Much weirder is voice editing in Google Docs, in a new feature called Docs Live. By describing with your voice what you want to write, an agent will dictate your words, generate text, pull in citations from the web, and aim to turn your stream-of-consciousness wishes into a coherent document.
(Reminder: All this stuff may eventually have ads.)
For Gemini power users, Google is creating a new subscription tier, the AI Ultra plan, for $100 a month. It is also dropping the price of its top Gemini AI Ultra from $250 a month to $200.
Google announced Gemini Omni, an AI video generator, akin to Sora 2. That was OpenAI’s generator that let you deepfake yourself, but was eventually killed by the company.
Google’s approach is building out a far more realistic video generator that can incorporate real video and extrapolate all manner of AI-powered weirdness on top of that. Google is eager for you to turn Omni’s eye on yourself, putting your face front and center. As such, selfie videos can be modified to add different backgrounds, styles, or environments, making it appear that you are somewhere other than your actual location.
The feature was demoed on stage with a video of someone recording themselves walking through a metal sculpture. They then asked Omni to change the structure to look like it was made of bubbles. You can also add images and video of yourself from your camera roll and generate just about any variety of cinematic style. Google says Omni is capable of advanced animations and fun typography.
Google’s approach is focusing Omni on video creation first, though it says still image and text capabilities will be coming later. Eventually, Google says it wants to let Omni create any output with any input.
Read more about Omni in Reece Rogers’ story on WIRED. OmniFlash, a starter version of Omni, is available starting today for Google AI+ Pro and Ultra subscribers.
Gemini Spark is Google’s answer to OpenClaw, the viral AI powered helper bot that could be used to help with real life needs like buying groceries or researching vacation options (and occasionally causing you to wind up in a scam).
Spark can write emails or plan a block party and pull information from files in your Google Drive. It is meant to be a personal agent just for you, keeping up with your schedule so it knows the rhythms of your life, learns what major events are coming up, and can help manage long-term or recurring tasks for you.
Spark runs entirely on Google Cloud, which Google says means it can process background requests without having to leave your device on. For now Spark just works with other Google software, though not with the Chrome browser quite yet. Google says that is coming, along with third party support, later this summer.
WIRED’s Reece Rogers has a deeper dive into Spark.
To help you manage all of your online shopping, Google will start deploying an agentic-powered shopping experience. As you search for products, Google will show you listings that it hosts for products for sale at various retailers. You can also shop the old fashioned way, by going to various websites and perusing the listings there.
The big difference is that now, Google will offer a universal shopping cart. Just add the products you’re interested in as you surf, and Google’s agent will keep your wish list organized. It can alert you to price changes and tell you when there’s a newer version or a new color option available. While products are sitting in your cart, you can engage Gemini to ask for more details about your potential purchases, add other products to the cart, or try to find better deals at other retailers.
A slow-moving delivery nearly grounded the hopes of some high-flying rocketry students from Bellevue, Wash., over the weekend. But a last-minute scramble and motor purchase saved the day and capped a strong showing for Washington state teams at the 2026 National Finals of the American Rocketry Challenge.
Washington Youth Aerospace, a Redmond, Wash.-based team made up of six ninth graders from Bellevue’s Interlake High School, finished second in the annual competition in The Plains, Va., on Saturday. The finals featured 100 teams from a record pool of 1,107 teams that competed in the overall challenge.
Washington was represented by 11 teams, including eight from the Eastside of the Seattle area. Four of those teams finished in the top 10.
The competition features middle and high school students who are tasked with designing, building, and launching model rockets. The goal is to inspire students to pursue careers in aerospace and STEM.
Washington Youth Aerospace earned $15,000 for the second-place finish — an impressive showing after the team was in danger of not even being able to launch its rocket.
Because of the hazardous nature of rocket motors, they had to be shipped via ground transportation from coast to coast.
“We mailed it about two and a half weeks back,” said Sudheer Sirivara, a parent advisor and chaperone for the team. “We were tracking it and somewhere it got lost in between for a week.”
The motors arrived in New Jersey on Thursday and on Friday they were tracked to Philadelphia. Sirivara and the team were frantically searching the Washington, D.C., area to find a motor that provided the specs the team had planned around. A vendor on site proved to be a hero just before the event on Saturday.
“He spent about 25 minutes searching for it, and deep in his truck was one box that had this one motor that we needed,” Sirivara said, adding that the team’s actual motors were finally delivered by the Postal Service — two days after the event concluded.
The Washington Youth Aerospace team consists of students Mikhail Antipin, Anay Mediwala, Nikhil Sirivara, Daniel Tadesse, Bao-Ky Tran, and Antoine Vigneron.
Sirivara said their rocketry success started with good mentoring they received from teachers while they were at Bellevue’s Odle Middle School. An executive VP at Warner Bros. Discovery and a Microsoft veteran, he also credited the concentration of tech and engineering parents on the Eastside from companies including Microsoft, Amazon, Google and others.
Good data collection doesn’t hurt either.
“You need to fire a lot of rockets to collect enough data to see how your rocket does in different wind conditions, weather conditions, temperatures,” Sirivara said.
In the finals, teams were scored on two launches. They needed to hit a target height of 730 feet for the first launch and 725 feet for the second. Rockets must stay airborne for between 36 and 39 seconds, and return to the ground safely with their unbroken cargo — an egg.
The Bishop’s School from La Jolla, Calif., took first place in the challenge and will represent the U.S. in the international finals — an event in which Bellevue’s Newport High School finished second just a few years ago.
Here are the final standings for Washington teams in the national finals:
The challenge’s top 25 finishers receive an invitation to participate in NASA’s Student Launch initiative to continue their exploration of rocketry with high-powered rockets and challenging mission parameters.
Google unveiled another AI content-generation tool on Tuesday at its Google I/O developer conference. Are we really surprised?
Gemini Omni was announced at the Shoreline Amphitheatre in Mountain View, California, alongside a slew of agentic AI features, such as Gemini Spark and Universal Cart (check out more on our live blog). Gemini Omni is Google’s new AI content-generation tool that creates graphics and videos based on your prompts.
I hate to burst Google’s bubble, but we already have enough AI-content generation tools, including Google’s own Nano Banana 2, an AI image generator. OpenAI, Shutterstock and Canva also have AI video generation capabilities.
Gemini Omni Flash is now available, so you can create and edit videos in the Gemini app, Google Flow and YouTube Shorts. Output modalities, like images and audio, will be available later. Google gave an example at Google I/O of creating a claymation-style video about how protein is created.
Maybe Gemini Omni will be a cool way to teach my 5-year-old about science, but we all know this is just another tool for more AI slop.
CNET found that many of us are already tired of the futuristic content on our social media timelines. Earlier this year, CNET found that 51% of US adults believe we need better AI labels online, and 21% believe there should be a total ban on AI-generated content on social media. Only 11% say AI content is useful, informative or entertaining.
Despite over half of US adults wanting tighter AI content restrictions, AI is still getting past us. Most (94%) of US adults believe they see AI-generated or AI-altered content on social media. However, only 44% say they can confidently tell real content from AI-generated photos and videos.
Google appeared to reveal a solution to that problem at Google I/O.
Google is adding Content Credentials verification across its Gemini app to show whether the content was created with AI or a camera. It will also detect whether it’s been edited with AI. The SynthID detector is still available on the app to verify AI-generated content.
It sounds like Google is trying to please everyone while still pushing its agenda of shoving AI into every tool and software update possible.
Between Nano Banana Pro and Gemini Omni, that feels like a paradox. The same tech giant that provides the tools to create AI-generated content is also developing tools to verify it.
Gemini Omni is one of many examples of how Google’s developer showcase is out of touch with everyday consumers. There’s still hesitancy and mistrust toward AI, despite how helpful it can be in some use cases.
We don’t need another AI content generation tool. We need to understand how companies are protecting our data, not by hosting demo planning parties or by using smart glasses to turn crowds into cartoons.
The FBI says Americans have lost over $388 million last year to scams using cryptocurrency kiosks, also known as crypto ATMs or Bitcoin ATMs.
Cryptocurrency kiosks are physical, standalone electronic terminals (which may or may not require identity verification to prevent money laundering) that resemble bank ATMs and allow users to buy or sell crypto assets using cash or debit cards.
They are often found around gas stations, convenience stores, and other easily accessible locations. Cybercriminals ask potential victims to deposit their cash into crypto kiosks that then transfer the funds to attacker-controlled crypto wallets.
Earlier this month, Minnesota lawmakers banned cryptocurrency kiosks statewide, after similar moves from Indiana in March and Tennessee in April.
In a public service announcement published on Friday, the bureau warned of a nearly 60% surge in reported losses to crypto ATM scams compared with the previous year.
“In 2025, the IC3 received more than 13,400 complaints reporting the use of cryptocurrency kiosks, with losses over $388 million — a 23% increase in complaints and a 58% increase in losses from 2024. More than half of the complaints involved individuals over 50, with losses over $302 million,” the FBI said.
“In typical IC3 complaints involving cryptocurrency kiosks, criminals give detailed instructions to individuals, including how to withdraw cash from their bank, how to locate a kiosk, and how to deposit and send funds using the kiosk.”
According to complaint data and adjusted losses shared by the FBI, Americans from Texas, Florida, and California filed over 3,300 crypto ATM scam complaints and reported over $112 million in estimated losses.
The FBI also shared some measures that anyone can take to protect themselves from falling victim to such scams, including not sending money to people you only know online, never scanning QR codes or following payment instructions from unknown individuals, always verifying phone calls directly, and not sharing any personal information over the phone.
It also recommended being wary when anyone claiming to be from the government or law enforcement demands cryptocurrency payments, stopping transactions if a kiosk operator warns you of fraud, and always keeping receipts for cryptocurrency transactions.
The FBI’s 2025 Internet Crime Report says the law enforcement agency received over 1 million complaints through its Internet Crime Complaint Center (IC3) last year, linked to almost $21 billion in losses from cyber-enabled crimes such as investment scams, tech support fraud, and business email compromise.
Automated pentesting tools deliver real value, but they were built to answer one question: can an attacker move through the network? They were not built to test whether your controls block threats, your detection rules fire, or your cloud configs hold.
This guide covers the 6 surfaces you actually need to validate.
Nvidia GeForce Game Ready Drivers include day-one support for the hugely popular Forza Horizon 6, and based on our testing, the update brings noticeable performance improvements. While Game Ready drivers do not always deliver meaningful launch-day gains, Nvidia appears to have implemented worthwhile last-minute optimizations this time.
A recent report from Taiwan paints a bleak picture of the current state of the memory industry. DRAM manufacturers are struggling to keep pace with supply chain disruptions and rapidly rising contract prices, forcing them to take extreme financial measures just to stay in business while waiting for better conditions.
Read Entire Article
Source link
The OS update integrates Gemini Intelligence and adds a new widget system.
Alongside a slew of new AI-focused announcements at I/O 2026, Google also announced Wear OS 7, the latest update to its wearable operating system. The software update carries over some of the design tweaks the company plans to introduce in Android 17, lays the groundwork for new Gemini Intelligence features and adds interface elements that display glanceable information in new ways.
The biggest of those additions is what Google calls Wear Widgets, an evolution of the informational “tiles” that have been the bread and butter of the Wear OS experience for years at this point. Wear Widgets are designed to be more dynamic and customizable for developers, and closely mirror what can be offered on smartphones. Widgets that users make with Google’s AI-powered Create My Widget tool will also be able to make the jump to smartwatches.
Live Updates, Google’s system for displaying real-time information on the lock screen of Android phones is also coming to Wear OS, along with a default workout tracker with built-in media controls that can be used across wearable fitness apps. Wear OS 7 will also have controls for deciding which apps automatically launch the Wear OS media controls interface and a new Remote Output Switcher for switching which headphones or speakers play the audio you’re streaming.
Critical to Google’s current focus on using Gemini for agentic AI experiences, Wear OS 7 includes several APIs that can be used by developers to hook up their apps to Google’s Gemini Intelligence system. Those include an AppFunctions API that can integrate features and functions of apps with Google’s assistant, and support for the ability to invoke task automation (like placing an order through a food delivery app, for example) directly from your wrist.
Google plans to detail more of the new features of Wear OS 7 during I/O 2026. A test version of the OS is also available to try now via the Wear OS 7 Canary Emulator, ahead of its release later this year.
Photo credit: NBC News
Months after paying a deposit and chasing down answers through repeated calls and emails, NBC News opened the box on the Trump Mobile T1 smartphone. Inside waited a gold-colored device that turns heads the moment it leaves the packaging. An American flag covers the back, though it carries only eleven stripes instead of the standard thirteen. Trump branding appears in four separate spots across the body, making the origin unmistakable from the first glance.
Unboxing the phone reveals a couple of essential accessories straight away, including a charging cord, wall block, and a clear protective cover, so you don’t have to go out and acquire any other stuff to get started. The item feels really substantial in the hand, however a bit of a handful due to its height. However, the 6.78-inch screen, which is somewhat taller than the new iPhones, gives you some extra real estate to play about with when browsing or checking out some material.
Sale
When you turn on the phone, you can see that it is running standard Android with all of the customizations that make it distinctive. Truth Social is already set up and ready to go, with little effort on your part. 512 gigabytes of storage is plenty for all of your programs, photographs, and data. The 50 megapixel camera performed well for ordinary photos, and they were particularly impressed with the wide angle photographs, which were sharp and had excellent colors.
Phone conversations were fine, and texts got through without any problems. Everything just worked as intended, with no hiccups or surprises. The inclusion of a 3.5 mm headphone connector will be a significant plus for those of you who still rely on wired earphones, especially since so many modern products have abandoned it. To add added security, the fingerprint sensor and facial unlock were both lightning fast.
The phone is presently on sale for $499, which is a reasonable price if you’re considering giving it a try. The construction quality appears sturdy enough for daily usage, and the gold finish appears to be durable enough to withstand the occasional scratch without exhibiting too much wear. The marketing pitch used to be all on how it was created in the United States, but now it appears to be about how it was developed there. The package boldly states ‘proudly assembled in the United States,’ yet when you look at the specs, they’re basically the same as the HTC U24 Pro from Taiwan.
[Source]
China has deployed a massive CPU-only supercomputer called LineShine that delivers 1.54 exaflops of AI training performance without using any GPUs at all.
The system packs 20,480 compute nodes, each containing two LX2 processors for a total of 40,960 chips across the entire machine.
Each LX2 processor has 304 CPU cores, meaning the whole supercomputer uses roughly 2.45 million Armv9 cores in total.
The processor was developed by Huawei or through a joint design with China’s National Supercomputing Center, though the exact origin remains undisclosed.
Each LX2 processor uses two compute chiplets with cores organized into eight clusters containing 38 cores per cluster.
Every core includes ARM‘s Scalable Vector Extension and Scalable Matrix Extension units that accelerate matrix operations used in AI training.
The processor delivers 60.3 teraflops of FP64 performance, 240 teraflops of BF16 throughput, and 960 teraops of INT8 performance from a single chip.
The memory subsystem combines 32GB of on-package HBM delivering up to 4TB/s of bandwidth with up to 256GB of off-package DDR5 memory.
CPU-only systems offer several advantages for complex scientific tasks that combine AI training with massive data ingestion and preprocessing.
Since everything runs on the same processor and memory space, they avoid costly and bandwidth-hungry CPU-to-GPU data transfers.
Homogeneous CPU-based systems can also expose much larger coherent memory pools by combining HBM with large DDR capacities.
This is useful for handling massive scientific datasets, retrieval augmented generation, and long context windows that GPU memory limitations cannot accommodate easily.
CPU-only systems are usually less power efficient and deliver lower-density AI throughput than GPU-based supercomputers.
This is the major reason most of the industry bets on heterogeneous CPU plus GPU architectures for large-scale AI workloads.
China is pursuing this path largely due to US bans on GPU exports, not because CPU-only systems are technically superior for AI tasks.
The LineShine shows that CPUs can successfully perform GPU jobs, but the efficiency gap between the two approaches remains substantial and unlikely to close anytime soon.
China is making a strategic trade-off, accepting lower performance and higher power consumption in exchange for independence from foreign hardware and software ecosystems like Nvidia‘s GPUs and CUDA.
Whether that trade-off makes sense for long-term AI development depends entirely on how quickly Chinese manufacturers can close the performance gap with their own GPU designs.
Until then, the LineShine will remain a remarkable engineering achievement and a practical necessity, but probably not a blueprint for how most of the world will build AI supercomputers.
Via Toms Hardware
Follow TechRadar on Google News and add us as a preferred source to get our expert news, reviews, and opinion in your feeds.
You step out of your car after a long haul in the middle of summer and realize the front end has turned into a magnet for all kinds of bugs. Everything from the bumper to the windshield is caked with bug splatter. Normally, the next course of action is to get the cleanup underway as quickly as possible. While it’s impossible to repel bugs when your car’s barreling down the highway at over 100 mph, there are ways to make dealing with them a lot easier later.
There are several techniques you can choose from, but let’s start with the cheapest: wax. Apply a layer, and the surface gets slick enough that splatters slide off instead of sticking. That’s crucial since bugs carry acidic compounds. Given enough time on hot paint, they can penetrate the clear coat, leaving behind etching, staining, and even some bubbling on the finish. Cicadas can make things even worse with their glue-like residue. Wax can help prevent these problems. And because it’s affordable and applying it is fairly straightforward, you can simply reapply it on your own every few months if it’s wearing off.
A step up from wax is ceramic coatings from reliable brands. Because they’re hydrophobic, they repel water and dirt as well, making it harder for bug juice to stick. But unlike wax, they require significantly more precision and time, which is why it’s best to leave application to a professional detailer. For even more thorough armor, paint protection film (or PPF) for cars works better still. Not only does it soak up impacts from road debris and insects alike, but some PPF products are also self-healing. The catch is that it’s significantly pricier.
On top of all that, you can use bug repellent sprays too — although they’re not to be confused with human bug sprays. The water-based sprays leave a slippery layer, with some holding up for a week or two. They work well when you don’t want to commit to anything permanent, making them ideal for a long road trip.
Then there’s the low-tech route of preventing bugs from ever getting close to the paint: bug screens. These are essentially mesh screens that strap onto the front of your car and can snap on and off relatively easily. You can simply keep them in the car, strap them on when the season is exceptionally buggy, and wash them after a trip. As an alternative, a fabric car bra over the hood does much the same thing, just with less coverage.
Some of the prevention is behavioral, too. The slower you drive, the fewer bugs your car will pick up along the way. That’s because at lower speeds, the airflow around the car gives them a chance to get pushed around. Even if they do hit, the chances of them splattering hard drop significantly. You also save money on actual repairs, since beetles with their hard shells upon impact can quickly become one of the reasons for your car’s chipped paint. If it’s lovebugs you’re dealing with, then timing can help too. Most of them hit their daily peak around 10 a.m. and stop flying at dusk, so during lovebug season, timing drives for the evening can help reduce the problem. When bugs do inevitably hit the paint, try to get them off as soon as possible. That’s because the longer they sit in the sun, the deeper the acid works into the finish.
Trustpilot claims businesses could effectively be “invisible” in AI-generated answers if they fail to build visible trust signals through customer reviews and engagement, indicating a new post-SEO era.
While optimizing for search engines remains key, generative engine optimization (GEO) has introduced yet more challenges for companies looking to survive in 2026 and beyond.
According to Trustpilot’s research, only those with active review profiles are most likely to appear in AI results per the analysis of 800K answers across ChatGPT, Gemini, Perplexity and Google’s AI Mode.
The alarming data reveals brands with no Trustpilot presence were cited in just 1% of AI-generated answers. Conversely, those with 80+ reviews were cited in over three-quarters of answers.
Thankfully, even laying out the foundations can create a meaningful uplift. By creating a Trustpilot account, or presumably a similar online reviews account, the company saw businesses being cited in more than half (53.5%) of AI-generated answers. More reviews, responses and engagement lead to even higher visibility.
This is all important information for companies looking to increase visibility, with more than half (58%) of consumers now using AI to find products and services – a number that’s expected to climb.
The report claims that brands increasingly need AI visibility as well as search visibility – not in place of it – as AI-generated answers quickly emerge as the new front page for businesses.
More broadly, review and trust platforms are playing bigger roles in AI discovery. Per the data, they’re now the second-most cited source type in AI-generated answers, accounting for 14% of all citations.
The availability of fresh content, detailed information and signals of public trust and legitimacy all influence the prevalence of review platforms in AI-generated answers, the firm said, summarizing the ‘3Rs’ – relevance, recency and ranking.
Trustpilot criticized company web pages for being static, corporate places of information, whereas public forums like review sites offer real-time, conversational and experience-based details to tick the relevance box.
Recency is where Trustpilot sings its own praises, amassing around 200K reviews daily in 2025, while a 94/100 domain authority score affords this website high ranking for information retrieval by AI engines.
As for the industry in general, the report details how AI systems are increasingly combining traditional search indexing (hence the continued importance of SEO), retrieval systems, LLMs and real-time web grounding.
Although opening a profile can boost AI visibility somewhat, Trustpilot found that responding to comments and engaging with customers gives businesses the best chances of being seen in AI-generated search results.
It’s not clear why, but the company surmises that the two-way interaction could reduce spam signals and demonstrate accountability.
Live feeds also show that a company is still operational, that customer support exists and that complaints are addressed.
“In an era of AI-powered buying journeys, trust is a quantifiable, high-value asset for businesses,” Chief Customer Officer Alicia Skubick commented.
While Trustpilot’s findings center around its own business model, the data does point to a broader shift in how businesses need to reach customers in an AI-first era.
Future discovery strategies must expand beyond SEO to include trust signals, customer engagement and real-time information to tackle the emerging challenge that is GEO.
Follow TechRadar on Google News and add us as a preferred source to get our expert news, reviews, and opinion in your feeds.
BloFin War of Whales 2026 Grand Prix opens registration for $5M trading championship
Weekend Open Thread: Theory – Corporette.com
E-Estate Announces 1 Year Live: Washington DC Summit as Real Estate Tokenization Enters Its Next Phase
Bitcoin Suisse expands with Digital Asset License and Investment Business Act Registration Approval in Bermuda
Tech Moves: Microsoft AI leader jumps to OpenAI; former AI2 exec joins Meta; and more
Pakistan to enter Chinese capital market as war inflation bites
Bitcoin Suisse expands with Digital Asset License and Investment Business Act Registration Approval in Bermuda
Google’s Gemini AI Predicts Incredible Solana Price by the End of 2026
Google reimburses Register sources who were victims of API fraud
H&R Real Estate Investment Trust (HR.UN:CA) Q1 2026 Earnings Call Transcript
Napoleonic enters 2026 Doomben 10,000 field via Abounding withdrawal
Zara Larsson Has Blunt Response To Chris Brown Diss
Palestine’s flag becoming a regular sight at European football stadiums
Two AI Tokens Lead May Rally, But Risks Are Rising
The Best-Kept Makeup Secret for a More Defined Face
BeInCrypto 100 Institutional Awards Nomination: KAST for Best Digital Assets Neobank and Best Digital Assets Fintech
On the Scene at Gucci’s Cruise Show in New York City: Mariah Carey, Kim Kardashian, Lindsay Lohan, Iman, and More!
Why AI is making typography a boardroom conversation
The 81 MPs (so far) who have demanded tick-like Starmer out
Senate Approves Kevin Warsh to Federal Reserve Board Seat
You must be logged in to post a comment Login