Brave built a 30-billion-page web index and licenses access to it through its Search API, which its own marketing says is "most commonly used in training and grounding" large language models. In early November 2026, major publishers took that business model to court.
2026-11-08 · by Moe Ameen
The business under scrutiny is simple to describe: Brave operates its own web search index — the complaint quotes Brave's own figure of "over 30 billion webpages, with millions of new pages indexed each day" — and sells access to it through the Brave Search API. That API exposes Web Search, News Search, LLM Context, and Answers endpoints, and a Brave guide quoted in the lawsuit says its "most common uses today are in training and grounding" large language models. Brave's About page has named Cohere, Mistral, Perplexity, and You.com as API users, and a November 2023 Brave blog post said Anthropic used the API to help train Claude 2. In other words, Brave positions a crawled copy of the open web as a licensable data product for AI builders — grounding (feeding live search results into a model's answer) and, on the appropriate plan terms, training.
On November 2, 2026, The New York Times Company sued Brave Software in the U.S. District Court for the Southern District of New York (Civil Action No. 26-cv-9549). Around the same time, Tribune Publishing's New York Daily News, Chicago Tribune, and Orlando Sentinel filed a separate case against Brave. The complaints allege that Brave built its index by crawling publishers' sites with a bot that, per Brave's own crawler documentation, does not advertise its own user agent — leaving site owners, the Times argues, with no clear way to identify or block it. The Times complaint lists Anthropic, Perplexity, Mistral, xAI, and Cohere as customers of the index; none of those companies is a defendant.
The legal claims go beyond copyright. The Times pleads direct, contributory, and vicarious copyright infringement; a DMCA anti-circumvention claim under 17 U.S.C. § 1201 tied to Brave's browser "Paywall blocker" feature, which Brave has promoted as a way to bypass paywalls; and an unjust-enrichment claim under New York law. The dispute is not Brave's first: in March 2025 Brave sued News Corp seeking a declaration that its indexing is fair use, and News Corp countersued. These are allegations; no court has ruled, and as of early November 2026 Brave had not filed a public response. Treat the case details as the plaintiffs' claims, and confirm current Search API plan names and terms on Brave's own pricing pages, which have changed more than once.
The useful reframe here is that AI grounding — the "most common use" Brave's own marketing cites — is the same mechanism behind getting your brand surfaced and cited in AI answers. You cannot control whether an index resells a crawl of your site, and you should not try to out-lawyer a publisher coalition. What you can control is being prolific and recognizable enough across the open web that the engines doing the grounding keep landing on your work with your name attached. That is [generative engine optimization](/glossary/generative-engine-optimization) and [answer engine optimization](/glossary/answer-engine-optimization), and the raw material it runs on is volume plus consistency — exactly the throughput problem [Kompozy](/) is built to solve. From one source it generates original [Blog Articles](/), text posts, and platform-native video and image formats, all governed by one [Persona Brief](/glossary/persona-brief) so a wide footprint still reads as a single, citable voice rather than scattered noise.
The second move is owning your distribution instead of depending on an index to represent you. Kompozy publishes that output across the eight social platforms plus blog and email, on [Autopilot](/glossary/autopilot) behind a per-post review gate, so your brand shows up natively wherever audiences and crawlers actually look — not only in whatever snippet a third-party API happens to resurface. The honest boundary: Kompozy does not license your content to AI firms, block an unnamed crawler, or manage your robots.txt — that is a legal-and-infrastructure question, and the related shift toward a shared training opt-out is covered in [the Google, Microsoft, and OpenAI robots.txt agreement](/news/shared-ai-training-opt-out). What Kompozy removes is the production ceiling that keeps most creators from publishing enough original, on-brand content to be the source worth grounding on in the first place.
Publishers allege Brave crawled their sites — with a bot that, per Brave's own documentation, does not advertise its own user agent — to build a web index of over 30 billion pages, then licensed access to that index through its Search API to AI companies for training and grounding large language models, all without permission or payment to the publishers. These are allegations; no court has ruled, and Brave had not publicly responded as of early November 2026.
The New York Times Company sued Brave Software on November 2, 2026 in the U.S. District Court for the Southern District of New York (Civil Action No. 26-cv-9549). Tribune Publishing's New York Daily News, Chicago Tribune, and Orlando Sentinel filed a separate case against Brave around the same time. The complaints plead copyright infringement, a DMCA anti-circumvention claim tied to Brave's paywall-blocker feature, and unjust enrichment.
The Times complaint lists Anthropic, Perplexity, Mistral, xAI, and Cohere as customers of the Brave index, and Brave's own materials have named Cohere, Mistral, Perplexity, and You.com as API users, with a 2023 post citing Anthropic's use in training Claude 2. None of these companies is a defendant — the suits target Brave as the alleged supplier, not the labs that bought access.
Blocking an unnamed crawler is hard, and litigation is a publisher-coalition move, not an individual one. The practical play is to lean into grounding on your own terms: publish enough original, on-brand content across platforms that AI engines cite you with attribution and audiences recognize your brand regardless of where a snippet surfaces. For the separate question of opting out of AI training, see the shared robots.txt opt-out that Google, Microsoft, and OpenAI agreed to honor.