for its index. Brave’s February 2026 Search API update puts the index at 40 billion pages, the figure used in our November 2 report. Both figures are therefore retained with their sources and dates, rather than one being replaced.
