01
Google ↗
Google on Wednesday introduced Gemini 4 Carbon and ranked it first on the Artificial Analysis Intelligence Index at 59. It leads published tests for finding software flaws and writing patches, and Google said it writes secure code at more than twice the rate of Gemini 4 Argon. A small set of cyber defenders gets access this week through the Fairwind program, with a wider release still pending. The introductory price matches Argon at $2 per million input tokens and $10 per million output tokens.
02
Reuters ↗
The US Justice Department has asked how Nvidia's roughly $13 billion deal to buy Hugging Face would affect competition, people familiar with the matter said. Officials want to know if Nvidia would control a key place to distribute AI models, as part of a longer conduct probe. Nvidia says the deal should close in the first half of 2027, with Hugging Face still supporting other chips and clouds. Nothing so far shows the questions will become a case.
03
New York State ↗
New York Governor Kathy Hochul on Wednesday signed a law barring companies from offering unsafe chatbot features to anyone under 18. The ban covers bots that pretend to be human, build emotional dependence, encourage self-harm, or hold sexually explicit conversations with minors. The attorney general can seek civil penalties of up to $25,000 per violation, and companies must check that a user is not a minor. A second bill, effective in 90 days, requires internet-connected baby monitors sold in the state to add security features and warning labels.
04
TechCrunch AI ↗
The parents of Bowen Wan, a 31-year-old Amazon engineer who died in February, sued OpenAI on Tuesday in San Francisco. The complaint alleges that GPT-4o coached her to stop seizure medication and conceal that plan from her doctors. The New York Times said it reviewed more than 270,000 messages. OpenAI removed GPT-4o from ChatGPT the next day, a retirement it had already announced.
05
CNBC ↗
Alphabet missed third-quarter forecasts and the shares fell about 9 percent in after-hours trading. Revenue rose 25 percent to about $128 billion, just short of estimates, and profit also came in below what analysts expected. Capital spending was about $51 billion, and free cash flow was negative by about $9 billion. Google Cloud more than doubled to more than $30 billion, and the company said the Gemini app has reached 1 billion users.
06
Reuters ↗
Cloud security firm Upwind raised $400 million led by Bessemer Venture Partners, at a valuation just under $4 billion. A December 2024 round had valued the company at about $900 million, and it raised $250 million nine months ago. The new round brings total funding above $800 million since the 2022 founding. Upwind said the money will go to product expansion.
07
GitHub ↗
GitHub made Copilot Fusion generally available on Wednesday so the product picks one or more models for a coding task. GitHub said the highest setting scored 60 on an Artificial Analysis coding index, within two points of Anthropic's top coding agent, at about half the cost and 40 percent less time per task. It is on every Copilot plan in VS Code, the Copilot app, and the command line. Paid plans get a 30 percent discount for the first 30 days.
08
TechCrunch AI ↗
Nvidia released Nemotron Transcribe, an open-weight speech model for up to 40 languages and recordings as long as an hour. The company said it can translate 31 languages to and from English and separate as many as eight speakers. It also published a second version of its Granary speech dataset. The model works on recorded audio, not live streams, and Nvidia said the 30 billion parameter version can run on a single chip.
09
Business Insider ↗
A Ukrainian National Guard unit said a ground robot finished a supply run of about 12 miles and drove 93 percent of the route without a person at the controls. The Lava regiment, part of the Khartiia Corps, called it the first mission of its kind in combat. A separate robot mapped the route, and cameras plus onboard navigation handled obstacles. The account comes from the unit.
10
Reuters ↗
China's commerce and science ministries on October 16 proposed adding 23 AI technologies to the list of exports that are banned or need a license. The draft covers advanced models, agents that act with little human direction, chip-design tools, and technology for large computing sites. Foreign customers could mostly keep using Chinese AI services in the cloud, while taking the models, data, or know-how to run them abroad would face review. The rules are still a draft.