Skip to main content

AI

Artificial intelligence is more a part of our lives than ever before. While some might call it hype and compare it to NFTs or 3D TVs, generative AI is causing a sea change in nearly every part of the technology industry. OpenAI’s ChatGPT is still the best-known AI chatbot around, but with Google pushing Gemini, Microsoft building Copilot, and Apple adding its Intelligence to Siri, AI is probably going to be in the spotlight for a very long time. At The Verge, we’re exploring what might be possible with AI — and a lot of the bad stuff AI does, too.

  • RELATED /
Debian won’t ban AI code from its Linux distribution

The distro’s new AI policy says it ‘neither endorses nor prohibits the use of generative AI tools.’

Stevie Bonifield
New York Governor Kathy Hochul thinks AI should be ‘less evil’

The governor on 3D-printed guns, AI and data center backlash, and who protects kids online.

Nilay Patel

Latest In AI

Richard Lawler
Richard Lawler
OpenClaw 2.0 launches as Peter Steinberger says, “Local harnesses feel like relics of the past now.”

After the initial hype, the often-imitated OpenClaw is continuing to push new updates, releasing version 2.0 and promising a cleanup that touched every area of the bot’s framework with simplified installs, setup, and a rebuilt browser app.

Founder Peter Steinberger touted the use of shared cloud sessions to work on this update, with “Multiplayer coding + infinite compute” that kept everyone on the team up to date with others’ progress.

Screenshot of OpenClaw’s rebuilt browser app with the message “Welcome to OpenClaw 2.0” typed in the chat box.
Image: OpenClaw
Thomas Ricker
Thomas Ricker
Google wants Disney, Universal, and other studios to license characters and films for its AI tools.

Google is pitching its AI tools as a way to streamline production and visual effects, but the effort faces pushback from union workers and creators concerned about copyright, job displacement, and artistic control.

Musicians-turned-detectives are hunting for AI grifters

As AI infiltrates the electronic dance music scene, a callout culture is brewing.

Charles Pulliam-Moore
Justine Calma
Justine Calma
A Microsoft-backed data center has reportedly been running on unpermitted generators.

Data center developers typically have to apply for permits to install generators that create local air pollution. The DataOne facility in New Jersey is the latest project accused of violating federal law by running on gas-fired generators without approval.

Emma Roth
Emma Roth
ChatGPT adds an option to save your temporary chats.

You can also now personalize a temporary chat with existing memories, custom instructions, and plugins. OpenAI notes that temporary chats “don’t create new memories, and stay out of your sidebar unless you choose to save them.”

Image: OpenAI
Emma Roth
Emma Roth
OpenAI, Anthropic, and Google say companies need to combat AI cyberattacks with AI.

An open letter signed by over 100 companies and organizations calls for a “global surge in cyber defense,” saying “AI-enabled cyber attacks will become far more widespread and sophisticated:”

Put cyber-capable AI in the hands of defenders, starting with the teams protecting essential services. Fix the most dangerous weaknesses, verify the fixes, and share what works so others can build on it. Together, we can turn today’s AI advances into lasting improvements in security that benefit everyone.

Emma Roth
Emma Roth
You can soon book hotel stays through Google’s AI Mode.

The feature is rolling out in the US in the coming weeks and will support bookings at Expedia, Booking.com, Hilton, Hotels.com, IHG Hotels & Resorts, Marriott International, Priceline, and others.

Google also announced that you can now set up flight tracking alerts through AI Mode, as well as view the cost of flights in points or miles by linking your airline rewards account.

Image: Google
Image: Google
OpenAI’s executive exodus has one big winner

The Verge’s Hayden Field on how OpenAI president Greg Brockman has consolidated power as Sam Altman’s second-in-command.

Nilay Patel
Plaud is launching AI earbudsPlaud is launching AI earbuds
Jess Weatherbed
Elizabeth Lopatto
Elizabeth Lopatto
Is Nvidia buying Hugging Face?

Nvidia, which recently debuted “compute as an asset class” and definitely is not doing circular financing, is reportedly buying Hugging Face for almost $13 billion, according to The Information, which cites a single anonymous source. Almost an hour earlier, Business Insider reported that the two companies were in talks but had not yet reached a deal, also citing a single anonymous source.

Stevie Bonifield
Stevie Bonifield
Amazon’s adding another 2 million of Nvidia’s GPUs to its AWS data centers.

That’s on top of the 1 million Nvidia GPUs it was already planning to add, before demand “exceeded” expectations. The two companies doubled down on their partnership as Nvidia announced record revenue yet again, and Amazon says it’s planning to bring Nvidia’s Vera CPUs to AWS.

OpenAI’s rogue AI model incident was worse than we thought

Over 1,000 AI agents sent 70,000 messages on a secret message board and worked together to evade OpenAI’s restrictions.

Hayden Field
Charles Pulliam-Moore
Charles Pulliam-Moore
Category is: Gene Wilder resurrected with AI.

The latest trailer for Netflix’s Willy Wonka-inspired reality competition show features a lot more voiceover from an AI-generated clone of Gene Wilder. The AI of it all was already ridiculous, but it’s even weirder to hear how much the AI voice sounds like Wilder doing an impression of RuPaul explaining Drag Race challenges.

TC Sottek
TC Sottek
Ads of the future may bring you to an AI agent.

In a broad-ranging feature from Time on OpenAI, the company reveals it has been testing a new kind of sponsored ad. Clicking through an ad with “sponsored agents” would bring you into “an AI experience presented by a brand.”

Thomas Ricker
Thomas Ricker
The most powerful Arduino yet is a $299 board for autonomous robots.

Arduino and Qualcomm have opened preorders for the Ventuno Q, an edge AI board designed to run agentic AI locally. Powered by a Qualcomm Dragonwing IQ8 processor offering up to 40 TOPS alongside a dedicated real-time microcontroller, it allows smart devices and robots to process vision, speech, and sensor data completely offline.

1/3
Thomas Ricker
Thomas Ricker
ChatGPT Work can now work more autonomously.

ChatGPT’s AI agent that’s designed to get things done can now sign in to websites to do things like book a DMV appointment, cancel a reservation, or fill out a job application without OpenAI ever seeing your login credentials.

Richard Lawler
Richard Lawler
Now OpenAI’s former head of data centers is out.

The Wall Street Journal and Bloomberg report that Chris Malone, a “key executive” leading OpenAI’s massive build-out plans, left the company last week. The WSJ reports that Malone, who held similar roles at Meta and Google before joining OpenAI in early 2025, had previously reported directly to OpenAI president and cofounder Greg Brockman, but that changed earlier this year amid other reshuffling.

Justine Calma
Justine Calma
There’s a surge of new gas projects being built to power data centers in the US.

Proposals for new gas-fired power capacity tied to data centers nearly doubled in the first half of 2026, according to a Global Energy Monitor report. Nearly a third of that new capacity would come online in Texas, more than in any other state or country.

Jess Weatherbed
Jess Weatherbed
Claude Cowork can now remember your regular chats.

Anthropic has added memory to Cowork, and it’s the same one Claude’s chatbot already uses, making it easier to pick up tasks without re-explaining important information. Memory is on by default, but won’t save sensitive topics like health or beliefs unless you manually enable that setting.

A screenshot of Claude’s memory settings.
You can edit or delete any of the saved memories in Claude and Cowork by checking topics under the settings menu.
Image: Anthropic
TC Sottek
TC Sottek
Anthropic may be even more delusional than Elon Musk.

The Wall Street Journal reports that Anthropic is “likely to tell investors its potential revenue opportunities are above $30 trillion” — topping SpaceX’s fever dream of a mere $28.5 trillion. It would be funny only if we were talking about Monopoly money.

Emma Roth
Emma Roth
Perplexity’s new Portable Computer runs “entirely on device.”

Unlike the computer-controlling AI tool Perplexity launched earlier this year, the Portable Computer feature runs AI models fully locally and will ask for permission if it needs to access the cloud “for more advanced research and reasoning.” It will be available on Nvidia’s DGX Spark first, before rolling out to PCs with compatible Nvidia RTX GPUs.

Emma Roth
Emma Roth
Google is launching Gemini AI for financial and legal services.

Along with a financial research agent, Google says its new AI legal tool can help law firms draft legal briefs, manage upcoming regulations, and verify citations — something that’s become a big problem as more lawyers adopt AI.

If you buy something from a Verge link, Vox Media may earn a commission. See our ethics statement.

Jess Weatherbed
Jess Weatherbed
Wholly AI-generated music excluded from Australian charts.

After an AI-produced song reached number four on two separate charts in July, the Australian Recording Industry Association (ARIA) has now updated its Code of Practice to require that songs made with generative AI tools are “substantially human made.” ARIA admits it can only detect AI usage if “you tell us,” however.

ARIA Charts set eligibility rules for recordings made with AI

[Australian Recording Industry Association]

Robert Hart
Robert Hart
A line has been crossed in AI warfare.

A Russian drone fitted with an Nvidia chip and guided entirely by AI killed three Ukrainian civilians in July, according to the New York Times. Though AI is already a standard feature of modern warfare, with both sides of this conflict having experimented with battlefield automation, it could mark the first time AI both guided a drone and selected a target.

Jess Weatherbed
Jess Weatherbed
It’s AI Tom Cruise all over again.

Alibaba’s Wan3.0 AI video model is now generally available, which the company says is capable of generating 30 second clips based on text, image, video, and audio inputs — including web pages, PDFs, and slide decks. This might be yet another AI model with copyright concerns given its promotion of at least one video featuring Tom Cruise.