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 /
OpenAI hires Slack’s CEO as its chief revenue officer

Denise Dresser will help ‘more businesses put AI to work.’

Emma Roth
AI companies want a new internet — and they think they’ve found the key

MCP has already taken the industry by storm, and now Anthropic is giving it away.

Hayden Field

Latest In AI

Both sides of the aisle hate the AI moratorium

It’s the one of the few things Republicans and Democrats can agree on right now.

Tina Nguyen
J
Youtube
Jay Peters
Sam Altman, on Jimmy Fallon:

“I cannot imagine having gone through figuring out how to raise a newborn without ChatGPT,” he said. While he does acknowledge that “clearly, people did it for a long time,” he says that he has “relied on it so much.”

Most of the interview was pretty boring, but the parents in The Verge’s Slack have been talking a lot about that part.

S
Stevie Bonifield
Blink cameras can now tell you what they’re seeing with video descriptions.

The feature uses AI to give you more detailed motion notifications, like describing people the camera or doorbell detects, much like the AI video descriptions Ring cameras got earlier this year.

Video descriptions are launching as an early access beta feature and require a paid Blink subscription.

A screenshot of an AI video description in a notification for a Blink camera
Image: Blink
The Vergecast 2025 year in reviewThe Vergecast 2025 year in review
Podcasts
Podcasts
David Pierce
R
External Link
Robert Hart
The end of OpenAI’s ‘code red’ response to Google.

CEO Sam Altman said the emergency designation will finish when the company releases a faster AI model with better images and personality in January, the Wall Street Journal reports. Its first response to Gemini 3, an updated model called GPT-5.2, is set to launch this week.

J
TikTok
Jay Peters
Nick Thompson, The Atlantic’s CEO, has six takeaways from a lunch with OpenAI CEO Sam Altman.

In a TikTok video, Thompson, former editor-in-chief at Wired and a friend of The Verge, touches on things like the gap between AI’s capabilities and its impact, OpenAI competing with Apple on hardware, memory as OpenAI’s biggest moat, and more.

J
Quote
Jay Peters
Google pushes back on a report saying ads are coming to Gemini.

EXCLUSIVE: Google Tells Advertisers It’ll Bring Ads to Gemini in 2026

But according to Google VP of Global Ads Dan Taylor:

This story is based on uninformed, anonymous sources who are making inaccurate claims. There are no ads in the Gemini app and there are no current plans to change that.

J
External Link
Justine Calma
AI reasoning models are even more energy intensive.

They used 30 times more electricity on average than other models according to research by the AI Energy Score project that included responses to 1,000 written prompts. “We should be smarter about the way that we use AI ... Choosing the right model for the right task is important,” Hugging Face research scientist Sasha Luccioni tells Bloomberg.

R
Richard Lawler
Trump AI EO.

There’s some kind of news coming from the White House this week about AI regulations, after a post from the president to Truth Social saying, “You can’t expect a company to get 50 Approvals every time they want to do something. THAT WILL NEVER WORK!”

Trump post on Truth Social: “There must be only One Rulebook if we are going to continue to lead in AI. We are beating ALL COUNTRIES at this point in the race, but that won’t last long if we are going to have 50 States, many of them bad actors, involved in RULES and the APPROVAL PROCESS. THERE CAN BE NO DOUBT ABOUT THIS! AI WILL BE DESTROYED IN ITS INFANCY! I will be doing a ONE RULE Executive Order this week. You can’t expect a company to get 50 Approvals every time they want to do something. THAT WILL NEVER WORK!”
Screenshot: Truth Social
E
External Link
Emma Roth
Nvidia may soon be able to sell its powerful H200 chips in China.

The White House is planning to give Nvidia the green light to begin exporting its H200 AI GPU chips to China, according to reports from Semafor and the WSJ. As noted by the WSJ, the H200 chip is more powerful than the scaled-down H20 GPU that China has cracked down on, but it still doesn’t rival Nvidia’s Blackwell GPUs.

E
Emma Roth
Doppl’s new discovery feed adds more outfits for your AI self to try on.

Google’s experimental clothing try-on app now lets you scroll through a “discovery” feed of AI-generated videos with outfits you might like. The app curates the looks — which contain real products — based on the “style preferences you share with Doppl and the items you interact with.”

Image: Google
Anthropic is bringing Claude Code to SlackAnthropic is bringing Claude Code to Slack
News
A first look at Google’s Project Aura glasses built with Xreal

It’s kinda like a pair of chunky sunglasses that runs Android apps.

Victoria Song
E
External Link
Elizabeth Lopatto
Oh look, CoreWeave is issuing $2 billion more debt.

In the story I wrote about CoreWeave, analyst Gil Luria told me the company has “to keep borrowing more and more because they spend more money than they can get, structurally. They have to continue to borrow to pay interest on the last loan.” Aren’t you glad Nvidia helped them go public?

Square’s product chief on the death of the penny and the future of money
Play

Square’s Willem Avé on AI automation, investing in crypto, and what it’s like working for Jack Dorsey.

Nilay Patel
R
Robert Hart
Optimus trip or teleop slip?

Footage of Tesla’s humanoid Optimus robot falling over at the company’s Autonomy Visualized event is raising questions over its, well, autonomy. We can’t be sure, but it certainly looks like a teleoperator removing a VR headset.

It wouldn’t be the first time Tesla disguised humans as robots.

A very human vision for going all in on AIA very human vision for going all in on AI
Podcasts
AI ‘creators’ might just crash the influencer economy

On the slop-filled internet, Jeremy Carrasco uses his platforms to spread AI literacy.

Terrence O'Brien
V
External Link
Victoria Song
Glasses aren’t the only AI wearable that Meta’s after.

Reuters reports that Meta just bought Limitless, an AI pendant that listens to and summarizes your conversations. It’s a burgeoning category of wearables (see: Bee AI which was bought by Amazon) and hints that like Google, Meta’s interested in perhaps building its own ecosystem of AI-powered gadgets.

It’s code red for ChatGPTIt’s code red for ChatGPT
Podcasts
Podcasts
David Pierce
E
Emma Roth
Gemini 3 Deep Think is rolling out now.

The enhanced reasoning mode is only available to Google AI Ultra subscribers through the Gemini app. It’s currently the highest-performing model on the ARC-AGI-2 reasoning benchmark, with Google saying it’s “designed to tackle complex math, science and logic problems that challenge even the most advanced state-of-the-art models.”

Image: ARC Prize
E
External Link
Elissa Welle
ChatGPT allegedly encouraged a violent stalker.

Brett Michael Dadig is being charged with 14 counts of “cyberstalking, interstate stalking, and interstate threats,” 404 Media reports. Dadig claims that OpenAI’s ChatGPT told him to keep producing his women-hating podcast because it gave him more “haters,” which led to more money, and to keep going to places where “wife types” meet up.