# The New York Times sues Brave, calling its search index a back door for AI companies

> Publishers say Brave crawled their sites, sold the results to AI labs and let users skip paywalls. The suits name Anthropic, Perplexity, Mistral, xAI and Cohere as buyers.

- URL: https://dev.devisdead.ai/news/the-new-york-times-sues-brave-over-its-search-index
- Published: 2026-11-02T19:51:10Z
- Author: DevIsDead Newsroom
- Section: Policy
- Tags: brave, copyright, new-york-times, search, anthropic
- Originally reported by The Verge (https://www.theverge.com/tech/1024588/brave-lawsuit-ai-scraping-new-york-times).

The copyright fight over AI training has a new defendant, and it isn't a model maker. The New York Times sued Brave Software on Monday, and a second suit arrived the same day from three regional papers, the Orlando Sentinel, the Chicago Tribune and the New York Daily News, as [The Verge](https://www.theverge.com/tech/1024588/brave-lawsuit-ai-scraping-new-york-times)'s Jay Peters reported, citing Reuters.

The complaints go after two things. The first is Brave's search index: according to the publishers, Brave sells AI companies paid access to fresh web content from it, which those companies then use for training, for answering user questions, or both. The suits list Anthropic, Perplexity, Mistral, xAI and Cohere among the buyers. The second is the browser, which advertises a paywall blocker that the complaint says works on hundreds of major publications.

## The middleman theory

Most AI copyright cases so far go after the lab that trained the model. This one goes after the supplier. In the publishers' telling, Brave never asked permission to crawl, made itself hard to block, never paid, and then resold what it collected. The attorney for the three regional papers summed it up: "Buying stolen goods secondhand doesn't make them legitimate."

If that theory holds, it reaches far past Brave. Plenty of AI products don't run their own crawlers. They buy search results or web data from someone else's API. A ruling that the reseller is liable would make "we just license an index" a much weaker defense for everyone downstream.

## What we don't know yet

Brave and most of the named AI companies hadn't responded when The Verge published. Perplexity did, with a jab at the Times' legal bills and its journalism rather than a rebuttal of the claims. Words like "stolen" and "laundering" are the publishers' framing; no court has tested any of it.

For developers, the practical point is about supply chains. If your product grounds answers with a third-party search API, it's worth knowing where that index came from and what the provider promises about how it was built. Provenance used to be a question for the model's training set. It's now a question for every API call that fetches the web.

**[Scoreboard](/scoreboard) update:** Humans +1. Publishers take the AI copyright fight to the companies that supply the data.

Stay undead.