Amazon has been trying to make virtual assistants happen for more than a decade. Alexa is, by many definitions, wildly successful, but it has so far failed to become the kind of omnipresent, omnipotent helper the company imagines. (It has also, by all accounts, failed to become a compelling business for Amazon.) This week, though, Amazon launched the most ambitious version of Alexa yet, with new technology underneath and some big new ideas about how you might interact with AI.
On this episode of The Vergecast, we talk a lot about whatâs next for Alexa. David Imel â who you might know as the co-host of the Waveform podcast â joins the show to help us figure out what to make of Alexa Plus, and the whole idea that large language models can make virtual assistants both more useful and more accessible. Amazonâs description of Alexa Plus makes a lot of sense, and sounds pretty compelling, but we have reservations both about the user experience and about Amazonâs ability to actually pull this off.
Officials will be able to tell how much time is left and get haptic alerts. | Image: NHL, Apple
The Apple Watch is now the smartwatch of choice for National Hockey League officials. Apple and the NHL just announced a collaboration where on-ice officials will wear Apple Watches that are running special software to receive important in-game information.
Using the custom-built NHL Watch Comms app, on-ice refereescan view the game clock directly from their wrist. Theyâll also be able to receive haptic alerts for when NHL players leave the penalty box or when time is running out in the period. The haptic patterns for each timer are different, so as not to confuse the officials. The Apple Watches will also be synced to the NHLâs Oasis feed, a cloud system that enables player tracking, game telemetry, and other data, which will ensure that all officials receive the same information.
âWe wanted to make sure that the officials had really good awareness and were able to keep their eyes on play,â says Andres de Corral, vice president of digital services at Presidio, a tech firm that helped develop the app. âSo by enabling haptic responses, we were able to provide non-visual cues to the officials.â
Situational awareness is a major challenge for officials on the ice. On t …
It’s the end of an era. Microsoft is shutting down Skype in May and replacing it with the free version of Microsoft Teams for consumers. Existing Skype users will be able to log in to the Microsoft Teams app and have their message history, group chats, and contacts all automatically available without having to create another account, or they can choose to export their data instead. Microsoft is also phasing out support for calling domestic or international numbers.
“Skype users will be in control, they’ll have the choice,” says Jeff Teper, president of Microsoft 365 collaborative apps and platforms, in an interview with The Verge. “They can migrate their conversation history and their contacts out and move on if they want, or they can migrate to Teams.”
If you choose to move on and bring your Skype data with you, the exported data will include photos and conversation history. Microsoft also made a tool to easily view existing Skype chat history if you don’t want to move to Teams.
Skype will remain online until May 5th, so existing users will have around 60 days to decide whether they want to switch to Microsoft Teams or export their data. “If they do want to come to Teams then the first-run is pretty instantaneous because we’ve already done the work on the backend to restore their contacts, message history, and call logs,” says Amit Fulay, vice president of product at Microsoft.
The transition to Microsoft Teams will keep Skype group chats intact, and during the 60-day window, Microsoft will also maintain interoperability so you can message contacts on Teams and those messages will be delivered to friends still using Skype.
If you do move to Microsoft Teams, there’s one big part of Skype that’s disappearing, though. Microsoft is removing the telephony parts that allow you to call domestic or international numbers or people’s cellphones. “Part of the reason is we look at the usage and the trends, and this functionality was great at the time when voice over IP (VoIP) wasn’t available and mobile data plans were very expensive,” explains Fulay. “If we look at the future, that’s not a thing we want to be in.”
Microsoft will honor existing Skype credits, but it will no longer offer new customers access to paid Skype features that allow you to make or receive international and domestic calls. Existing Skype subscription users will be able to use their Skype credits and subscriptions inside Microsoft Teams until the end of their next renewal period. Existing Skype Number users will also need to port their number over to another provider, as Microsoft is no longer supporting this, either.
The Skype Dial Pad will be part of Teams temporarily for existing credits and subscriptions, but Microsoft isn’t going to offer calling plans to Teams consumers like it does for businesses. “The world has really moved on,” says Teper. “Probably the biggest thing is higher bandwidth and lower data plan cost, from us and others, has really driven almost all of the traffic to VoIP.”
The admission of consumers moving on from calling phone numbers from Skype is also a large part of why the service is shutting down nearly 14 years after Microsoft first acquired it for $8.5 billion. Over the last decade, services like FaceTime, Messenger, and WhatsApp have made it simple to connect with friends through messaging, calls, and video chats in a way that Microsoft struggled to compete with through Skype and its many design iterations.
This was particularly evident in the early stages of the covid-19 pandemic, when consumers flocked to Zoom instead of Skype. “The Skype userbase actually grew at the beginning of the pandemic, and has been pretty flat since,” admits Teper. “It’s not shrunk in some dramatic way. It has been relatively flat over the last few years. We hope we’ll migrate most Skype users… but we want to make sure the users know they’re in control.”
Microsoft will now be fully focused on Teams for consumers, after launching the personal version in 2020. At the time, Microsoft said it was still fully committed to Skype, but it’s been clear in recent years that the company was preparing for the eventual retirement of Skype. In December, Microsoft killed off Skype credits and phone numbers in favor of subscriptions, another sign that the end of Skype was nearing.
“Initially the vision was to have one experience across work and life… but Teams was new and that was not realistically where we were in 2020,” reveals Teper. “So we continued to invest in Skype, and about two to three years ago we started bringing in the free Teams consumer experience with the new client. We wanted to wait until the adoption was at the scale where we could be very convinced it was the right time.”
The Skype retirement won’t result in job cuts, either, at least not immediately. “There’s one team, which is Microsoft Teams and Skype. On the backend it has actually evolved to a common team,” says Teper. “There won’t be layoffs, those folks are going to be working on making things better — whether it’s fun end user features or AI innovation, it’s really about doubling down on Teams.”
According to the Securities and Exchange Commission, meme coins don’t meet the requirements to be protected by federal securities law. In new guidance issued on Thursday, the SEC announced that it doesn’t view most meme coins — cryptocurrencies that originate from internet memes or cultural phenomena — as securities, and that transactions around them do not need to be registered with the commission.
The SEC said it reached this decision because the coins do not “generate a yield or convey rights to future income, profits, or assets of a business,” and therefore cannot be considered a security. Instead, the SEC describes meme coins as “more akin to collectibles” and “typically have limited or no use or functionality.”
The clarified guidance could impact crypto regulations and shield companies and individuals that create meme coins from potential litigation. The digital currencies tend to experience volatile price swings and are popularly used in “pump and dump” schemes, in which a token is artificially inflated via insider promotion to cash in on the buying frenzy.
“Meme coins typically are purchased for entertainment, social interaction, and cultural purposes, and their value is driven primarily by market demand and speculation,” said the SEC. “Given the speculative nature of meme coins, they tend to experience significant market price volatility, and often are accompanied by statements regarding their risks and lack of utility, other than for entertainment or other non-functional purposes.”
AMD’s answer to Nvidia’s RTX 50-series GPUs is arriving next week — and its new graphics cards are aggressively priced against Nvidia’s $749 GeForce RTX 5070 Ti and $549 RTX 5070. The AMD Radeon RX 9070 and 9070 XT ship on March 6th for $549 and $599, respectively, one day after the RTX 5070 arrives.
AMD says they both offer “4K gaming at a 1440p price,” though it’s making some unusual comparisons to do so — according to AMD, the 9070 XT is 51 percent faster on average than a four-year-old RX 6900 XT at 4K and max settings and 26 percent faster than a four-year-old RTX 3090 at the same settings, cards that filled a pricey niche two generations ago.
While leaks had suggested the pricing of AMD’s 9070 could start at $649, AMD surprised everyone today with suggested retail prices at $549 for this card — putting it head-to-head with Nvidia’s $549 RTX 5070. The $599 9700 XT could even challenge Nvidia’s $749 pricing for its RTX 5070 Ti, as long as AMD has managed to deliver performance that can challenge both of these RTX 50-series cards.
We won’t know the full answer to that until reviews for AMD’s cards appear next week as well as reviews for Nvidia’s RTX 5070. AMD is dropping some performance hints about where its 9070 series cards might fit in, though. AMD says the 9070 XT is 42 percent more potent at 4K Ultra (and 38 percent at 1440p Ultra) than AMD’s RX 7900 GRE, a card that challenged the RTX 4070 at $549 but vanished after less than a year. And it’s “just barely slightly slower” than a 7900 XTX, AMD’s 2022 flagship, we heard on a call.
The vanilla RX 9070, meanwhile, is 21 percent faster than the same 7900 GRE at 4K, suggesting it’s not far off the XT in performance: 38 percent faster than the AMD 6800 XT and 26 percent faster than the Nvidia RTX 3080, according to AMD. At 220 watts of board power, AMD’s director of graphics product management, Scott Olschewsky, says the 4nm monolithic chip is “the most efficient GPU we’ve ever built.”
Each card also offers 16GB of GDDR6 memory, DisplayPort 2.1a and HDMI 2.1b ports, and are PCIe 5.0 cards — not that your motherboard needs a 5.0 new slot or would necessarily benefit from one. They will use standard 8-pin PCIe power connectors.
While both the 9070 and 9070 XT have far fewer graphics compute units than the 7900 GRE (55, 64, and 80, respectively), they don’t need as many because they’re the first cards to ship with RDNA 4, which AMD says offers twice the rasterization (read: non-ray traced) graphics performance per compute unit as RDNA 2. (RDNA 3 was closer to 1.4x that of RDNA 2.)
AMD says it’s taken larger steps than that in ray tracing, too, and almost fully doubled its FP16 machine learning performance since RDNA 3, and up to 779 TOPS, though performance will differ from app to app: AMD’s only seeing 12 percent more performance in Adobe Lightroom super resolution over the 7900 GRE, as one example.
But even if you’re not playing with AI very intentionally, AMD will put that hardware to use in gaming with FSR 4, which comes with a new AI upscaling algorithm for its super resolution tech that’ll exclusively run on these RDNA 4 and newer cards.
Olschewsky says FSR 4 can increase frame rates “while looking just as good as native rendering” at 4K resolution. AMD used some images from Warhammer 40,000: Space Marine 2 — a title that averages just 53fps at 4K Ultra settings — to show more fine detail in the distant background of an image while averaging at 182fps with FSR 4 and frame generation turned on.
And while AMD didn’t share expected performance for that game without the fake frames added, it did break out the differences for a handful of other games like Call of Duty: Black Ops 6 and Ratchet & Clank:
AMD says it’ll have 30 FSR4 games at launch and over 75 by the end of the year — one obvious theme is a lot of PS5 games, making us wonder how closely the PS5 Pro’s “PlayStation Spectral Super Resolution,” an AI upscaler powered by AMD hardware, might be related to it. But AMD also says FSR 3.1 games can be updated to FSR 4 “with a simple toggle.”
The new cards also have an enhanced media engine with higher image quality for gameplay recording, as the last one “did not meet quality expectations from the gamers trying to record and stream.” AMD says its Fluid Motion Frames (AFMF), its generic fake frame generation you can apply to any game yourself, is being upgraded to AFMF 2.1, with less ghosting and smearing of details.
As with any graphics card these days, the actual amount you pay will be determined by whether AMD can actually make enough of the things, and if AMD’s board partners don’t jack up the price. Olschewsky had some fighting words for Nvidia there, though: “We believe that our 9070 XT will be going toe-to-toe with what you’ve seen from the 5070 Ti that users may or may not be able to buy,” adding that the 9070 “is going to look very strong against their upcoming 5070.”
Unlike Nvidia, which warned of shortages with its new cards, he says AMD expects “strong availability at launch.”
“We are working with all of our partners to go and deliver the most competitive prices on GPUs around the world; our focus is ensuring cards hit target price points around the world. It’s difficult to predict what’s going to happen, but we’ll be working with partners as we see pricing show up.”
At the end of AMD’s event today the company also teased that its RX 9060 series cards will arrive in the second quarter to “broaden the family even further.” AMD plans to disclose more information about these cards “later,” but it certainly looks like it’s getting ready to challenge Nvidia’s unannounced RTX 5060 cards.
The shadow president paced around the stage after his speech, sunglasses on, mouth frozen in a grin, raising a chainsaw overhead to the delight of an adoring crowd as a large rectangular canvas made its way from the back of the audience toward him. He grabbed the painting, visibly thrilled. On the canvas, as onstage, he was the focal point. Beams of light emanated from his head, which the artist had superimposed over scenes from the world his real-world counterpart had promised to build: an astronaut surveying a barren red planet, a futuristic civilization complete with flying cars. In the painting, he wore a suit and tie. His real-world attire was more casual: a black blazer over a novelty T-shirt that read âIâm not procrastinating, Iâm doing side quests,â a gold chain, his signature âdark MAGAâ hat, and the aforementioned sunglasses. In the painting, he was triumphant. Onstage at the Conservative Political Action Conference (CPAC), however, Elon Musk had appeared incoherent. Fifteen or so minutes into the interview, a reporter in the media pit turned to me and mimed smoking a joint, mouthing, âIs he high?â
Still, the audience was in his thrall. At one point, whe …
Airbnb co-founder Joe Gebbia has joined President Donald Trump’s Department of Government Efficiency (DOGE) service. Gebbia, a close friend of Elon Musk and fellow billionaire, announced that he’s been tasked with “improving the slow and paper-based retirement process,” though the specifics of his involvement are unclear.
“Excited to share I’m bringing my designer brain and start-up spirit into the government,” Gebbia announced on X. “I can think of few more important ones than volunteering to improve the user experience within our government.”
Musk has complained that the current system for processing retirement applications is too slow and restrictive due to using manually checked paper records. A converted mine in Pennsylvania is currently used to store and process 400 million printed government documents, taking up 26,000 filing cabinets. The US Office of Personnel Management announced on Thursday that it had processed an entire retirement application digitally for the first time, completing the task “without printing a single piece of paper” within two days, instead of the several weeks it usually takes.
The retirement process for every federal employee involves a paper application, filing cabinets, and a mainframe in a mine.
It’s unclear if Gebbia’s role at DOGE is a paid position. Gebbia remains a significant shareholder and board member of Airbnb despite leaving his operating role at the company in 2022 — the same year he joined Tesla’s board of directors. In response to Gebbia’s DOGE appointment, some Airbnb users and hosts have posted on the rental platform’s community forums with threats to boycott the service and called for Airbnb to distance itself from the co-founder.
Gebbia attracted similar ire from the Airbnb community last month when he revealed that he had voted for Trump in the November elections. Gebbia had been a Democratic donor until 2023 and scrutinized Trump for his “heartless, cruel, and immoral” family separation policy during his previous presidency. Having professed to having a “woke-up call,” his DOGE appointment further cements his pledge to support the MAGA ethos.
The original Red Alert is going open source for the second time.
Electronic Arts (EA) is releasing the source code for four Command & Conquer titles under the open-source GPL license. The original Command & Conquer (since subtitled Tiberian Dawn) is joined by Red Alert, Renegade, and Generals, the code for all of which can now be found on EA’s GitHub page. Only the code has been open sourced, not the games’ assets and cinematics, but it will help modders and the game restoration community keep the games playable.
This isn’t actually a first for EA. Back in 2020 the company released the source code for its Command & Conquer Remastered Collection, made up of Tiberian Dawn and Red Alert. That code had already been adapted for the remaster’s engine however, while the new releases are the “fully recovered source code” of the series’ first two games, according to Luke “CCHyper” Feenan, a Command & Conquer community member who proposed and orchestrated the release together with EA.
Renegade and Generals, meanwhile, have been released under an open-source license for the first time. Renegade is a 2002 first- and third-person shooter set in the franchise’s Tiberium universe, while Generals is a 2003 strategy game that eschewed the Tiberium and Red Alert worlds for a near-future setting depicting a war between the United States, China, and the fictional Global Liberation Army. Its expansion Zero Hour is also included in the open source release.
Alongside the open sourcing, EA has also opened Steam Workshop support, and released a ‘Modding Support’ pack that includes the source XML, schema, script, shader and map files, for all the games that use the SAGE engine:
C&C Renegade
C&C Generals & Zero Hour
C&C 3 Tiberium Wars and Kane’s Wrath
C&C Red Alert 3 & Uprising
C&C 4 Tiberian Twilight
That move should make it easier to create mods and maps for the games, and to share some of those creations through Steam. To cap off the announcement, EA released a 35-minute video of archival gameplay footage from the early development of Renegade and Generals:
Tri-folds are having a moment. Thereâs that impressive Huawei device, my favorite 3-in-1 Apple charger, and now this: the Zip tri-fold projector from a company called Aurzen. Itâs the most gadgety gadget Iâve tested in a long time.
The Zipâs heft, texture, and hinge stiffness evokes quality at first touch and itâs impressively bright for a compact battery-powered projector that initially costs $249.
Using it is also a joy. It connects quickly to iPhones over AirPlay and to Android devices over Miracast, Smart View, or similar using Wi-Fi Direct â no hotspot required. It then automatically focuses and aligns the image on any available flat surface including walls, t-shirts, and pillows. It works in both landscape and portrait modes and pairs with Bluetooth headsets for private audio or Bluetooth speakers for a shared experience. The built-in rechargeable battery lasts about 80 minutes in my real-world testing, but you can always plug it into a powerbank or wall jack to extend that.
Using it sucks, however, if youâre trying to stream from services like Netflix and Disney Plus or trying to watch Spotify music videos. Thatâs because all those companies protect their …
The EV4 sedan was first announced as a concept in 2023, but it’s now a real car. It comes in two versions. One has a funky rear that stretches out to look like a traditional 4-door car. The other, a “five-door” hot-hatch-looking version will primarily be for the European market, according to Kia. These aren’t going to be Kia’s most-performant EVs though, as they will be single-motor front-wheel drive vehicles with just 150kW of power and a 0-62 mph acceleration time as quick as 7.4 seconds.
The EV4 will also run on a lesser 400-volt version of the company’s E-GMP platform instead of the faster-charging 800-volt versions used in the EV6, Hyundai Ioniq 5, and others. Kia says the system can still charge 10 to 80 percent in as little as 31 minutes.
The vehicles could shine in battery range. Kia says they’ll get up to 630km (about 391 miles) on a single charge with the larger 81.4 kWh battery option, and 430km (about 267 miles) on the smaller 58.3 kWh battery. However, those estimates are based on Europe’s more generous WLTP standards.
The EV4 still offers some of Kia’s cooler EV tech. It has a 30-inch widescreen display (really three side-by-side instrument cluster and infotainment screens) that run Kia’s latest “connected car Navigation Cockpit (ccNC) software. You get entertainment options like YouTube, Netflix, games, and more that will be available from an app store.. You can also get an AI voice assistant, Apple Watch digital key access, V2L for powering household devices, and a “Smart Cruise Control 2” advanced driver-assistance system with lane keeping.
Kia’s concept EV2, which had purportedly been spotted in camouflage last year, is supposed to be even cheaper than the EV4. The company says the mini-SUV will be its “smallest EV yet” and will have options to reconfigure seating to maximize front row space for lounging or for rear cargo. It will also have perks like removable portable speakers for tailgating.
The EV4 sedan will initially be built in Korea and will first launch there in the middle of March. The EV4 hatchback will be produced in Slovakia and will launch in Europe in the second half of 2024. The North American version of the sedan will be produced “later in the year.” The EV2, meanwhile, is coming in 2026 for Europe and other regions, although its US availability is yet to be confirmed.
Kia President Ho Sung Song announced at the event that EV4 pricing will start at 37,000 euros (about $38,500), Electrek reports. The initial run will include 160,000 units, of which 80,000 are destined to ship to North America.
With the EV4, US car buyers might soon have another affordable mass-market electric car option against the Tesla Model 3. The EV4 could fill some vacancies, too, as automakers like Volkswagen, which is no longer bringing the ID.7 stateside, re-tune their EV strategies amidst the Trump administration’s interests to eliminate electric vehicle incentives.
Meta is planning to launch a dedicated app for its AI chatbot, according to a report from CNBC. The Verge can also confirm that Meta is working on the standalone app. The new app could launch in the second quarter of this year, CNBC says, joining the growing number of standalone AI apps, including OpenAI’s ChatGPT, Google Gemini, and Microsoft Copilot.
Meta has already brought its AI chatbot across Facebook, Instagram, Messenger, and WhatsApp, but launching a standalone app could help the company reach people who don’t already use those platforms. Similar to rival chatbots, Meta AI can answer questions, generate images, edit photos, and more. It recently gained the ability to use its “memory” to provide better recommendations.
In a response to CNBC’s report, OpenAI CEO Sam Altman joked, “ok fine maybe we’ll do a social app.” Meta declined to comment.
Meta has ramped up its efforts to compete in the AI industry in recent months, with CEO Mark Zuckerberg announcing plans to invest up to $65 billion to further the company’s AI ambitions. The company also plans on holding an event dedicated to AI on April 29th.
Amazon is gearing up to launch new hardware to go along with its AI-upgraded Alexa. During an interview with Bloomberg, Amazon CEO Andy Jassy said the company has a “brand new lineup of devices that are coming in the fall that are beautiful.”
On Wednesday, Amazon revealed Alexa Plus, a more conversational version of the smart assistant that’s capable of performing a wider range of tasks, such as ordering an Uber or finding concert tickets. Amazon says “almost every” Alexa device will support Alexa Plus, save for older Echo generations. Alexa Plus will cost $19.99 per month but will be included with a subscription to Prime.
Though Jassy didn’t share any other details about the new devices, it seems like the company plans to put an even bigger focus on displays. When asked about Amazon’s next-generation hardware, Panos Panay, Amazon’s head of devices and services, told my colleague Jennifer Tuohy that, “I believe in screens. I think they matter in a massive way.”
The Cricut Maker 4 and Explore 4 crafting machines in their new seashell and sage color options. | Image: Cricut
Cricut has announced new versions of its crafting machines designed to print, cut, and emboss various materials using a collection of swappable tools. The new Cricut Maker 4 and Cricut Explore 4 are the first updates to both machines in nearly four years and offer faster cutting speeds for some materials, improved accuracy, and cheaper price tags.
The machines will be available starting on February 28th, 2025. They’ll still start at $249.99 for the Cricut Explore 4, and at $399.99 for the Cricut Maker 4, and both will be available in sage and seashell color options. The four-year-old Cricut Explore 3 sells for $319, while the Maker 3 is $439, so the cheaper price tags for the 4-series line could help make the new machines more accessible to new users. Both models come with a bundle of crafting materials to complete 10 initial projects.
The Cricut Maker 4 and Explore 4 look nearly identical to their predecessors but are up to twice as fast when cutting through materials like cardstock and vinyl, Cricut says. The speed improvements won’t be quite as dramatic when using thicker materials, but the upgrade will still appeal to crafters who sell their creations on sites like Etsy and are looking to upgrade their output.
Cricut has also introduced a new optical sensor on both models that works alongside a light that better illuminates materials, according to CNET. The upgrade should help improve cutting accuracy, no matter what the lighting conditions are in your crafting room.
Although there are differences between both machines’ capabilities, they’re mostly dependent on the types of projects you’re looking to create. The pricier Cricut Maker 4 is designed to work with over 300 different types of materials, including thicker options like leather and balsa wood, and a wider variety of cutting and embossing tools. The cheaper Cricut Explore 4 is limited to around 100 different materials, including thinner stock like vinyl, card stock, and iron-ons.
While the new machines don’t require subscriptions, the company still offers a Cricut Access subscription for users who heavily rely on pre-made designs and projects available through its Design Space app. Cricut Access also expands the number of fonts and images available through the app and includes discounts on materials.
Apple announced in a whitepaper that it plans to introduce a bunch of new child safety features, including letting parents share their kids’ age ranges with apps, refreshing the App Store’s age ratings system, and making it easier for parents to set up Child Accounts for their kids. The company says it will introduce the features “this year.”
Companies like Meta, Snap, and X have called for platforms to be responsible for verifying the ages of users at the OS or app store level. Apple also reportedly lobbied against a proposed bill in Louisiana that would have required the company to enforce age restrictions.
In the whitepaper, Apple argues that age verification “at the app marketplace level” wouldn’t be ideal, as it would require users to hand over “sensitive personally identifying information” to the company. “That’s not in the interest of user safety or privacy,” Apple says.
The age sharing system gestures in that direction without going so far as to fully verify each user’s age. With the age range feature, “parents can allow their kids to share the age range associated with their Child Accounts with app developers,” Apple says.
The age range will “be shared with developers if and only if parents decide to allow this information to be shared,” and parents will be able to disable sharing. The feature also won’t “provide kids’ actual birthdates.” Developers will be able to request the age ranges with a new API that Apple says is a “narrowly tailored, data-minimizing, privacy-protecting tool to assist app developers who can benefit from it.”
“Today’s announcement is a positive first step, however, developers can only apply these age-appropriate protections with a teen’s approval,” Meta spokesperson Jamie Radice says in a statement to The Verge. “Parents tell us they want to have the final say over the apps their teens use, and that’s why we support legislation that requires app stores to verify a child’s age and get a parent’s approval before their child downloads an app.”
As for App Store ratings will expand from four thresholds to five; the new categories will be Age 4 plus, 9 plus, 13 plus, 16 plus, and 18 plus. In their app listings, developers will be asked to highlight “whether apps contain user-generated content or advertising capabilities that can impact the presence of age-inappropriate content” and if apps have their own content controls.
Apple says that the App Store won’t show kids apps with age ratings “in the places where we feature apps on our storefront” that are higher than what their parents set on their accounts.
As for Child Accounts, Apple says that it will introduce a new setup process and let parents fix the age associated with the account if it wasn’t set up correctly.
Microsoft is releasing a native Copilot app for macOS today. Much like the Windows app, the Copilot version for Mac will provide access to the web-based version of Microsoft’s AI assistant, where you can upload images and generate images or text.
The macOS version of Copilot also includes a dark mode and a shortcut command to activate the AI assistant by using Command + Space — much like the Alt + Space on the Windows version. Microsoft is launching this new Copilot Mac app in the US, UK, and Canada today, and the iPad version is also being updated with a split screen mode.
You’ll also now be able to log into Copilot on an iPhone or iPad with an Apple ID, and upload text or PDF files to ask questions about the documents or generate a summary about them. This document summarization feature is also coming to the macOS app soon.
The launch of Copilot on macOS comes just days after Microsoft made Copilot Voice and Think Deeper free with unlimited use. Previously, both Think Deeper (powered by OpenAI’s o1 model) and Voice in Copilot had limits for free users, but Microsoft has now removed these to allow Copilot users to have extended conversations with the company’s AI assistant.
The Echo Show 21, Amazon’s newest smart display, shows the new user interface for viewing your calendar, playing music, and other tasks.
Oh, Alexa, how you’ve changed. The long-awaited new Alexa, Alexa Plus, is set to bring a more conversational, context-aware, and capable assistant to your smart home. With a new voice (eight of them, in fact) and a new attitude, this is the biggest change to the voice assistant since it debuted in 2014. And it all sounds very impressive.
Announced at a press conference in New York City this week, Alexa Plus offers several new generative AI-powered abilities to help you manage your life, plus some major smart home upgrades. I was at the event and saw several staged demos of the new features but also got to try out some of the smart home improvements for myself.
The biggest change is how Alexa can respond to natural language; the demo showed that you can talk to it and say what you want rather than having to remember specific commands. I saw the new Alexa understand and execute commands such as, “Bring the lights up in here and set to a warm glow.” The GE Cync smart bulbs and light strips in the room responded, despite the request not including a room or specific names for each device.
Then, when instructed to “Turn on the lamp in the sitting area,” Alexa was apparently able to “reason” that meant the lamp named “Sofa lamp.” This should mean no more memorizing specific device names, making it easier for anyone in the home to control devices with their voice.
I was also able to talk to the assistant myself and try out its new ability to follow multiple commands at once without needing to repeat the wake word, Alexa. I asked it to dim the lights and “make it a little warmer.” The thermostat adjusted while the lights dimmed. Alexa said, “I dimmed the lights in the living room and increased the temperature by two degrees; is there anything else you need?” I then said, “Can you vacuum the floor?” It replied, “Okay,” and the Roomba started a job.
This should mean no more memorizing specific device names
Another new feature I tried was the ability to set up a smart home routine just by using voice. I told Alexa I’d been having trouble waking up recently, and after some back-and-forth, it created a “Good Morning” routine that set an alarm to wake me up to Taylor Swift’s “Shake It Off” and adjust the smart lights in my room.
That’s a fairly simple one because there weren’t a lot of devices connected to the Show. But Amazon says that, courtesy of its knowledge of hundreds of smart home APIs, Alexa is capable of creating more complex routines through voice. This should make it easier for people to do more with their connected devices and not have to spend time programming an app.
The other exciting upgrade is the new Echo Show UI. This is launching on the 15 and 21, but Scott Durham of Amazon told me it will come to the Show 8 and 10 at some point. With a cleaner, sleeker full-screen UI with larger widgets to take advantage of the screen size (I saw it demoed on the 21-inch screen), it’s now more customizable and feels more like a tablet interface than a smart display. During the demo, it appeared to move smoothly and quickly with limited lag.
The UI now has a much larger calendar and smart home widgets. A handy new feature is the ability to send images, documents, and notes to your [email protected] email address or through the Alexa app or new web interface. From there, Amazon says it can parse things like events and add them to your calendar, as well as let you ask questions about the information. Apparently, it can even decipher all the info in that lengthy school email and set reminders to tell you what you need to send in on which days.
The smart home control UI has been lifted from the excellent interface on the Echo Hub, giving you easy touch control of devices in your home when you don’t want to use voice or a smartphone. And when you take it full screen, it’s now much easier to switch between rooms and devices. Alexa’s Map View is here, too, and it looks great on the big screen.
A new Ring camera integration feature lets Alexa Plus query Ring’s Smart Video Search to show you summaries of events that happened around your house or pull specific instances like, “Did a package arrive?” or “Did someone let the dog out?”
Another big improvement coming with Alexa Plus is new cooking controls. Currently, following recipes on an Echo Show can be fiddly and frustrating; with Alexa Plus, the assistant gets more proactive. It can now take ingredients from recipes and add them to a shopping list, let you use natural language to add additional items, and arrange to have them delivered to you (with Whole Foods and Amazon Fresh as well as “several other grocery providers.”)
Alexa can also come up with recipes based on ingredients you tell it you have on hand and suggest substitutions for items you’re out of. The kitchen is one of the most useful places for hands-free voice control, and if this works well, I can see it being very helpful. I’m most excited about the new timer feature, which takes all the time-based steps in the recipe — such as whisk for two minutes or bake at 350 degrees for 35 minutes — and automatically sets them for you to kick off when you’re ready.
Aaron Rubenson of Amazon told me that thanks to Alexa’s improved natural language skills, it interfaces better with smart kitchen appliances. So, instead of having to use specific nomenclature to get it to preheat my Thermador oven, it should respond to any command that implies I want my oven on. For example, “Alexa, can you set the oven to the right temperature for this recipe?”
I’ve used Alexa for close to a decade now, and while it has its uses, it’s never felt indispensable. This is largely because of how tricky it is to talk to correctly. I’ve had to learn Alexa-speak to get it to do anything reliably, often making it more frustrating than useful. If the new Alexa can work as well in my home as it did in the demos I saw this week, this will be a major shift in home automation.
Alexa Plus pricing and availability
Amazon Alexa Plus costs $19.99 a month and is included in Prime membership. It will be available via an early access program in late March, in the US only, to customers with an Echo Show 8, 10, 15, or 21.
It’s also accessible in a new Alexa app and at Alexa.com. Amazon says it will come to other Echo devices, including Echo Buds and Echo Frames, and will be compatible with Fire TVs and Fire tablets.
OpenAI is launching GPT-4.5 today, its newest and largest AI language model. GPT-4.5 will be available as a research preview for ChatGPT Pro users to start. OpenAI is calling the release its “most knowledgeable model yet,” but initially warned that GPT-4.5 is not a frontier model and might not perform as well as o1 or o3-mini.
GPT-4.5 will have better writing capabilities, improved world knowledge, and what OpenAI calls a “refined personality over previous models.” OpenAI says interacting with GPT 4.5 will feel more “natural,” adding that the model is better at recognizing patterns and drawing connections, making it ideal for writing, programming, and “solving practical problems.”
However, OpenAI notes it won’t introduce enough new capabilities to be considered a frontier model. “GPT-4.5 is not a frontier model, but it is OpenAI’s largest LLM, improving on GPT-4’s computational efficiency by more than 10x,” OpenAI said in a document that leaked ahead of its announcement. “It does not introduce 7 net-new frontier capabilities compared to previous reasoning releases, and its performance is below that of o1, o3-mini, and deep research on most preparedness evaluations.” OpenAI has since removed this mention from an updated version of the document.
It was previously reported that OpenAI was using its o1 reasoning model, codenamed Strawberry, to train GPT-4.5 with synthetic data. OpenAI says it has trained GPT-4.5 “using new supervision techniques combined with traditional methods like supervised fine-tuning (SFT) and reinforcement learning from human feedback (RLHF), similar to those used for GPT-4o.”
Despite some of its limitations, GPT-4.5 hallucinates a lot less than GPT-4o, according to OpenAI, and slightly less than its o1 model. “We aligned GPT-4.5 to be a better collaborator, making conversations feel warmer, more intuitive, and emotionally nuanced,” Raphael Gontijo Lopes, a researcher at OpenAI, said during the company’s livestream. “To measure this, we asked human testers to evaluate it against GPT-4o, and GPT-4.5 outperformed on basically every category.”
In a post on X, OpenAI CEO Sam Altman acknowledged that GPT-4.5 is a “giant, expensive model” and that it “won’t crush benchmarks.”
Following its launch for Pro users, OpenAI says GPT-4.5 will roll out to Plus and Team users next week and then to Enterprise and Edu users after that. It’s also available now in Microsoft’s Azure AI Foundry platform, along with new models from Stability, Cohere, and Microsoft.
We revealed in Notepad last week that OpenAI was planning to launch GPT-4.5 by the end of February, and GPT-5 as soon as late May. OpenAI CEO Sam Altman has referred to GPT-5 as a “system that integrates a lot of our technology,” and it will include OpenAI’s new o3 reasoning model, which the company teased during its 12 days of Christmas announcements in December.
While OpenAI released o3-mini last month, OpenAI is only shipping o3 as part of its upcoming GPT-5 system. This aligns with OpenAI’s goal to combine its large language models to eventually create a more capable model that could be labeled as artificial general intelligence, or AGI.
Update, February 27th:Noted that OpenAI changed its GPT-4.5 system card and added information from Sam Altman.
I tell Amazon’s Alexa to shut up on a near daily basis. I have almost zero interest in speaking to Gemini after our first awkward chat. The hitches, misunderstandings, and lag in any given AI “conversation” mean I’m always wasting time speaking when I could be texting instead.
But speaking to “Maya,” one of two voices from a new startup headed by the man who built Oculus VR and sold it to Facebook, is the first time I’ve been left wanting more. Like I could just talk to it, or at least play a genuinely fun game of testing its limits, like I did with Bing before Microsoft decided to tame down its unhinged persona.
I don’t have to describe it to you: you can try it, and you can listen to my first conversation yourself just below. Fair warning: I am a nerd! Confronted with a new voice assistant, I will ask it to dream up a Dungeons & Dragons-esque adventure and quiz it about small Android phones.
But here you go:
While I could absolutely still hear some chatbot nonsense coming through the cracks, I could easily interject – I asked Maya to inject “herself” into the adventure “she” was describing, and it did so without a hitch, immediately coming up with a Gnome engineer named Maya cobbling together deathtraps to protect my castle from incoming Orc invaders. Combined with the AI’s natural-sounding pauses, it felt more like a real conversation than anything I’ve had so far. Compared to my colleague Kylie Robinson’s conversation with ChatGPT’s Advanced Voice Mode last year, it feels like we’re somewhere much more compelling.
The company behind this is called Sesame, and it’s coming out of stealth today with an undisclosed amount of funding from Andreessen Horowitz, Spark Capital, and Matrix Partners —- all of which were big Oculus VR investors — with Oculus co-founder and former CEO Brendan Iribe, former Ubiquity6 CTO and co-founder Ankit Kumar, and former Meta Reality Labs research engineering director Ryan Brown in charge.
And the company says it’s building AI glasses to go along with its new voice assistant, too, ones “designed to be worn all day, giving you high-quality audio and convenient access to your companion who can observe the world alongside you.” So far, it’s only sharing a few small images of what look like early prototypes:
Sesame has a mini white paper you can read on its website, describing the model and its dataset of around one million hours of “publicly available audio.” It says it plans to both open source its models, and expand from just English to over 20 languages “in the coming months.”
Is this “crossing the uncanny valley of conversational voice,” as Sesame titles its blog post? Perhaps check it out and decide for yourself.
Meta has fired âroughly 20â employees who leaked âconfidential information outside the company,â according to a spokesperson.
“We tell employees when they join the company, and we offer periodic reminders, that it is against our policies to leak internal information, no matter the intent,â Meta spokesperson Dave Arnold tells The Verge exclusively. âWe recently conducted an investigation that resulted in roughly 20 employees being terminated for sharing confidential information outside the company, and we expect there will be more. We take this seriously, and will continue to take action when we identify leaks.”
Meta has ramped up its efforts to find leakers due to a recent influx of stories detailing unannounced product plans and internal meetings, including a recent all-hands led by CEO Mark Zuckerberg. After we and other outlets reported on what Zuckerberg said during that meeting, employees were warned not to leak. In comments that were subsequently leaked, CTO Andrew Bosworth then told them that the company was âmaking progress on catching people.â
PlayerUnknown Productions, the studio from PUBG creator Brendan Greene (aka PlayerUnknown), has shared a gameplay trailer for Prologue: Go Wayback!, a single-player survival game that’s launching in early access on Steam this year.
The studio previously shared that the game would be built with “machine-learning-driven terrain generation technology,” and in this new trailer, you can see some gorgeous, wooded areas on display. In the game, you’ll also have to deal with various types of weather, including pouring rain and heavy snow. The point of the game, according to a fact sheet, is to explore and survive so that you can find a weather station to call for help.
Prologue is just one of three games in the works from PlayerUnknown Productions. The “ultimate project,” dubbed Project Artemis, is going to be “a massive multiplayer sandbox experience” that builds on the technology featured in Prologue and a tech demo, Preface: Undiscovered World, that’s available on Steam.
