Nvidia's new open models compete with the labs it invests in
Apple's new App Store rule targets what apps do after approval, and Anthropic's 5% fund has a conspicuous gap: nobody has said who decides where the money goes.
Daily technology briefings from Silicon Opera.
Apple's new App Store rule targets what apps do after approval, and Anthropic's 5% fund has a conspicuous gap: nobody has said who decides where the money goes.
Also today: Exchange admins lose a weekend to bad patches, a NetScaler hole demands an urgent fix, and Microsoft reportedly drops a UK AI supercomputer over the power grid.
Also: a Kickstarter-famous $3,999 projector can't decide who it's for, and climate change is coming for the world's best surf breaks.
Meanwhile, data center lobbyists are handing out coloring pages to first graders, and Meta deleted a scraping clause it says changed nothing.
Also: a Meta glasses bug turned a privacy fear real, Node.js's creator walks away from Deno, and lawyers get a tool to check their AI's homework.
Replit's CEO says he dumped Anthropic over cost and performance. At least one person familiar with the partnership says it's still in place.
Benchmark admits the venture pact against backing rivals is cracking, and Cloudflare will start scoring Verified bots on whether they actually back off when told.
Meta says an AI influencer reciting a comic's joke verbatim isn't impersonation, and an a16z partner called the account a genius marketing hack.
OpenAI says 100-plus files followed an ex-employee to Apple, and the UK Legal Aid Agency lost records dating back to 1978.
Cloudflare data says strikes on Yandex didn't stop bots: one gambling-site botnet reappeared on other Russian hosts with the same fingerprint.
Plus: Ukraine's Starlink troubles and a €115 billion warning from the companies Europe is trying to depend on less.
The DOJ says North Korean operatives got U.S. companies to ship laptops to fake hires, and Netflix is cutting 1,000 jobs after losing its Warner Bros. bid.
Meanwhile, hackers pushed supermarket freezers to 50°C, and Microsoft is now counting AI chip racks in gigawatts.
Amex wants a signed proof of intent before an AI agent spends your money, though it says it hasn't yet seen disputes rise from agentic commerce.
Ambi's $26M round shows where ambient AI puts the consent problem: on the user, who has to tell everyone else they're being recorded.
Cloudflare lets you delete log data before you pay to store it, and Ultrahuman wants to test a biological-age model on an entire country.
The DNS root key changes in four days, and Cloudflare built a check you can run now. Also: a counterfeit-certificate report with thin details, and Google's open SynthID checker.
LibreOffice made 'no AI' a feature, and Vinod Khosla is betting an agent startup wins on trust even though its agent can hire a human to finish your errands.
Join thousands of readers who get our weekly breakdown of the most important stories in technology.
Free forever. Unsubscribe anytime.