Google just introduced a new smart home speaker yesterday, but the arrival of a new product heralds the end of others. The company has confirmed that it will end production of the Nest Home Mini and Nest Audio. These Nest devices have been on the market for around five years (or longer, if you count 2017's Google Home Mini) and it seems logical for Google to prioritize the newer generation given the current strategy of putting its Gemini chatbot everywhere it possibly can. Although the pair of speakers has been discontinued, any of the two smart home products in the wild will remain operational. Existing Nest Mini and Nest Audio devices will continue to be fully supported with regular software updates, security patches and customer care, a rep from Google told Engadget.

Upgrades include 360-degree audio and deeper integration with Gemini. Last fall Google teased that it was working on an all-new Google Home speaker due out sometime in mid 2026. And while it took a tiny bit longer than expected, today the company began taking pre-orders for its latest speaker ahead of its official on-sale date of June 25. While the new Google Home serves a similar role to the existing Nest Audio, including the ability to function as linked stereo speakers when paired with a second unit, there are a number of other design changes and upgrades. Instead of relying on directional sound, the Google Home was created to deliver clear 360-degree audio in an even more compact chassis.

Arriving six years after Google's last smart speaker, the new HomePod-style device was redesigned to play host to Gemini's chatbot. The last time Google released a smart speaker, the world was in the throes of a pandemic . Yes, it's been six years since the company trotted out a dedicated speaker. However, this newest Google Home Speaker brings a big change with it: The device has been redesigned to showcase the new Gemini assistant instead of the Google Assistant that powered all previous speakers and smart displays. Google announced the speaker last fall alongside new Nest smart home cameras and video doorbells, promising a spring 2026 launch.

Behavioral cloning is a simple yet effective technique for learning sequential decision-making from demonstrations. Recently, it has gained prominence as the core of foundation models for the physical world, where achieving generalization requires countless demonstrations of a multitude of tasks. Typically, a human expert with full information on the task demonstrates a (nearly) optimal behavior. In this paper, we propose to hide some of the task's information from the demonstrator. This ``blindfolded'' expert is compelled to employ non-trivial *exploration* to solve the task. We show that cloning the blindfolded expert generalizes better to unseen tasks than its fully-informed counterpart. We conduct experiments of real-world robot peg insertion tasks with (limited) human demonstrations, alongside videogames from the Procgen benchmark. Additionally, we support our findings with theoretical analysis, which confirms that the generalization error scales with $\sqrt{I/m}$, where $I$ measures the amount of task information available to the demonstrator, and $m$ is the number of demonstrated tasks. Both theory and practice indicate that cloning blindfolded experts generalizes better with fewer demonstrated tasks.

After WIRED reported last week that Meta's smart glasses app contained code that would enable the company to activate face-recognition features on the devices, the company removed the code this week without commenting on why or whether it plans to add such functionality back into the app later. Another WIRED investigation this week found that xAI's Grok is still hosting sexualized deepfakes, including "nudified" images and videos, of celebrities and at least one prominent US politician. After limiting the release of its new Mythos-class AI model over concerns about its potential impacts on cybersecurity, Anthropic announced a model upgrade for partners in its limited-access group this week and launched a "safe" version of the model to the public with guardrails meant to keep the system from being used to fuel cyberattacks. Meanwhile, the United States Cybersecurity and Infrastructure Security Agency issued a new directive to federal agencies this week in reaction to new AI threats that includes a requirement to fix the most urgent software vulnerabilities in as little as three days. As Europe looks to separate and insulate itself from US Big Tech, WIRED created a timeline that tracks all the ways EU governments, companies, and other organizations are moving away from US tech.

The ruling holds that a company that designs, trains, operates, and manages an AI system must assume legal liability for any damages caused by the responses it generates. A local court in Germany has issued a ruling that could reshape the operation of search engines and artificial-intelligence-based chatbots worldwide. The Munich Regional Court preliminarily ruled that Google is liable for a series of false statements generated by its AI Overviews feature, requiring the company to prevent the dissemination of erroneous or inaccurate claims through its search engine. The ruling stems from a case first reported by the Decoder, in which two publishers discovered that Google's AI-generated summaries linked them, in certain searches, to questionable business practices, scams, and subscription-related frauds, without any basis for doing so. Earlier this year, the affected companies sent the tech giant a cease-and-desist letter, according to the report.

Google's Debug project is seeking EPA approval to release millions of sterile male mosquitoes in New Jersey, California and Florida to reduce disease-spreading populations.

Apple's Camera Chief Thinks AI Can Give You Superpowers The generative features in iOS 27's new Photos app will add fake pixels to some of your shots, but Apple's Jon McCormack says the company isn't using AI "for the sake of AI." What even is a photograph these days? As tech giants pack generative AI capabilities into our phones and their camera software, the line between what is a real image and what isn't continues to blur. Phones from Google and Samsung, for example, now come with features that let you drastically alter a photo by erasing people, moving people around in the shot, and even adding new objects to the scene. Apple is getting in on the action by adding new generative features to its Photos app, though the company's iPhone camera chief, Jon McCormack, stresses that Apple is taking a more measured approach than its competitors and isn't "doing AI for the sake of AI."

OpenAI is in the midst of overhauling ChatGPT . The goal is to transform the chatbot's simple interface into a personalized AI agent that can handle tasks in every facet of your personal and professional life. The company has taken to calling this new product, privately and publicly, a "super app." The all-in-one platform represents one of the biggest bets OpenAI has ever made, and one engineering leader now holds enormous sway over whether it pays off: Thibault Sottiaux. Last month, Sottiaux was appointed OpenAI's head of core products, overseeing both ChatGPT and Codex, as well as combining them into the future super app.

Apple took the wraps off iOS 27 at its WWDC event, and the iPhone update is chock-full of smart upgrades, with a big focus on improvements to Siri. While the news coming out of WWDC 2026 is not as dramatic as last year's iOS 26 update, where the new Liquid Glass redesign caused a big stir, Apple's presentation still made waves. During the WWDC keynote, Apple walked through the changes coming to the next version of its mobile operating systems, iOS 27 and iPadOS 27. These updates will roll out to iPhones and iPads later this year, likely in September. There are several smaller quality-of-life improvements, but the big highlight is the Siri AI upgrade .