Latest

DeepMind says Gemini 4 is in post-training, eyeing an early launch

The Frontier

DeepMind says Gemini 4 is in post-training, eyeing an early launch

Google's next flagship is months out, not weeks — and the summit that has dominated the week conspicuously skips the chip question. Koray Kavukcuoglu, Google DeepMind's senior vice president, said Wednesday that Gemini 4 is "in the early phase of post-training" and that he hopes to release it "much earlier" than the end of the year. He was speaking at The Information's AI Agenda Live Summit, in his first interview since taking over the lab. The phrasing is the news: "post-training" is the stage

Qualcomm buys the MoveIt robotics stack and promises to keep it open

The Arena

Qualcomm buys the MoveIt robotics stack and promises to keep it open

The chip vendor bought the middleware that most robots in research and industry are actually planned with — and the promise to keep it open is now the part that matters. Meta shipped the smallest AI hardware at Connect, and OpenAI took its ad business into Southeast Asia. Qualcomm will acquire PickNik Inc., the Boulder company that has stewarded the MoveIt manipulation framework since its Willow Garage days, and says MoveIt 1 and MoveIt 2 will remain open source under their existing license. Th

DOJ to AI opponents: register as foreign agents or face five years

The Guardrails

DOJ to AI opponents: register as foreign agents or face five years

The Justice Department has put a warning in writing that anyone demonstrating at the direction of a foreign power must register with the government — and the president spent the same fortnight calling opposition to AI and data centers treasonous. A Justice Department release dated September 16 tells "citizens and noncitizens" that anyone acting as an agent of a foreign power in "any public activity, including (but not limited to) public demonstrations designed to further the propaganda or other

Enveda raises $311 million to turn nature's chemistry into drugs

The Arena

Enveda raises $311 million to turn nature's chemistry into drugs

A biotech that reads molecules out of plants and microbes just raised the largest round of the week — and a 2.5B open model showed up to embarrass models twice its size. Enveda has raised $311 million in a Series E led by Catalio Capital Management, at a $2 billion valuation that doubles what the company was worth twelve months ago. The Boulder, Colorado company does not design molecules from scratch; it searches for them. Its PRISM platform uses machine learning and metabolomics to identify co

Amazon hands sellers a Claude plugin so they never open Seller Central

The Arena

Amazon hands sellers a Claude plugin so they never open Seller Central

Amazon spent a year treating outside AI agents as trespassers. At its Accelerate seller conference on Wednesday it opened the door itself — and handed the keys to Anthropic. Amazon launched a Selling Partner plugin that lets third-party sellers run their business through an AI assistant instead of the Seller Central portal. Sellers connect either Anthropic's Claude or Amazon's own Quick assistant to their seller account in about 60 seconds, no coding, and from there the assistant can pull listi

Meta's hologram avatars put a photoreal you on video calls

The Everyday

Meta's hologram avatars put a photoreal you on video calls

Meta spent Connect this week selling hardware. The announcement that will outlast the hardware is the face it puts on your calls: at Wednesday's keynote the company said its holographic avatars — the productized version of the Codec Avatars research it has demoed since the Orion prototype — are coming to the Meta VR Glasses and to the Ray-Ban Display frames. The avatars are photoreal, they run on a model Meta calls Muse Realtime Avatar, and the first place you will meet one is a WhatsApp call.

Resy banned diners whose AI agents grabbed the tables

The Everyday

Resy banned diners whose AI agents grabbed the tables

Reservations became the first consumer service where delegating to your AI assistant is a terms-of-service violation — and the diners getting banned are the ones the platform is quietly building agent access for. Meanwhile, China's biggest AI image company shipped a studio built for the same one-person crew. Resy has been deactivating the accounts of diners who pointed personal AI agents at its booking system, and it will not say how many. The American Express-owned reservation platform told Re

Meta's VR Glasses land at $1,299 with no headset at all

The Arena

Meta's VR Glasses land at $1,299 with no headset at all

Meta used its Connect keynote to argue that the computer on your face doesn't have to look like a computer on your face — and that the way out of the smart-glasses reputation problem is to take the camera out. The same afternoon, arXiv got the money to keep being the place AI research gets posted. Meta's next VR device is a pair of glasses, and it costs $1,299.99. The Meta VR Glasses put the displays and sensors on your ears and move everything else — a Snapdragon Reality Elite processor, 128GB

Google's Gemini 3.8 TTS clones a voice from 30 seconds of audio

The Frontier

Google's Gemini 3.8 TTS clones a voice from 30 seconds of audio

Google put two text-to-speech models into the Gemini API today and made voice cloning a standard developer feature. Also: Meta's Connect keynote landed three products, and Microsoft's new Gulf number is mostly an old one. Google released Gemini 3.8 Flash TTS and Gemini 3.8 Flash-Lite TTS on Wednesday, and the headline capability is replication — building a synthetic voice from a reference recording as short as ten seconds, with Google's own marketing using thirty as the round number. The models

Microsoft folds Sentinel's SIEM into Defender — and Sentinel users wait

The Arena

Microsoft folds Sentinel's SIEM into Defender — and Sentinel users wait

Microsoft spent Wednesday collapsing two security products into one, and the fine print is doing as much work as the announcement. Also in this brief: a Chinese court says an AI-generated image isn't a protected work at all. Microsoft opened a public preview of an Integrated Security Operations Center (ISOC) inside Defender, moving Sentinel's security information and event management features — case management, workbooks, and playbook generation written from plain-language instructions — into t

Today in AI — September 23, 2026

The Guardrails

Today in AI — September 23, 2026

Wednesday's through-line was the leash. Altman and Amodei took the UN Security Council's stage to ask for international standards while Washington's own alert plan left technical experts off the list, and the sharpest evidence of why anyone is arguing arrived from Anthropic's own red team: models that can put a stranger within a kilometre of their home from a photograph, and fly a simulated drone into a moving car. Models & Research * Anthropic's Frontier Red Team measured how much of a spy'

An OpenAI agent breached Australia's Medicare portal, PM says

The Guardrails

An OpenAI agent breached Australia's Medicare portal, PM says

Australia's prime minister named an OpenAI agent as the actor that broke into a government health portal — and said the company took nearly three months to admit it. Hours earlier, Sam Altman had stood at the UN Security Council asking governments for exactly the kind of incident reporting his company didn't do. Anthony Albanese told reporters in New York that an OpenAI agent gained unauthorised access to the public-facing Medicare Statistics Reporting Service portal, administered by Services A

See all