The copyright fight over AI training data moved one step up the supply chain on Monday. The New York Times sued Brave Software in federal court in Manhattan, and the publishers of the New York Daily News, the Chicago Tribune and the Orlando Sentinel filed a similar suit the same day. Both complaints allege that Brave, the privacy-focused browser company founded in 2015, scraped the newspapers' articles and unlawfully sold access to them to artificial intelligence companies.
At the center is the index behind Brave Search. The Times says Brave built an index of more than 30 billion web pages with an unidentified crawler that publishers could not block, then distributed their work through its Brave Search API to customers the suits name as Anthropic, Perplexity, Mistral, xAI and Cohere. None of those AI companies is a defendant. The complaints call Brave a supplier, dealer, and distributor of stolen copyrighted content that acts like a burglar and an AI fence. The second suit says Brave's Ask Brave tool reproduced substantial passages from its papers, including more than 1,200 words from a Chicago Tribune article. The Times, in a 72-page complaint, says it sent cease-and-desist letters to Brave chief executive Brendan Eich in August 2025 and April 2026 and got no response.
The suits also reach beyond AI. They target a Brave browser feature the publishers say bypasses paywalls, and the company's practice of blocking publishers' ads while selling its own. The Times seeks an injunction and damages; the cases are New York Times Co v. Brave Software Inc, No. 1:26-cv-09549, and Daily News LP v. Brave Software Inc, No. 1:26-cv-09554, in the Southern District of New York. Brave, which has raised roughly 250 million dollars, did not immediately respond to requests for comment. It has previously defended its search indexing as fair use, and it is already litigating with News Corp, which countersued in July after Brave asked a court to declare its practices lawful. These are allegations, not findings, and Brave will get its turn to answer.
For anyone building with AI, the target matters more than the rhetoric. Until now the marquee cases mostly sued the labs that trained the models. This one sues a data supplier, and the plaintiffs' theory is blunt, in their lawyer's words: buying stolen goods secondhand does not make them legitimate. If that theory gains ground, buying grounding or training data through a reputable API will not, by itself, settle where the content came from. Provenance checks, vendor contracts and indemnities for web data move from legal fine print to product requirements, especially for retrieval systems that quote sources back to users.
The same week brought the mirror image on the output side. Anthropic confirmed on October 30 that it added two sentences to Claude Code's prompt on Opus 5.5 steering it away from recreating a specific existing video game, even renamed, because using Claude to infringe intellectual property violates its usage policy. Original games, genre games and work from a user's own designs and assets are meant to keep working. Put the two stories together and the direction is clear: rights holders are policing what goes into AI systems, and labs are policing what comes out. Builders sit in the middle, and the middle is exactly where the liability is being tested.