THREAT ASSESSMENT: HIGH. The legal fight over AI training data has a new kind of target. It isn’t a model maker this time. It’s the company that allegedly supplies them.
The New York Times and a group of other publications have filed lawsuits against Brave Software, the company behind the Brave web browser, The Verge AI reports, citing Reuters. The complaints accuse Brave of two things: scraping news sites to feed AI companies, and letting users get around paywalls.
📍 Situation Report
Here’s what the publishers are alleging:
- The index is the product. Brave allegedly makes money from its index of scraped websites by selling AI companies paid access to “real-time online content from the Brave Index.”
- The buyers are big names. According to the complaint, Brave provided “stolen copyrighted content” to “many” AI companies, including Anthropic, Perplexity, Mistral, xAI, and Cohere.
- There are two uses. AI firms allegedly use the index to train models, to answer specific user queries, or both.
- The paywall blocker. Brave promotes “hundreds of top publishers” as reachable through its “Paywall blocker” browser feature.
- No permission, no payment. “Brave has never asked for the Publishers’ permission to access their websites, much less to copy the material and provide it to users or other AI companies,” the complaint says. It adds that Brave took steps to stop publishers from blocking its access.
There are two separate filings. The New York Times filed one. The New York Daily News, the Chicago Tribune, and the Orlando Sentinel filed the other.
🎯 Who Said What
The Times didn’t hold back. “Brave has flouted established copyright law and ignored repeated requests by The Times to stop its infringing conduct,” spokesperson Graham James told The Verge.
Steven Lieberman, the attorney for the other three papers, went further.
“Buying stolen goods secondhand doesn’t make them legitimate,” he said. “Brave must be held responsible for laundering pirated reporting for the AI industry.”
Perplexity responded with a jab. Its communications chief, Jesse Dwyer, said the Times “is overpaying its attorneys and underpaying its journalists.” Brave, Anthropic, Mistral, and Cohere hadn’t commented at the time of reporting.
🧭 Why This Matters
What stands out here is the strategy. Publishers have spent years suing model builders directly. The Times’ case against OpenAI and Microsoft, filed in late 2023, is the best-known example. This lawsuit goes after the middle of the supply chain instead.
That matters because Brave runs one of the few independent web indexes outside Google and Microsoft’s Bing. Its search API has become a common way for AI products to ground answers in fresh web content. If a court decides that selling access to scraped copyrighted material is infringement, the theory reaches well past one browser company.
The “laundering” framing is the key legal move. Publishers are arguing that an AI company can’t clean up its data by buying it from someone else. If that argument holds up, buyers carry the risk too.
⚠️ Tactical Implications
If you build with AI, watch these pressure points:
- Search and grounding APIs are now legal exposure. If your product pulls live web content through a third-party index, you’ll want to know how that index was built.
- Retrieval is in scope, not just training. The complaint targets content used to “respond to specific user queries.” RAG (retrieval-augmented generation, where a model looks up documents before answering) isn’t a safe harbor.
- Licensing deals get more valuable. Every lawsuit like this pushes AI firms toward paid publisher agreements. Expect more of them.
- Browser features are fair game. A paywall bypass tool sitting right next to an AI data business makes the publishers’ case easier to tell a jury.
- Vendor due diligence is coming. Procurement teams will start asking data suppliers where their content comes from and whether they have rights to it.
🔭 Outlook
Brave’s response will set the tone. It may argue fair use, or it may say that indexing public pages is ordinary search engine behavior. Either way, the case tests a question the industry has mostly dodged: who’s liable when copyrighted content passes through several hands before it reaches a model?
The answer could change how every AI company sources the web. Full details are in the original report from The Verge AI.