Engadget

Threads users may soon have a new way to clean up their timelines. The app is testing a new archive feature that can be used to manually archive individual posts or automatically hide posts after a set period of time, Adam Mosseri shared.

As on Instagram, archiving a post on Threads will hide it from public view, though the post will be available to the original author to view or undo. The Instagram head said the feature will be available to a “small number of people” to start. Though optional, Mosseri has made no secret that his preference is for all posts on the service to be somewhat ephemeral. “I think we should move to automatically archive posts on Threads after a month or so,” he wrote in a post back in February.

But, in a poll shared with that earlier post, Threads users overwhelmingly said they would “never” want their posts automatically hidden from public view. In his latest update, Mosseri noted that “the resounding feedback was not to make this the default” so the company will try out archiving “as an option” to start.

While archiving old posts is popular on Instagram, the feature could be somewhat more controversial on Threads. As the service grows and adds more features geared toward real-time information, posts that automatically archive could make it harder to go back and revisit the original source of an update. (If Mosseri, for example, were to automatically archive all his old posts, it would be much more difficult to track his evolving thoughts on Threads, which he often shares publicly on the platform.) At the same time, allowing posts to be ephemeral often encourages users to share more often, which is even more important for Threads’ continued growth.

This article originally appeared on Engadget at https://www.engadget.com/threads-is-testing-automatic-archiving-for-posts-184243484.html?src=rss

Several months after Meta brought WhatsApp passkey logins to Android, it’s doing the same on iOS. The company is rolling out the feature now, and it will be available to all users in the coming weeks.

Passkeys are seen as a more secure login method as (for one thing) it mitigates the risk of scammers convincing users to share their SMS passcode. Instead, you can log in by verifying a passkey using facial recognition, biometrics or a PIN stored on Apple’s passkey manager. Passkeys work by securely storing an authentication credential on your device that pairs with one saved by the service you want to log into.

To turn on passkeys for WhatsApp on iOS, go to the Account section of the app’s Settings, and then tap Passkeys. “Passkey verification will make logging back into WhatsApp easier and more secure. We’re excited to launch this on WhatsApp and give users an added layer of security,” Alice Newton-Rex, WhatsApp’s head of product, said in a statement.

This article originally appeared on Engadget at https://www.engadget.com/whatsapp-is-enabling-passkey-support-on-ios-182527962.html?src=rss

The Biden administration is tackling the monumental task of making America’s industrial freight system more environmentally friendly. The White House said on Wednesday that it aims to have 30 percent of industrial truck sales produce zero emissions by 2030 and 100 percent by 2040.

In addition to those non-binding targets, the White House is meeting on Wednesday with stakeholders from the commercial vehicle, shipping and infrastructure industries to help execute its agenda. The roundtable is designed to advance the Biden Administration’s goal of “supercharging the buildout of the infrastructure necessary to make a zero-emissions freight ecosystem a reality in the United States.”

Unsurprisingly, the freight industry uses a lot of energy and produces a lot of pollution to match. Bloomberg notes that the transportation sector emits about 29 percent of US greenhouse gas emissions, and freight (including shipping, trucking and trains) makes up about a third of that figure. So, you can ballpark that the American freight industry is responsible for roughly 10 percent of the country’s carbon emissions.

As part of the election-year rollout, the Biden Administration plans to ask the public to comment on charging infrastructure for heavy-duty vehicles, signaling that the specifics of the plan aren’t yet finalized. The White House wants to avoid a fragmented industrial EV charging system without a universally agreed-upon standard. The industry has seemingly settled on Tesla’s NACS as the de facto choice in the lightweight consumer sector.

Alongside the newly announced industrial goals, the Biden Administration’s Environmental Protection Agency (EPA) is opening up about $1 billion in Inflation Reduction Act (IRA) funding to replace Class 6 and 7 vehicles (school buses, garbage trucks and delivery trucks) with electric equivalents.

The IRA requires that at least $400 million of that funding goes to local communities hit the hardest by industrial pollution. The White House says 72 million Americans live near truck freight routes and bear the brunt of their short-term output. Sadly but unsurprisingly (given the nation’s history), people of color and those from low-income households are most likely to be heavily affected by high environmental toxin levels.

The White House’s goals are admirable, given the urgency of the global climate crisis and the freight industry’s role. However, one significant problem remains: These are voluntary, non-binding resolutions that could — and, given public comments, almost certainly would — be undone by a second Trump Administration, should the serial napper return to office next year. As with many other aspects of the nation’s and world’s future, US voters will decide the outcome this November.

This article originally appeared on Engadget at https://www.engadget.com/the-white-house-wants-a-zero-emission-freight-industry-by-2040-180401537.html?src=rss

Parents throughout the country just breathed a sigh of relief at the prospect of allowing more screen time for their kids that doesn’t revolve around some really loud and annoying YouTube dude. PBS just launched a new free ad-supported channel called PBS Retro, as reported by The Wrap. As the name suggests, this is a repository for all of your favorite edutainment classics from the 1980s, 1990s and beyond. The nostalgia is strong with this one.

PBS Retro is available via The Roku Channel, which is a streaming service on smart TVs and, of course, Roku devices. The 24/7 channel airs all of the shows you’re likely picturing right now, including Reading Rainbow and Mister Rogers’ Neighborhood. You’ll also be able to introduce your kids to Thomas and Friends, Zoboomafoo and Kratt's Creatures, among others. It’s a collection of old-school classics.

This isn’t the only PBS-adjacent channel available on Roku’s platform. It’s home to other ad-supported channels like PBS Antiques Roadshow, Antiques Road Trip, PBS Food, Julia Child and PBS Nature. There are also plenty of PBS-related channels available via subscription, including PBS Masterpiece, PBS Kids, PBS Living, PBS Documentaries and PBS America. Some of these are available on Hulu+ Live TV.

PBS may become an even bigger fixture in the free ad-supported streaming television (FAST) space in the future. The organization says it's in the “early days of experimentation” with a goal of making PBS content “available in new places.” It will “continue to consider additional opportunities” if these FAST channels increase revenue and brand awareness. In the meantime, Mister Rogers’ Neighborhood makes for some mighty fine ASMR.

This article originally appeared on Engadget at https://www.engadget.com/pbs-retro-is-a-new-fast-channel-playing-just-the-classics-173125500.html?src=rss

At CES this January, startup Rabbit unveiled its first device, just in time for the end of the year of the rabbit according to the lunar calendar. It’s a cute little orange square that was positioned as a “pocket companion that moves AI from words to action.” In other words, it’s basically a dedicated AI machine that acts kind of like a walkie talkie to a virtual assistant.

Sound familiar? You’re probably thinking of the Humane AI Pin, which was announced last year and started shipping this month. I awarded it a score of 50 (out of 100) earlier this month, while outlets like Wired and The Verge gave it similarly low marks of 4 out of 10.

The people at Rabbit have been paying close attention to the aftermath of the Humane AI Pin launch and reviews. It was evident in founder and CEO Jesse Lyu's address at an unboxing event at the TWA hotel in New York last night, where the company showed off the Rabbit R1 and eager early adopters listened rapturously before picking up their pre-orders. Engadget's sample unit is on its way to Devindra Hardawar, who will be tackling this review. But I was in attendance last night to check out units at the event that industry peers were unboxing (thanks to Max Weinbach for the assistance!).

What is the Rabbit R1?

As a refresher, the Rabbit R1 is a bright orange square, co-engineered by Teenage Engineering and Rabbit. It has a 2.88-inch color display built in, an 8-megapixel camera that can face both ways and a scroll wheel reminiscent of the crank on the Playdate. The latter, by the way, is a compact gaming handheld that was also designed by Teenage Engineering, and the Rabbit R1 shares its adorable retro aesthetic. Again, like the Humane AI Pin, the Rabbit R1 is supposed to be your portal to an AI-powered assistant and operating system. However, there are a few key differences, which Lyu covered extensively at the launch event last night.

Rabbit R1 vs Humane AI Pin

Let's get this out of the way: The Rabbit R1 already looks a lot more appealing than the Humane AI Pin. First of all, it costs $199 — less than a third of the AI Pin's $700. Humane also requires a monthly $24 subscription fee or its device will be rendered basically useless. Rabbit, as Lyu repeatedly reiterated all night, does not require such a fee. You'll just be responsible for your own cellular service (4G LTE only, no 5G), and can bring your own SIM card or just default to good old Wi-Fi. There, you'll also find the USB-C charging port.

The R1's advantages over the Pin don't end there. By virtue of its integrated screen (instead of a wonky, albeit intriguing projector), the orange square is more versatile and a lot easier to interact with. You can use the wheel to scroll through elements and press the button on the right side to confirm a choice. You could also tap the screen or push down a button to start talking to the software.

Now, I haven’t taken a photo with the device myself, but I was pleasantly surprised by the quality of images I saw on its screen. Maybe my expectations were pretty low, but when reviewers in a media room were setting up their devices by using the onboard cameras to scan QR codes, I found the images on the screens clear and impressively vibrant. Users won’t just be capturing photos, videos and QR codes with the Rabbit R1, by the way. It also has a Vision feature like the Humane AI Pin that will analyze an image you take and tell you what’s in it. In Lyu’s demo, the R1 told him that it saw a crowd of people at “an event or concert venue.”

Cherlynn Low for Engadget

We’ll have to wait till Devindra actually takes some pictures with our R1 unit and downloads them from the web-based portal that Rabbit cleverly calls the Rabbit Hole. Its name for camera-based features is Rabbit Eye, which is just kind of delightful. In fact, another thing that distinguishes Rabbit from Humane is the former’s personality. The R1 just oozes character. From the witty feature names to the retro aesthetic to the onscreen animation and the fact that the AI will actually make (cheesy) jokes, Rabbit and Teenage Engineering have developed something that’s got a lot more flavor than Humane’s almost clinical appearance and approach.

Of all the things Lyu took shots at Humane about last night, though, talk of the R1’s thermal performance or the AI Pin’s heat issues was conspicuously absent. To be clear, the R1 is slightly bigger than the Humane device, and it uses an octa-core MediaTek MT6765 processor, compared to the AI Pin’s Snapdragon chip. There’s no indication at the moment that the Rabbit device will run as hot as Humane’s Pin, but I’ve been burned (metaphorically) before and remain cautious.

I am also slightly concerned about the R1’s glossy plastic build. It looks nice and feels lighter than expected, weighing just 115 grams or about a quarter of a pound. The scroll wheel moved smoothly when I pushed it up and down, and there were no physical grooves or notches, unlike the rotating hinge on Samsung’s Galaxy watches. The camera housing lay flush with the rest of the R1’s case, and in general the unit felt refined and finished.

Most of my other impressions of the Rabbit R1 come from Lyu’s onstage demos, where I was surprised by how quickly his device responded to his queries. He was able to type on the R1’s screen and tilted it so that the controls sat below the display instead of to its right. That way, there was enough room for an onscreen keyboard that Lyu said was the same width as the one on the original iPhone.

What’s next for the Rabbit R1?

Rabbit also drew attention for its so-called Large Action Model (LAM), which acts as an interpreter to convert popular apps like Spotify or Doordash into interfaces that work on the R1’s simple-looking operating system. Lyu also showed off some of these at the event last night, but I’d much rather wait for us to test these out for ourselves.

Lyu made many promises to the audience, seeming to acknowledge that the R1 might not be fully featured when it arrives in their hands. Even on the company’s website, there’s a list of features that are planned, in the works or being explored. For one thing, an alarm is coming this summer, along with a calendar, contacts app, GPS support, memory recall and more. Throughout his speech, Lyu repeated the phrase “we’re gonna work on” amid veiled references to Humane (for instance, emphasizing that Rabbit doesn’t require an additional subscription fee). Ultimately, Lyu said “we just keep adding value to this thing,” in reference to a roadmap of upcoming features.

Hopefully, Lyu and his team are able to deliver on the promises they’ve made. I’m already very intrigued by a “teach mode” he teased, which is basically a way to generate macros by recording an action on the R1, and letting it learn what you want to do when you tell it something. Rabbit’s approach certainly seems more tailored to tinkerers and enthusiasts, whereas Humane’s is ambitious and yet closed off. This feels like Google and Apple all over again, except whether the AI device race will ever reach the same scale remains to be seen.

Last night’s event also made it clear what Rabbit wants us to think. It was hosted at the TWA hotel, which itself used to be the head house of the TWA Flight Center. The entire place is an homage to retro vibes, and the entry to Rabbit’s event was lined with display cases containing gadgets like a Pokedex, a Sony Watchman, a Motorola pager, Game Boy Color and more. Every glass box I walked by made me squeal, bringing up a pleasant sense memory that also resurfaced when I played with the R1. It didn't feel good in that it's premium or durable; it felt good because it reminded me of my childhood.

Whether Rabbit is successful with the R1 depends on how you define success. The company has already sold more than 100,000 units this quarter and looks poised to sell at least one more (I’m already whipping out my credit card). I remain skeptical about the usefulness of AI devices, but, in large part due to its price and ability to work with third-party apps at launch, Rabbit has already succeeded in making me feel like Alice entering Wonderland.

This article originally appeared on Engadget at https://www.engadget.com/rabbit-r1-hands-on-already-more-fun-and-accessible-than-the-humane-ai-pin-163622560.html?src=rss

Google keeps promising to phase out third-party cookies on Chrome but not actually doing it. The company vowed to deprecate cookies back in 2020, pushing the date back to 2023 and then 2024. We did get some traction earlier this year, when Google disabled cookies for one percent of Chrome users, but those efforts have stalled. Now, the company says it won’t happen until next year.

It’s easy to drag Google for this but it’s not entirely in the company’s hands. The tech giant is working closely with the UK’s Competition and Markets Authority (CMA) to ensure that any tools it implements to replace the cookie’s tracking and measurement capabilities aren’t anti-competitive. These tools are known collectively as the Privacy Sandbox and Google says it has to wait until the CMA has had “sufficient time to review” results from industry tests that’ll be provided by the end of June.

Google’s Privacy Sandbox has stirred up some controversy in recent years. The proposed tools have drawn complaints from adtech companies, publishers and ad agencies, on the grounds that they are difficult to operate, don’t adequately replace traditional cookies and give too much power to Google. To that end, the company said that it recognizes “ongoing challenges related to reconciling divergent feedback from the industry, regulators and developers.” This is another reason given for the delay until next year.

The CMA isn’t the only regulatory agency giving the side-eye to the current iteration of these Privacy Sandbox tools. The UK-based Information Commissioner’s Office drafted a report that indicated these tools could be used by advertisers to identify consumers, as suggested by the Wall Street Journal.

Those in the ad industry want to see cookies given the heave-ho, despite complaints about Privacy Sandbox. Drew Stein, CEO of adtech data firm Audigent, told Engadget that it’s time for Google “to deliver on the promise of a better ecosystem” by implementing its plans to eliminate third-party cookies.

The CMA, on the other hand, has indicated a willingness to keep third-party cookies in play, particularly if Google’s solution does more harm than good. Craig Jenkins, the CMA’s director of digital markets, recently said the organization would delay implementation of Privacy Sandbox tools if “we’re not satisfied we can resolve the concerns”, as reported by Adweek. We’ll see what happens in 2025.

This article originally appeared on Engadget at https://www.engadget.com/google-has-delayed-killing-third-party-cookies-from-chrome-again-155911583.html?src=rss

The bill that will force a sale or ban of TikTok in the United States is now law. President Joe Biden signed a package of foreign aid bills that included the “Protecting Americans from Foreign Adversary Controlled Applications Act,” one day after the legislation was approved by the Senate.

In a statement, TikTok said it would challenge the law in court, which could delay an eventual sale or ban. “This unconstitutional law is a TikTok ban, and we will challenge it in court,” the company said. “We believe the facts and the law are clearly on our side, and we will ultimately prevail. The fact is, we have invested billions of dollars to keep U.S. data safe and our platform free from outside influence and manipulation. This ban would devastate seven million businesses and silence 170 million Americans.”

The law gives TikTok’s parent company ByteDance, which is based in China, up to a year to sell the app to a new owner. If the company fails to divest, then TikTok will be banned from US app stores and web hosting services.

Unlike previous attempts to force a sale or ban of the app, the “Protecting Americans from Foreign Adversary Controlled Applications Act,” had overwhelming bipartisan support and was able to move through Congress with remarkable speed. The original version of the bill, which called for a six-month window to divest, passed the House in March, just days after it was introduced. An updated version, which allows up to 12 months for a divestment, passed over the weekend.

In a video shared on TikTok, CEO Shou Chew called it a “disappointing moment” for the company. “Make no mistake, this is a ban on TikTok and a ban on you and your voice,” he said. “It's actually ironic because the freedom of expression on TikTok reflects the same American values that make the United States a beacon of freedom.”

This article originally appeared on Engadget at https://www.engadget.com/joe-biden-signs-the-bill-that-could-ban-tiktok-in-the-united-states-154106950.html?src=rss

Valve has closed a loophole in Steam's refund policy that effectively allowed players to beat a game before its official release date and get their money back. The company has long had a policy in place whereby users could refund a game, no questions asked, as long as they haven’t played for more than two hours within 14 days. Until now, the refund policy was far more lenient for games in early access or advanced access, but Valve has nipped that in the bud.

"When you purchase a title on Steam prior to the release date, the two hour playtime limit for refunds will apply (except for beta testing), but the 14-day period for refunds will not start until the release date," Valve's updated policy reads, as noted by IGN. "For example, if you purchase a game that is in Early Access or Advanced Access, any playtime will count against the two-hour refund limit. If you pre-purchase a title which is not playable prior to the release date, you can request a refund at any time prior to release of that title, and the standard 14-day/two-hour refund period will apply starting on the game’s release date."

Early access enables players to try an incomplete version of a game. It’s helpful for developers as they can take feedback from players and use that to improve their project before ramping up the marketing campaign ahead of the official release. Supergiant famously used this strategy to tremendous success with Hades (and is perhaps looking to repeat that trick with the sequel). But a game can remain in early access for years. Under the previous policy, players could put many hours into an early access game and still claim a refund on Steam.

As for advanced access, that relates to playing a full version of a game before its proper debut. It's pretty common for publishers to sell a deluxe edition of a game that lets players dive in a few days early. However, Steam made it possible for someone to beat a game in advanced access and get their money back before the standard version was available to everyone.

Now, the two-hour time limit applies to games in early access and advanced access. There's also a new advanced access label to make it clear when a game offers that.

There is one other key issue with the otherwise generous two-hour refund policy that Steam hasn't fully addressed, however. It's not uncommon for players to roll credits on very short games, typically indie titles, and then get their money back. That leaves the game's developer and publisher out of pocket.

This article originally appeared on Engadget at https://www.engadget.com/steam-closes-an-early-access-loophole-in-its-refund-policy-150003143.html?src=rss

Last fall, Qualcomm revealed a major upgrade for its laptop chips with the Snapdragon X Elite. And while we’re still waiting for those processors to make their way into retail devices, today Qualcomm is expanding the line with the Snapdragon X Plus, which I had a chance to test out ahead of its arrival on gadgets later this year.

Similar to the X Elite, the X Plus is based on the same 4nm process and Arm-based Oryon CPU architecture as its sibling. The difference is that the new chip is meant to be used in slightly more affordable mainstream laptops, and as such it only has 10 CPU cores (vs 12 for the X Elite) and reduced clock speeds (3.4Ghz vs 3.8Ghz for the X Elite). This positioning is a lot like what Qualcomm’s rivals have been doing for a while, with the X Elite serving as the flagship chip (like Intel’s Core Ultra 9 series) and the X Plus sitting just below that (which would be equivalent to the Core Ultra 7 line).

Qualcomm

However, one thing that hasn’t changed is that just like the X Elite, the X Plus’ Hexagon NPU puts out the same 45 TOPS of machine learning performance. This is particularly notable as Microsoft recently suggested that laptops would require at least 40 TOPS in order to run various elements of its Copilot AI service on-device. Qualcomm is also making some big claims regarding power efficiency, with the X Plus chip said to deliver 37 percent faster CPU performance compared to an Intel Core Ultra 7 155H when both chips are running at the same wattage. And when put up against other Arm-based chips, Qualcomm says the X Plus is 10 percent faster than Apple’s M3 processor in multi-threaded CPU tasks.

Photo by Sam Rutherford

Unfortunately, the X Plus is not expected to show up in retail devices until sometime in the second half of 2024. That said, at a hands-on event, I was able to run a few benchmarks on some early Qualcomm-built reference devices. And to my pleasant surprise, the X Plus performed as expected with multi-core scores in Geekbench of 12,905 and multi-thread performance in Cinebench 2024 of 852. (Note: Because the processor has not been released yet, there’s an error in Cinebench that results in the chip’s GPU incorrectly being listed as from the X Elite instead of the X Plus.)

This is a promising showing for Qualcomm’s second and less expensive chip featuring its Oryon architecture. Though as always, the real test will come when the X Plus starts showing up in proper retail hardware. That’s because even if it boasts impressive benchmark figures, these processors will still need to play nicely with Windows, which has not had nearly as smooth a transition to Arm-based silicon as Apple’s macOS.

Photo by Sam Rutherford/Engadget

But with renewed support for Windows on Snapdragon PCs and Qualcomm recently working with major players like Google to bring “dramatic performance improvements” in Chrome for devices running its laptop chips, things may be smoother this time.

This article originally appeared on Engadget at https://www.engadget.com/qualcomm-is-expanding-its-next-gen-laptop-chip-line-with-the-snapdragon-x-plus-130018288.html?src=rss

It used to be that you could pay for a retail version of Windows 11 and expect it to be ad-free, but those days are apparently finito. The latest update to Windows 11 (KB5036980) comes out this week and includes ads for apps in the "recommended" section of the Start Menu, one of the most oft-used parts of the OS.

"The Recommended section of the Start menu will show some Microsoft Store apps," according to the release notes. "These apps come from a small set of curated developers." 

The app suggestions are enabled by default, but you can restore your previously pristine Windows experience if you've installed the update, fortunately. To do so, go into Settings and select Personalization > Start and switch the "Show recommendations for tips, app promotions and more" toggle to "off."

The new "feature" arrives just weeks after it appeared as an Insider beta, showing how quickly Microsoft can implement things when it wants to. It certainly wasn't enough time to receive the kind of user feedback the Insider program is designed for.

The update is bound to rub customers the wrong way, considering that Windows 11 starts at $139 for the Home version. While removing it isn't a huge deal, it may also remind folks of the needless time they spent stripping bloatware from OEM Windows installations. Microsoft previously tested ads in the Windows 11 File Explorer, but ended the experiment shortly afterward.

This article originally appeared on Engadget at https://www.engadget.com/windows-11-now-comes-with-its-own-adware-124531977.html?src=rss

The US Federal Trade Commission (FTC) has banned noncompete clauses in a move to "drive innovation" and protect workers' rights and wages, the regulator said in a press release. The new rule will free most new and current employees from such agreements, with the exception of "policy-making" executives earning more than $151,164 per year. 

"Noncompete clauses keep wages low, suppress new ideas, and rob the American economy of dynamism," said FTC Chair Lina M. Khan. The agency estimated that the new rule will allow the creation of 8,500 new businesses each year, increase worker earnings by $524 per year and lower health care costs by $194 billion over the next decade. 

Noncompete clauses, widely used in the tech industry, keep employers from freely changing to similar jobs or starting a business in the same field. The result is that workers must often stay in jobs they don't want, switch to a lower-paid position, relocate, or defend against costly litigation. "An estimated 30 million workers — nearly one in five Americans [in the workforce] —are subject to a noncompete," according to the FTC.

The Commission found that noncompetes tend to negatively affect competitive conditions in labor markets by inhibiting efficient matching between workers and employers. There is also evidence that noncompetes lead to increased market concentration and higher prices for consumers.

Companies must now cancel existing noncompete clauses and notify employees about the change. The ruling applies to most employees and future hires, but current deals with senior executives still apply on the grounds that such agreements are likely to have been agreed upon by both parties. 

Tech companies ostensibly use noncompetes as a way to protect IP, but they function in reality to lock in workers. The FTC said that trade secret laws and non-disclosure agreements (NDAs) are a better way to protect IP, and "employers that wish to retain employees can compete on the merits for the worker's labor services by improving wages and working conditions."  

Microsoft, the third largest tech industry employer in the US, eliminated such clauses back in 2022. "While our existing employee agreements have noncompete obligations, we do not endorse the use of such provisions as a retention tool," the company said at the time. 

The FTC vote went 3 to 2 along party lines. Republican commissioner Melissa Holyoke said the Commission "overstepped the boundaries of its power" and estimated the ruling would be challenged in court and struck down. 

This article originally appeared on Engadget at https://www.engadget.com/ftc-bans-employers-from-using-noncompete-clauses-123045777.html?src=rss

Finding the right gift for a new graduate in your life can be tough. Default ideas like a fancy watch or a monogrammed bag can be pricey and, let’s face it, boring. Tech can be a good option, and no, you don’t have to spend a fortune to get a solid gadget that they’d like. While expensive stuff like iPhones, smartwatches and game consoles may come to mind immediately, they aren’t the only options out there. And sometimes all it takes is a practical gadget like a power bank to make someone’s life a little easier — that’s especially applicable to new grads who are focusing all of their attention on snagging that new job or applying for further education. Here’s Engadget’s list of the best gifts under $50 for new graduates.

This article originally appeared on Engadget at https://www.engadget.com/best-gifts-for-grads-under-50-114506320.html?src=rss

The Mercedes-Benz G-Class, aka the Gelandewagen (which means "all-terrain vehicle" in German) has been in regular production since 1979. It's changed a lot since then, evolving from a utilitarian off-roader to a desirable luxury icon, but it's never seen a change quite like this.

Meet the Mercedes-Benz G 580 with EQ Technology. That's a long and mild name for a pretty radical reinvention of the classic G. This is the first battery-powered G-Wagen, driven by a whopping four electric motors that draw juice from a 116-kilowatt-hour battery pack. Total output is 580 horsepower.

Those four motors enable the electric G to do some amazing things, like a so-called G-Turn, where it spins in place. The idea is to help this big rig get out of some tight off-road situations, but we think it'll be even more effective at impressing your neighbors. There's a bevy of other tech here too, including dedicated off-road driving modes and a series of cameras that allow drivers to spot every rock and rut when crossing the trails. No formal word on pricing just yet, but it's important to note that the gas-powered versions of the G-Class live on, so if you're really attached to internal combustion there's still a G for you.

Really, though, if you can't do a tank turn, then what's the point? Watch the video above for the full story.

This article originally appeared on Engadget at https://www.engadget.com/mercedes-benz-quad-motor-g-class-could-be-the-ultimate-ev-off-roader-120024168.html?src=rss

The Senate approved a measure that will require ByteDance to sell TikTok or face a ban, in a vote of 79 to 18. The Protecting Americans from Foreign Adversary Controlled Applications Act next goes to President Biden. The president has already said he’ll sign the bill into law. (Yes, as predicted, I'm writing about this again.)

TikTok has faced the ire of US politicians for a few years now, but this bill has picked up support across both political parties. It sailed through the House of Representatives before being approved (bundled with a package for foreign aid) by the Senate on Tuesday.

The bill states that TikTok would have up to 12 months to divest from its parent company ByteDance, or face a ban in US app stores and web hosting services. The company, naturally, has protested this push, calling the bill unconstitutional and vowing to mount a legal challenge if the bill is signed into law. If it does so, it could bounce around courts for years before any eventual ban, if the company declines to sell. A few years is a long time in social media. Ask Snap, or worse, Vine.

And who would buy TikTok? While many major tech companies might love to grab the social network’s engaged young audience, many politicians would balk at making a Big Tech company even bigger. Steve Mnuchin, who was Treasury secretary in the Trump administration, told CNBC he was putting together an investor group. What could go wrong?

— Mat Smith

The biggest stories you might have missed

X, for some reason, has a TV app now

The best travel gear for graduates

Adobe Photoshop’s latest beta makes AI-generated images from simple text prompts

​​You can get these reports delivered daily direct to your inbox. Subscribe right here!

Microsoft’s new lightweight AI model can run on smartphones It can supposedly respond similarly to models 10 times its size.

Microsoft has unveiled its latest light AI model, called the Phi-3 Mini, for smartphones and other local devices. The aim is to provide a cheaper alternative to cloud-powered large language models (LLMs), allowing smaller organizations to adopt AI, with presumably lower energy burdens and without heady processing costs. According to Microsoft, the new model handily outperforms its previous Phi-2 small model and is on par with larger models like Llama 2. In fact, the company says the Phi-3 Mini responds close to the level of a model 10 times its size. The trick is apparently in the data Microsoft used to train its tiny model.

Continue reading.

Tesla teases its ride-hailing app ahead of robotaxi unveiling The feature would be available in the Tesla app ‘in the future.’

Tesla teased ride-hailing features coming to its app ahead of an August robotaxi unveiling. The company released mock-ups of the upcoming feature, which showed the ability to “summon” a ride from the Tesla app. The company has been promising self-driving taxi services for years. Tesla didn’t offer many details, but it seems to have Uber-like functionality and the ability to remotely set the car’s temperature before arrival.

Continue reading.

Ray-Ban Meta smart glasses get improved AI smarts that don’t require a subscription Take that, Humane. Meta

After a few months of testing, Meta is bringing multimodal AI to its smart glasses. Multimodal AI means the system can process multiple types of information, including photos, videos, text and audio. You might have seen feature showcases of AI-connected devices that can view what a device is looking at and offer extra information — that kind of thing. Meta also announced hands-free video call integration with WhatsApp and Messenger and a few more frame designs.

Continue reading.

This article originally appeared on Engadget at https://www.engadget.com/the-morning-after-senate-passes-the-bill-that-could-ban-tiktok-111556543.html?src=rss

Back in 2022, Mercedes-Benz announced that it was going to release an electric G-Class by the end of 2024. Now, the automaker has formally introduced the electrified version of its iconic luxury vehicle that's known for its exclusivity. The Mercedes-Benz G 580 with EQ Technology comes equipped with a 116 kWh lithium-ion battery that gives it enough energy to run for up to 473 kilometers (294 miles) on a single charge. It also has a maximum total output of 432 kW and a maximum torque of 1,164 Nm.

The electric model looks pretty much like the gas-powered G-Class, with its sharp angles and its distinctive door handles, grilles and round headlights. Mercedes offers optional lighting exclusive to the EQ version of the vehicle, though, so you can change its looks and give it a design that's considerably different from a standard G-Class. The vehicle is powered by four electric motors located near each wheel and has several modes for off-road use: G-TURN, which will allow you to turn the vehicle almost on the spot, G-STEERING, which could eliminate the need for multi-point turns, and the intelligent off-road crawl function that provides cruise control for off-road drives. 

Mercedes-Benz made sure the vehicle's battery is ready for off-road journeys, as well, and encased it in a torsion-resistant casing that protects it from water and dirt. Since the vehicle can be driven in up to 33.5 inches of water, the battery definitely needs that kind of protection. It can charge with both alternating current and direct current, and when plugged into a fast-charging DC system, it can go from 10 to 80 percent in just 32 minutes. The G-Class can convert kinetic energy into electrical energy when you coast or hit the brakes, as well. 

EDITION ONE, the G-Class with EQ Technology coming out at launch later this year, will have an "an expanded palette of standard features." A company spokesperson told TechCrunch that a range-extended version with a battery that uses a more energy-dense silicon-anode technology from Mercedes' partner Silas will also be available in the next few years.

This article originally appeared on Engadget at https://www.engadget.com/mercedes-benz-finally-unveils-its-electric-g-class-luxury-off-roader-110040316.html?src=rss

The European Union has adopted a right-to-repair directive that will make it easier for consumers to get their devices fixed. The new rules extend a product's guarantee if it breaks under warranty, while obliging manufacturers to repair devices no longer covered. The law still needs to be approved by member nations. 

Devices sold in Europe already offer minimum two-year warranties, but the new rules impose additional requirements. If a device is repaired under warranty, the customer must be given a choice between a replacement or a repair. If they choose the latter, the warranty is to be extended by a year. 

Once it expires, companies are still required to repair "common household products" that are repairable under EU law, like smartphones, TVs and certain appliances (the list of devices can be extended over time). Consumer may also borrow a device during the repair or, if it can't be fixed, opt for a refurbished unit as an alternative.

The EU says repairs must be offered at a "reasonable" price such that "consumers are not intentionally deterred" from them. Manufacturers need to supply spare parts and tools and not try to weasel out of repairs through the use of "contractual clauses, hardware or software techniques." The latter, while not stated, may make it harder for companies to sunset devices by halting future updates. 

In addition, manufacturers can't stop the use of second-hand, original, compatible or 3D-printed spare parts by independent repairers as long as they're in conformity with EU laws. They must provide a website that shows prices for repairs, can't refuse to fix a device previously repaired by someone else and can't refuse a repair for economic reasons.

While applauding the expanded rules, Europe's Right to Repair group said it there were missed opportunities. It would have liked to see more product categories included, priority for repair over replacement, the right for independent repairers to have access to all spare parts/repair information and more. "Our coalition will continue to push for ambitious repairability requirements... as well as working with members focused on the implementation of the directive in each member state."

Along with helping consumers save money, right-to-repair rules help reduce e-waste, CO2 pollution and more. The area is currently a battleground in the US as well, with legislation under debate in around half the states. California's right-to-repair law — going into effect on July 1 — forces manufacturers to stock replacement parts, tools and repair manuals for seven years for smartphones and other devices that cost over $100.

This article originally appeared on Engadget at https://www.engadget.com/eus-new-right-to-repair-rules-force-companies-to-repair-out-of-warranty-devices-081939123.html?src=rss

Most microSD cards are fast enough for stashing photos, recording video and transferring files, but some will get you a little more bang for your buck than others. If you’re looking to boost the internal storage of your Nintendo Switch, Steam Deck, GoPro, Raspberry Pi or any other device that still has compatibility with microSD cards, we can help. We’ve thoroughly researched the market and put more than a dozen microSD cards through a range of benchmark tests: Here are the best that we’ve tested, along with some general advice on what to look for when buying a new card.

Editor’s note (4/24/24): Samsung has released an upgraded version of our “best value” pick, the Evo Select. We plan to test it soon and update our guide accordingly. Samsung says the new model can reach sequential read speeds up to 160 MB/s, so it should still be a step down from our other top picks, which we continue to recommend

What to look for in a microSD card Capacity

The first thing to figure out when buying a microSD card is how much space you need. Modern cards are usually available in sizes ranging from 32GB to 512GB, while a handful offer 1TB models as well. 2TB cards are theoretically possible but still in the prototype stage.

For most, a 128GB or 256GB model should be the sweet spot between price and storage space. But if you need more room — say, for stashing a bunch of games on a Steam Deck — a 512GB card could make more sense and often provides a better cost-per-GB ratio. These days, you can find a decent 128GB card for around $15 or less, a good 256GB card for less than $30 and a solid 512GB card for less than $40 (though many cost closer to $50 or $60). There’s a starker increase when you go up to 1TB cards, which typically cost closer to $100.

Note that a microSD card’s performance may differ depending on what capacity you buy. SanDisk says its 128GB Extreme card delivers sequential write speeds up to 90 MB/s, for example, while the higher-capacity models in the same line offer up to 130 MB/s.

When we talk about microSD cards today, we usually refer to cards that use the microSDXC (eXtended Capacity) standard, which have a capacity between 32GB and 2TB. Your device needs to support this for it to work with a microSDXC card. This will almost never be an issue these days, but some older devices (a Nintendo 3DS, for instance) are only compatible with microSDHC (High Capacity) cards, which range from 2GB to 32GB.

Read and write speeds

MicroSD cards are primarily judged on their read and write speeds, which are usually measured in megabytes per second (MB/s). Generally, most microSD cards have faster read speeds than write speeds.

These metrics can then be broken down into sequential and random performance. Sequential read and write speeds matter when you’re trying to access or save long, constant streams of data, such as opening a large video or copying a big batch of files from a PC. If you want to use a microSD card for media storage, this is particularly important. Random performance, meanwhile, is about how quickly a card can read and write small files scattered throughout the device.

Since random read/write speeds are much lower than sequential ones, storage device makers tend not to advertise them as loudly. But they’re important if you use a card with a gaming device or a single-board computer like the Raspberry Pi, where it often has to rapidly save and access small bits of data in random locations.

SD Association Speed ratings

If you look at a microSD card, you’ll see a buffet of numbers, letters and symbols. Most of these refer to the card’s speed class and performance ratings, which are determined by the SD Association.

A card’s Video Speed Class, or V-rating, details its minimum sequential write speed, which is especially important when recording video from a camera. It ranges from V6 to V90. Most of the cards we tested had a V30 rating, so they have a sequential write speed of at least 30 MB/s. This should be enough to support up to 4K video at lower bitrates. Higher-rated V60 and V90 cards are usually better for capturing 8K, but they come at a much higher cost.

The UHS Speed Class, or U-rating, also refers to a card’s minimum sequential write speed. It comes in two varieties: U3, which mandates a minimum of 30 MB/s, and U1, which is rated for 10 MB/s.

The older Speed Class rating overlaps with the other two systems. It’s signified by a C symbol and goes from Class 2 to Class 10, with the number (again) indicating minimum sequential write speed. This rating is less relevant nowadays, but you may still see a “C10” logo on some cards.

The Application Performance spec, marked by an A symbol, is an indicator of random read/write speeds. This is measured in IOPS, or input/output operations per second, rather than MB/s. There are two categories here: A1 cards offer a minimum random read speed of 1,500 IOPS and a minimum random write speed of 500 IOPS, while A2 cards bump those up to 4,000 IOPS and 2,000 IOPS, respectively. Both ratings also guarantee sequential write speeds of at least 10 MB/s.

To keep it simple, most people should look for a card with V30, U3 and A2 ratings. It’s totally possible to get a solid card without those: A U1 card might be worth it if you just need a cheap, high-capacity option, for example. V60 and V90 cards are worth a look if you’re serious about shooting high-resolution photos and video as well. But overall, cards with the certifications above should provide the best blend of price and performance today.

It’s important to emphasize that these ratings are baselines. Most V30 cards offer significantly higher write speeds than 30 MB/s, for instance, and some A1 cards can outperform some A2 models in practice. The speeds advertised by manufacturers aren’t always 100 percent accurate, either: Sometimes the card will be slower in real-world use, other times it may actually be a bit faster.

UHS bus speeds

The other spec to note is the card’s bus interface. Most microSD cards available today are UHS-I, which have a theoretical maximum speed of 104 MB/s. There are also UHS-II cards, which have an extra row of pins on the back and can reach up to 312 MB/s. (A UHS-III standard exists as well but hasn’t seen wide adoption.) These are labeled on the card with a Roman numeral I or II.

The fastest microSD cards you can buy right now are UHS-II cards, and they’re usually the ones with V60 or V90 ratings. If you shoot lots of 4K to 8K video or frequently use burst mode to capture ultra high-res photos, the performance gains of a good UHS-II card can save you time.

But these are much more expensive than UHS-I cards: This 128GB Lexar Professional Gold model, for instance, is a relative bargain at $40. While that's less than many UHS-II models we’ve seen in the past, it's still more than double the typical price of our top pick mentioned below. You need a device that’s compatible with the UHS-II interface to see any benefits, too, and stock for UHS-II cards is generally spottier. For now, the higher speeds aren’t worth the price premium for most people, so we stuck mostly to UHS-I cards here.

We’ll also note Samsung’s recent announcement of a new 256GB microSD card based on an entirely different standard called SD Express. This is theoretically much faster than any UHS-I or UHS-II option: Samsung claims this specific model can reach sequential read speeds up to a whopping 800 MB/s. For context, that’d be quicker than some older SSDs. SD Express has technically been around for several years but hasn’t seen wide adoption, so exactly how this card performs in reality — and how much it’ll cost — remains to be seen. We’ll test the device once it becomes available, which Samsung says will be sometime later in 2024.

Kris Naudus / Engadget A note on card readers, the Nintendo Switch and Steam Deck

While the UHS-I spec has a theoretical maximum of 104 MB/s, some UHS-I cards can exceed that speed through proprietary extensions. You need a compatible card reader and host device to take advantage of that extra performance, though. If you find a UHS-I card advertising speeds higher than 104 MB/s, this is what’s going on. You can see these limits in action with a Nintendo Switch or Steam Deck: Both of those gaming devices support the UHS-I interface and don’t go beyond its official speed, flattening any sequential gains some cards may have elsewhere. The broader takeaway: Your microSD card will only be as fast as the slowest link in your chain.

Warranty

Many microSD cards are designed to be durable, with protection from water, extreme temperatures, X-rays and drops. Still, in case of catastrophe, a long warranty is always good to have. Many manufacturers offer lifetime limited warranties, though we’ve noticed that “endurance” cards marketed to withstand more hours of writing are usually covered for a much shorter period of time.

Avoiding counterfeits

The memory card market has had a particular problem with scammers selling fake products. To guard against this, only buy from a known brand and a reputable retailer such as Best Buy, B&H Photo or Adorama. If you shop at Amazon, only buy if the shipper and seller is Amazon.com. (Though a handful of users have reported receiving counterfeits even from Amazon directly in the past.) Remember: If a price seems too good to be true, it probably is. Be wary of any retailer offering significantly a lower price than others.

Once you receive a card, check its packaging for any irregularities. You can run benchmark tests like CrystalDiskMark or BlackMagic Disk Speed Test to verify its speeds aren’t drastically lower than what’s advertised (or possible, given its specs). You can also use software that’s designed to verify the true capacity and performance of your card, such as H2testw and FakeFlashTest.

How we tested

We put 13 microSD cards through a series of tests to verify their sequential and random performance. These included benchmarks like CrystalDiskMark, BlackMagic Disk Speed Test, ATTO Disk Benchmark and AJA System Test, as well as a few “real-world” tests. We copied and pasted a small folder of photos about 1.15GB in size to and from each card, then did the same with a larger 12.2GB folder containing multiple file types and subfolders, timing the process each time. We also checked how each card performed on the Steam Deck, downloading games of varying sizes — including Stardew Valley, Aperture Desk Job, Metal Gear Rising: Revengeance and Apex Legends — then timing how long it took to launch each game and load save files.

We used a Kingston USB 3.2 UHS-II reader to test each card on both Windows 11 and macOS Sonoma. For the former, we used an Alienware gaming PC with an Intel Core i9-10900F, Nvidia GeForce RTX 3080 GPU, 32GB of RAM and a 1TB SSD. For the latter, we used a 2021 16-inch MacBook Pro with an Apple M1 Pro chip, 16GB of RAM and a 512GB SSD. To use our card reader on the MacBook, we used Apple’s USB-C Digital AV Multiport Adapter.

We tested the 128GB version of each card wherever possible, though for a few cards — SanDisk's Extreme and Samsung's Pro Plus and Pro Ultimate — we were only able to test their 256GB models. We also reformatted each card before testing with the SD Association’s Memory Card Formatter tool.

Other notable microSD cards The SanDisk Extreme and PNY Elite-X.Photo by Jeff Dunn / Engadget Delkin Devices Power

We didn’t use it ourselves, but if you’re willing to pay for a more powerful UHS-II card built for heavy-duty video recording, the Delkin Devices Power has tested well elsewhere and should deliver significantly faster sequential write speeds than our picks above. It’s one of the few UHS-II cards we could actually find in stock, but it costs a ton, with a 128GB model normally priced at $93.

SanDisk microSDXC Card for Nintendo Switch

The SanDisk microSDXC Card for Nintendo Switch is another decent option if you ever see it available for less than the Pro Plus and Canvas Go Plus. Its sequential read speeds were about the same as the latter in our benchmarks, but its sequential writes were slightly slower (and farther behind the Pro Plus). Its random read/writes were a bit behind according to CrystalDiskMark as well. The 128GB version of this card is priced around $18 as of this writing, while the 512GB model costs around $50; both are higher than the Pro Plus. SanDisk backs the card with a lifetime warranty, however, plus it’s available in a 1TB capacity. Note that we tested the Apex Legends version of the 128GB card; SanDisk also sells a Super Mario model, but we can’t speak to whether that one performs any differently.

SanDisk's Nintendo Switch-branded microSDXC card.Photo by Jeff Dunn / Engadget Lexar Professional 1066x

Similarly, the Lexar Professional 1066x is a decent alternative to the Pro Plus if our main picks are unavailable. It’s another V30, U3 and A2 model, and like Samsung’s card, its sequential write speeds were a bit faster than the Canvas Go Plus and SanDisk Switch card in our benchmarks. Those write speeds weren’t as quick as the Pro Plus, and its sequential reads trailed all three cards. Random read/writes also lagged behind the Pro Plus and Canvas Go Plus, and we noticed its speeds peak and dip more noticeably in our file transfer test. Still, it’s not slow, so depending on its price, it might be a good bargain if you need fast write speeds for video recording and the like. It’s also available in 1TB, plus it comes with a lifetime limited warranty.

SanDisk Extreme

The SanDisk Extreme effectively matched the Pro Plus in a few of our sequential write tests and delivered sequential reads about on par with the Canvas Go Plus, but that was partly due to us only being able to secure the card’s 256GB model, which is higher-rated than the 128GB version. It’s a fine choice if you see it on sale at a reputable seller, but its random performance lagged behind the Samsung Pro and Kingston cards, and it typically costs more. It does have a 1TB model, though.

SanDisk Extreme Pro

The SanDisk Extreme Pro is a rival to the Samsung Pro Ultimate but, as of this writing, is either unavailable at most trusted retailers or priced too high by comparison.

PNY Elite-X

The PNY Elite-X wasn’t too far off the random read/write performance of the Pro Plus and Pro Ultimate in CrystalDiskMark, and it often goes for cheap, but it was well behind in our sequential tests.

This article originally appeared on Engadget at https://www.engadget.com/best-microsd-card-130038282.html?src=rss

A bill that could ban TikTok is now all but certain to become law. The Senate approved a measure that requires ByteDance to sell TikTok or face a ban, in a vote of 79 - 18. The “Protecting Americans from Foreign Adversary Controlled Applications Act,” will next head to President Joe Biden, who has said he would sign the bill into law.

While it’s far from the first effort to force a ban or divestment of the social media app, the bill managed to draw far more support than previous attempts. The bill was introduced in March and sailed through the House of Representatives with overwhelming bipartisan agreement. A slightly revised version was approved as part of a package of foreign aid legislation on Saturday.

Under the updated terms, TikTok would have up to 12 months to divest from parent company ByteDance or face a ban in US app stores and web hosting services. The company has called the bill unconstitutional and indicated it would mount a legal challenge to such a law, which could further delay an eventual sale or ban.

The company didn't immediately respond to a request for comment.

TikTok has long been viewed with suspicion by lawmakers and the intelligence community. Ahead of votes in the House and Senate, members of Congress were briefed by intelligence officials on the alleged national security threat posed by the app. The exact nature of those concerns is still unclear, though some members of Congress have asked for details from the briefings to be declassified.

At the same time, some lawmakers have expressed skepticism, saying that the alleged threat posed by TikTok is largely hypothetical. Free speech and digital rights groups also oppose the bill, noting that comprehensive privacy legislation would be a more effective way of protecting Americans’ personal data. TikTok CEO Shou Chew has made a similar argument, telling Congress last year that a forced sale wouldn’t resolve data concerns about the app.

But TikTok’s recent efforts to muster opposition to the bill may have backfired. Lawmakers rebuked the company for sending in-app notifications to users about the bill after the alerts resulted in a flood of calls to Congressional offices. And the app may have drawn even more suspicion when Politico reported last week that Chinese diplomats were lobbying Congressional staffers to oppose the bill. Officials in China have condemned the measure. A Chinese law, passed in 2020, could prevent ByteDance from including TikTok’s recommendation algorithm in a sale of the app.

This article originally appeared on Engadget at https://www.engadget.com/senate-passes-bill-that-could-ban-tiktok-014124533.html?src=rss

Leading artificial intelligence companies including OpenAI, Microsoft, Google, Meta and others have jointly pledged to prevent their AI tools from being used to exploit children and generate child sexual abuse material (CSAM). The initiative was led by child-safety group Thorn and All Tech Is Human, a non-profit focused on responsible tech.

The pledges from AI companies, Thorn said, “set a groundbreaking precedent for the industry and represent a significant leap in efforts to defend children from sexual abuse as a feature with generative AI unfolds.” The goal of the initiative is to prevent the creation of sexually explicit material involving children and take it off social media platforms and search engines. More than 104 million files of suspected child sexual abuse material were reported in the US in 2023 alone, Thorn says. In the absence of collective action, generative AI is poised to make this problem worse and overwhelm law enforcement agencies that are already struggling to identify genuine victims.

On Tuesday, Thorn and All Tech Is Human released a new paper titled “Safety by Design for Generative AI: Preventing Child Sexual Abuse” that outlines strategies and lays out recommendations for companies that build AI tools, search engines, social media platforms, hosting companies and developers to take steps to prevent generative AI from being used to harm children.

One of the recommendations, for instance, asks companies to choose data sets used to train AI models carefully and avoid ones only only containing instances of CSAM but also adult sexual content altogether because of generative AI’s propensity to combine the two concepts. Thorn is also asking social media platforms and search engines to remove links to websites and apps that let people “nudity” images of children, thus creating new AI-generated child sexual abuse material online. A flood of AI-generated CSAM, according to the paper, will make identifying genuine victims of child sexual abuse more difficult by increasing the “haystack problem” — an reference to the amount of content that law enforcement agencies must current sift through.

“This project was intended to make abundantly clear that you don’t need to throw up your hands,” Thorn’s vice president of data science Rebecca Portnoff told the Wall Street Journal. “We want to be able to change the course of this technology to where the existing harms of this technology get cut off at the knees.”

Some companies, Portnoff said, had already agreed to separate images, video and audio that involved children from data sets containing adult content to prevent their models from combining the two. Others also add watermarks to identify AI-generated content, but the method isn’t foolproof — watermarks and metadata can be easily removed.

This article originally appeared on Engadget at https://www.engadget.com/the-worlds-leading-ai-companies-pledge-to-protect-the-safety-of-children-online-213558797.html?src=rss

Tesla has shown off a preview of an upcoming ride-hailing feature in its app ahead of an August robotaxi unveiling. The company released mock-ups of the upcoming feature, which showed the ability to “summon” a ride from the Tesla app.

The company didn’t offer many details about how it would work, but images show Uber-like functionality, with the ability to remotely set the car’s temperature ahead of its arrival. Tesla CEO Elon Musk announced earlier this month that the company would unveil its long-promised robotaxis August 8.

Tesla has been promising a self-driving ride hailing service for years, with Musk promising the company’s robotaxi would start picking up riders back in 2019. For now, it’s unclear exactly when such a service might launch. “We are currently working on ride-hailing functionality that will be available in the future,” the company wrote in its latest earnings report. “We believe the Tesla software experience is best-in-class across all our products, and plan to seamlessly layer ride-hailing into the Tesla App.”

The update was shared in Tesla’s first-quarter earnings report for 2024. The automaker reported a revenue decline of 9 percent from last year, its steepest loss in more than a decade.

During a call with investors, Musk suggested the ride hailing service, which he referred to as "Cyber Cab," would function like “some combination of Airbnb and Uber,” so Tesla owners could choose to make their vehicles part of the service. “There'll be some number of cars that Tesla owns itself and operates in the fleet,” Musk said. “There'll be a bunch of cars where they're owned by the end user, but that end user can add or subtract their car to the fleet whenever they want. And they can decide if they want to only let the car be used by friends and family, or only by five star users, or by anyone.”

This article originally appeared on Engadget at https://www.engadget.com/tesla-previews-ride-hailing-experience-ahead-of-august-robotaxi-unveil-212738199.html?src=rss