Google Search has rolled out Gemini 4 Hydrogen to users worldwide. Google says the update helps Search answer complex questions, find relevant websites and present results with more images and tables.
Google DeepMind has introduced Gemini 4 Hydrogen, an AI model built for fast coding and other multistep tasks. It is rolling out across Google’s developer tools, business services, Gemini app and Search.
Hugging Face CEO Clément Delangue said the company will keep supporting open-source AI after Jensen Huang’s departure from Nvidia. His pledge comes while Nvidia’s planned purchase of Hugging Face is still pending and under scrutiny from U.S. antitrust officials.
SoftBank reported a $7.1 billion quarterly loss, including more than $3 billion tied to deposits on natural gas turbines it ordered for AI data center power projects. Its costly over-ordering suggests demand for gas-fired power may grow more slowly than it expected.
Google has opened preorders for Gemini-powered smartglasses from Warby Parker and Gentle Monster, starting at $395. The glasses use Samsung hardware and let wearers talk to Gemini or use a built-in camera, a feature that may renew privacy concerns about eyewear.
The FAA has expanded its SMART AI tool nationwide to help manage air traffic and suggest flight routes. FAA staff check its advice before controllers receive it, but the controllers’ union says it was not involved in developing or testing the system.
Anthropic is testing a feature that lets Claude join Google Meet calls to draft documents and whiteboards as teams talk. Available to Claude Team and Enterprise subscribers, it lets participants revise the work during the meeting.
An analysis found signs of restricted Nvidia Blackwell chips in six models released since April by DeepSeek, Moonshot, MiniMax and Ant Group. The patterns suggest the chips set the models’ starting values, but do not establish where the hardware was used or whether export rules were broken.
TSMC’s board approved plans to invest up to $1 billion in OpenAI and Anthropic, with up to $500 million for each. The chipmaker already produces many chips used in AI data centers, and analysts warn that investments between suppliers and customers can make demand harder to measure.