Industry
New York Times sues Brave over paywall bypass and AI content sales
The publisher alleges the browser maker scrapes articles, sells index access to AI firms and lets users read paywalled stories for free.
Still
Crunchy. Tuesday, 3 November 2026
YESTERDAY, STILL WORTH KNOWING
The New York Times sues Brave over AI scraping, Anthropic discloses nearly $900bn in compute commitments, and the NCSC warns AI malware may hit UK infrastructure by 2028.
Publishers take Brave to court
The New York Times is suing Brave Software, maker of the Brave browser, alleging it scrapes news articles, helps users bypass paywalls and sells publishers' content to AI firms. Filed in the US District Court for the Southern District of New York, the complaint says Brave sells AI companies paid access to its index of online content, naming Anthropic, Perplexity, Mistral, xAI and Cohere as alleged recipients, and claims, citing third-party revenue estimates, that Brave has made more than $100m from copied work while publishers received nothing. Ars Technica tested the browser and found its optional paywall blocker, enabled by default, did allow reading a paywalled NYT article; the feature can be turned off. The NYT is seeking an injunction, damages and restitution of profits, and says it had repeatedly asked Brave to stop its infringing conduct before filing. Brave, Anthropic, Mistral and Cohere did not immediately respond to requests for comment; Perplexity, which faces a separate NYT lawsuit, called the case disinformation without identifying inaccuracies.
Three other publishers, the New York Daily News, Chicago Tribune and Orlando Sentinel, filed a separate suit the same day in the same court. As the publishers frame it, the cases will test whether selling scraped publisher content to AI companies carries the same legal exposure as scraping it directly.
The bill for AI arrives
Anthropic, the AI company behind Claude, has filed an updated IPO prospectus revealing nearly $900bn in compute and data centre commitments alongside a $23.3bn quarterly net loss. Quarterly revenue of almost $17bn is twelve times a year earlier, but growth slowed to 46 per cent quarter-on-quarter while compute costs grew 77 per cent, and the filing warns comparable growth rates are not expected. Deals with Microsoft, Google, Amazon and Broadcom total $437bn and are non-cancellable. The IPO, which the New York Times reports could raise around $100bn, will test whether markets buy the bet that enough compute turns this into a sustainable business.
Drax, owner of Britain's largest power plant, plans a datacentre development with investor Brookfield and its AI infrastructure business Radiant at its North Yorkshire site, potentially exceeding 1 gigawatt. The £20bn headline figure is ambition rather than committed funding, with the first phase still needing planning permission and a tenant. Separately, The Information reports some Huawei customers must source memory for its AI chips themselves, effectively pushing buyers into black markets, where smuggled HBM3E has reportedly climbed to roughly $4,000 per unit.
Warnings and classroom controls
The UK's National Cyber Security Centre warns AI-powered malware is highly likely to be used in a ransomware attack on critical infrastructure such as power stations and hospitals by 2028, with Russia and China expected to use AI routinely in cyber-attacks. Chief executive Richard Horne said AI will let attackers exploit long-tolerated vulnerabilities at a scale that was once unimaginable.
Google will turn its AI products off by default for K-12 student accounts, but schools keeping AI disabled will lose Google Search entirely, as the two are integrated. The change reverses this summer's backlash-prompting decision to enable AI features by default in Classroom.