New York Times and Daily News sue Brave over AI scraping and API syndication

The New York Times and the Daily News have filed a joint copyright infringement lawsuit against Brave Software.

The complaint, lodged in federal court, alleges that Brave systematically harvested millions of copyrighted articles without authorisation. The publishers argue that Brave used this corpus to train proprietary machine learning models. Crucially, the filing also accuses Brave of commercialising the scraped text by syndicating data feeds to third-party artificial intelligence developers, specifically naming Anthropic and Perplexity AI.

The legal action expands the ongoing copyright conflict between news organisations and AI developers into the data infrastructure tier. Where prior lawsuits focused on frontier labs scraping the web directly, this case targets the intermediaries that aggregate, process, and resell web data.

Brave launched its independent search engine in 2021, promoting it as a privacy-centric alternative to Google and Bing. To deliver search results without relying entirely on third-party indexes, the company deployed its own web crawler. The publishers contend that this crawler routinely extracted protected articles across both titles, circumventing paywalls and scraping terms.

According to the complaint, Brave did not restrict the use of this harvested material to traditional search retrieval. Instead, the company allegedly incorporated the articles into training datasets for its internal summarisation features, including its "Answer with AI" product. These features generate synthetic overviews directly on the search engine results page, presenting detailed synopses of publisher reporting.

The plaintiffs argue that these synthetic summaries function as direct market substitutes. When a user reads an AI-generated synthesis of an investigative report on Brave Search, they have little commercial incentive to click through to the original publisher. The complaint maintains that this dynamic diverts traffic, suppresses digital subscriptions, and damages core advertising revenue.

Beyond first-party use, the lawsuit places heavy emphasis on Brave’s commercial Application Programming Interfaces (APIs). Brave markets its Search API to external developers seeking real-time web retrieval data to ground their large language models. The plaintiffs allege that Brave packaged and sold programmatic access to copyrighted reporting to downstream artificial intelligence vendors.

The filing identifies Anthropic and Perplexity AI as major recipients of these data feeds. The publishers argue that Brave profited directly by monetising access to proprietary intellectual property that it had neither licenced nor paid to acquire. By syndicating this material, the plaintiffs assert, Brave enabled downstream developers to train, fine-tune, and ground their own commercial generative engines on unlicenced journalism.

The causes of action span direct, contributory, and vicarious copyright infringement. The publishers also allege violations of the Digital Millennium Copyright Act (DMCA), citing the intentional removal or alteration of copyright management information. The suit claims Brave stripped out author bylines, terms of service notices, and copyright metadata during its extraction process.

The New York Times has pursued an aggressive intellectual property enforcement strategy over the past eighteen months. The publisher previously sued OpenAI and Microsoft in December 2023, initiating a high-profile battle over fair use and training data ingestion. The Daily News and other publications owned by Alden Global Capital joined that legal offensive in subsequent actions.

The action against Brave, however, alters the litigation map. Search engine operators have long relied on legal precedents that protected indexing and snippet generation under fair use doctrines. Brave has historically maintained that indexing the public web is fundamental to preserving competition against tech monopolies.

Yet the integration of retrieval-augmented generation and commercial data syndication complicates traditional safe harbour defences. If the court treats search-based data syndication as distinct from indexation, data brokers face significant exposure under federal copyright law.

The plaintiffs are seeking statutory damages, an accounting of Brave’s commercial profits from API syndication, and permanent injunctive relief. Such an injunction would compel Brave to purge protected publisher content from its search index and model training pipelines, as well as cease programmatic syndication of that data to partners.

As the legal framework for retrieval-augmented generation takes shape in federal courts, the case poses an unresolved question for the data supply chain: can web-scale indexers continue to package the open web into commercial model feeds without direct publisher consent?