The Den Reviewthe briefing for executives who run AI
Wednesday, November 11, 2026

The Denominator

Subscribe
Trending

Model Labs · AI Copyright Suits

Lawsuit Calls Brave an 'AI Fence' Selling Scraped News to Anthropic and Perplexity

The complaint calls those AI buyers 'knowing receivers of that stolen content.'

· 2 min read

A neat stack of folded newspapers, blank on top, rests on a dark marble desk in a dim, wood-paneled room.
Photograph: AI-generated

Four newspaper publishers sued Brave Software, saying its paid search feed returned 1,291 of the 1,295 words in one of their articles.

Anthropic, Perplexity, Mistral, xAI and Cohere are named as the feed’s customers. The publishers ask a court for damages and a permanent injunction. If Brave had to carve publishers’ content out of its index, the products built on it could change with it, which makes the supplier’s sourcing a question for every buyer.

What do the complaints allege?

The two complaints, filed in federal court in Manhattan, say Brave acts “both like a burglar that misappropriates the content of publishers, and an AI ‘fence’ that sells stolen content to third parties.” The publishers bring five claims, among them direct, contributory and vicarious copyright infringement and circumvention of paywalls.

Brave’s index holds more than 30 billion web pages, and it sells access through a feed that returns search results and page text to other companies. A Brave guide quoted in the filings says the feed’s most common uses are training AI models and “grounding” them, which means supplying live web results for a model to answer from. Brave is private and does not have to disclose its revenue, the complaints say, and they rely on outside estimates of more than $100 million for 2025.

How has Brave responded?

Brave did not respond to requests for comment, according to a published report. Its position in a separate court fight with News Corp is that indexing articles and showing short excerpts and summaries is fair use.

Brave’s help page says its crawler carries no distinct name because many sites admit only Google, and that the crawler skips any page Googlebot is barred from. The complaints say the missing name leaves publishers no way to block the crawler. One of the publishers adds that Brave ignored its letters of August 2025 and April 2026.

What changes for buyers?

Earlier publisher suits mostly named model makers and answer engines. These name the supplier the publishers say crawls the web and resells what it finds, and the five customers appear only as buyers.

The publishers want an injunction, and one that forced Brave to drop publishers’ content or domains could change what its customers’ products can retrieve. Whether a buyer’s contract covers that is the question the suits put to supply deals.

Could the company name the search suppliers behind its AI products and say what its contract does if a court cuts their data?

NVIDIAMicrosoftOpenAIAnthropicGoogleAWSJPMorganCapital OneAT&TCharterComcastT‑MobileUnitedHealthCVS HealthTravelersWalmartSalesforceDatabricksSnowflakeCoreWeaveStripeAccenture