Publishers Sue Brave Over AI Training Data and Paywall Bypass
The New York Times and other publishers have sued Brave, the maker of the Brave web browser, over alleged unlawful scraping of content to train AI models and over alleged enabling of users to bypass paywalls. The filings were reported on November 2, 2026, according to The Verge.
The Times complaint, published as a PDF by the paper's corporate assets host, centers on the claim that Brave copied protected news content without authorization and used it for model training. The paywall allegation is separate. It concerns the browser's role as a means of reaching articles that publishers place behind subscription barriers.
A second complaint came from the companies that operate the New York Daily News, the Chicago Tribune and the Orlando Sentinel. It names Brave Software as defendant and makes a more specific claim about commercial structure. According to the complaint, Brave monetized its index of scraped websites by offering AI companies paid access to real-time online content from the Brave Index. The complaint says that content was licensed for training models or for answering user queries, which is a different use from simple crawling for search.
The same complaint alleges that Brave supplied stolen copyrighted content to AI companies including Anthropic, Perplexity, Mistral, xAI and Cohere. These are allegations. No court has ruled on the merits.
On the paywall question, The Verge reports that Brave promotes hundreds of top publishers as accessible through the browser's Paywall blocker feature. That marketing detail matters to the legal theory because it links the product's stated functionality to the publishers' subscription revenue. It is the publishers' reading of the feature, presented as part of their case.
For practitioners, the distinction between the two lawsuits is the one to track. Scraping for training is a question about the copying and ingestion of copyrighted works, which courts have been working through in multiple AI cases. A paywall bypass claim turns on how a browser renders or retrieves content that a site has chosen to gate, and whether the browser's features are the mechanism. The Daily News, Tribune and Sentinel complaint adds a third element: a downstream licensing market in which an intermediary sells live or near-live web content to model developers and to retrieval-augmented generation (RAG) systems, which pull fresh documents into a model's answers. If that characterization holds, the question shifts from what a crawler collected to who profited from redistributing it.
The broader context here is that publishers have increasingly treated the web index itself as an asset, not only the articles on individual sites. Robots.txt, the plain-text file where a site states which bots may crawl which pages, and site terms of service were designed for a crawl-and-link bargain. They offer limited control once content has been indexed, cached and sold on. Whether these complaints succeed will depend on facts the court has yet to test, including what was copied, how it was delivered to each AI company, and whether any licensing terms existed.
In my view, the cases will be read closely by anyone who builds or operates a browser, search index or retrieval pipeline, because they press on the seam between user agent and data broker. A browser that fetches pages on a user's behalf sits in a different legal position from an index that sells access to its contents in bulk. Where a given product falls along that line is the question the complaints put before the courts.
There is a practical upside for the wider ecosystem if the cases produce clear rules. Publishers, model developers and browser vendors would gain predictable terms for licensing, opt-out signaling and paywall handling, which is cheaper than litigating each dispute from scratch. Until then, engineers building crawlers or content-aware features should expect their data provenance to be examined more closely than it was a few years ago.
