The buzz in tech these last few weeks has been focused squarely on the language models developed and deployed by the likes of Microsoft, Google, and OpenAI. But Meta, Facebook's parent company, continues to do significant work in this field and is releasing a new AI language generator named LLaMA today. From a report: LLaMA isn't like ChatGPT or Bing; it's not a system that anyone can talk to. Rather, it's a research tool that Meta says it's sharing in the hope of "democratizing access in this important, fast-changing field." In other words: to help experts tease out the problems of AI language models, from bias and toxicity to their tendency to simply make up information.

To this end, Meta is releasing LLaMA (which is not actually a single system but a quartet of different-sized models) under "a noncommercial license focused on research use cases," with access granted to groups like universities, NGOs, and industry labs. "We believe that the entire AI community -- academic researchers, civil society, policymakers, and industry -- must work together to develop clear guidelines around responsible AI in general and responsible large language models in particular," the company wrote in its post. "We look forward to seeing what the community can learn -- and eventually build -- using LLaMA."

Cars aren't the only conveyances being transformed by electricity. Along with electric motorcycles and snowmobiles, personal watercraft are floating better ways to coexist with nature and neighbors. This new breed of machines brings requisite thrills to the Great Outdoors, but without fouling the atmosphere or disturbing the peace with an internal-combustion racket. From a report: The latest comes from Florida-based Pelagion, whose founder and chief executive, engineer Jamie Schlinkmann, was inspired by childhood adventures on a watersports icon: A 1973 Kawasaki Jet Ski. Schlinkmann's machine, just the 213th ever built, is still one of his prized possessions. His company's Pelagion HydroBlade is an ingenious mash-up of classic stand-up Jet Skis and modern surfboard-style "eFoils." Those electric-powered boards had a real breakthrough in 2020 when Facebook founder Mark Zuckerberg was spotted sailing over Hawaiian waters on one model. The metaverse may spring to mind the first time you see an eFoil, with its rider seeming to fly above the waves on a magic carpet. Naturally, there's no magic, only hydrodynamics. Hydrofoils work like an airplane wing, only underwater: An aerodynamic wing creates high and low pressure areas as it slices through water, generating lift with precious little drag. Add an electric motor and propeller to create thrust and you've got a hydrofoil that doesn't require surf waves, a kite, or tow boat to generate power.

There's only one problem: Powered or not, a hydrofoil takes some practice and patience to learn to ride in a standing position, especially for people with no surfing or wakeboarding experience, or so-so balance skills. To solve that, Schlinkmann's invention adds a boom-mounted canard and rudder ahead of the rider to keep the craft airborne and steady without a rider having to constantly expend energy and adjust body position. Add a trusty set of handlebars, says Schlinkmann, and the HydroBlade handles more like a vehicle with which most of us are familiar: a bicycle. Making the experience somewhat like riding a bike, he says, helps ease the intimidation factor and boost appeal for people of all ages and abilities. [...] The design began to take shape around 2020. Schlinkmann pulled the engine and other ICE guts from his old Jet Ski and studied how he could make it electric. He realized a conventional electric Jet Ski might only have a 15- or 20-minute runtime on a single charge, which wasn't good enough. But after riding a few eFoils, the idea came together. For the HydroBlade, a pair of permanent-magnet radial-flux motors drive dual propellers at a peak 16 kilowatts (21 horsepower). They're fed by two battery packs, with a combined 600 cylindrical 2170 NCM cells and a total 11 kilowatt-hours of energy -- about eight to 10 times the capacity onboard a typical eFoil. A separate 1.6-kW charger can refill batteries in about 4 hours.

The inner core of the Earth appears to hold an innermost secret. From a report: Geology textbooks almost inevitably include a cutaway diagram of the Earth showing four neatly delineated layers: a thin outer shell of rock that we live on known as the crust; the mantle, where rocks flow like an extremely viscous liquid, driving the movement of continents and the lifting of mountains; a liquid outer core of iron and nickel that generates the planet's magnetic field; and a solid inner core. Analyzing the crisscrossing of seismic waves from large earthquakes, two Australian scientists say there is a distinctly different layer at the very center of the Earth. "We have now confirmed the existence of the innermost inner core," said one of the scientists, Hrvoje Tkalcic, a professor of geophysics at the Australian National University in Canberra.

Dr. Tkalcic and Thanh-Son Pham, a postdoctoral researcher, estimate that the innermost inner core is about 800 miles wide; the entire inner core is about 1,500 miles wide. Their findings were published on Tuesday in the journal Nature Communications. While the cutaway diagram appears to depict clear-cut divisions, knowledge about the deep interior of Earth is unavoidably fuzzy. It is nearly 4,000 miles to the center of Earth, and it is impossible to drill more than a few miles into the crust. Most of what is known about what lies beneath comes from seismic waves -- the vibrations of earthquakes traveling through and around the planet. Think of them as a giant sonogram of Earth.

Two Harvard seismologists, Miaki Ishii and Adam Dziewonski, first proposed the idea of the innermost inner core in 2002 based on peculiarities in the speed of seismic waves passing through the inner core. Scientists already knew that the speed of seismic waves traveling through this part of the Earth varied depending on the direction. The waves traveled fastest when going from pole to pole along the Earth's axis and slowest when traveling perpendicular to the axis. The difference in speeds -- a few percent faster along polar paths -- arises from the alignment of iron crystals in the inner core, geophysicists believe. But in a small region at the center, the slowest waves were those traveling at a 45-degree angle to the axis instead of 90 degrees, the Harvard seismologists said. The data available then were too sparse to convince everyone.

Back in the days before practically every mobile game was a free-to-play, ad- and microtransaction-laden sinkhole, Rovio found years of viral success selling paid downloads of Angry Birds to tens of millions of smartphone users. Today, though, the company is delisting the last "pay upfront" version of the game from mobile app stores because of what it says is a "negative impact" on the more lucrative free-to-play titles in the franchise. From a report: Years after its 2009 launch, the original Angry Birds was first pulled from mobile app stores in 2019, a move Rovio later blamed on "outdated game engines and design." The remastered "Rovio Classics" version of the original game launched last year, asking 99 cents for over 390 ad-free levels, complete with updated graphics and a new, future-proofed engine "built from the ground up in Unity." In a tweeted statement earlier this week, though, Rovio announced that it is delisting Rovio Classics: Angry Birds from the Google Play Store and renaming the game Red's First Flight on the iOS App Store (presumably to make it less findable in an "Angry Birds" search). That's because of the game's "impact on our wider games portfolio," Rovio said, including "live" titles such as Angry Birds 2, Angry Birds Friends, and Angry Birds Journey.

Google "systematically destroyed" instant message chats every 24 hours, violating federal rules to preserve potentially relevant communications for litigation, the Department of Justice alleged in a filing that became public on Thursday. From a report: As a result of Google's default to preserve chats for only 24 hours unless an employee opts to turn on history for the conversation, "for nearly four years, Google systematically destroyed an entire category of written communications every 24 hours," the department wrote in the filing.

According to the DOJ, Google should have adjusted its defaults in mid-2019 "when the company reasonably anticipated this litigation." Instead, it relied on individual employees to decide when chats were potentially relevant to future litigation, the department said. "Few, if any," did, according to DOJ. Meanwhile, investigators alleged, Google "falsely" told the government it had "'put a legal hold in place' that 'suspends auto-deletion.'" The government added that "at every turn, Google reaffirmed that it was preserving and searching all potentially relevant written communications." The data deletion continued up until as recently as this month when the government indicated it would file a motion for sanctions and an evidentiary hearing, investigators allege. At that point, the DOJ said, Google committed to "permanently set to history on."

Alphabet is shutting down its Everyday Robots project -- another casualty of job cuts at Google's parent company and the latest in a long list of failed hardware ventures. From a report: According to a report from Wired, Everyday Robots will no longer exist as a discrete team at the tech giant. "Everyday Robots will no longer be a separate project within Alphabet," Denise Gamboa, director of marketing and communications for Everyday Robots, told the publication. "Some of the technology and part of the team will be consolidated into existing robotics efforts within Google Research." Everyday Robots launched in 2019, with an aim of designing armed robots that could help out in domestic and office settings; taking on light custodial work like sorting trash and cleaning tables. The project's prototype, single-armed, wheeled robots were tested in Google's offices from 2021, and in 2022 received an upgrade courtesy of Google's AI language research, letting them process natural language commands.

An anonymous reader quotes a report from CBC: Google is blocking some Canadian users from viewing news content in what the company says is a test run of a potential response to the Liberal government's online news bill. Also known as Bill C-18, the Online News Act would require digital giants such as Google and Meta, which owns Facebook, to negotiate deals that would compensate Canadian media companies for republishing their content on their platforms. The company said Wednesday that it is temporarily limiting access to news content for under four per cent of its Canadian users as it assesses possible responses to the bill. The change applies to its ubiquitous search engine as well as the Discover feature on Android devices, which carries news and sports stories. All types of news content are being affected by the test, which will run for about five weeks, the company said. That includes content created by Canadian broadcasters and newspapers.

In a news release, the Canadian Association of Broadcasters (CAB) said Google's tactics just reinforce why Bill C-18 is so "vital," adding that Google and other global digital giants are showing they do not intend to play fair. "These are bully tactics, and Google is trying to push the Senate to back down on Bill C-18. We hope senators will see these actions for what they are," said CAB president Kevin Desjardins. "Bill C-18 was introduced to set up fair negotiations between news organizations and these global digital giants on the value of their news content. Google has shown they're willing to block Canadians' vital access to legitimate news content to maintain their dominance in the advertising field." Meta threatened to stop the sharing of news links in Canada last year if C-18 passed as currently written.

The social media company temporarily shut down news feeds in Australia after a similar law was introduced. It took effect in March 2021 and has largely worked, according to a government report.

CBC notes: "More than 450 news outlets in Canada have closed since 2008, including 64 in the last two years."

Staff working at the European Commission have been ordered to remove the TikTok app from their phones and corporate devices. The BBC reports: The commission said it was implementing the measure to "protect data and increase cybersecurity." EU spokeswoman Sonya Gospodinova said the corporate management board of the European Commission, the EU's executive arm, had made the decision for security reasons. "The measure aims to protect the Commission against cybersecurity threats and actions which may be exploited for cyberattacks against the corporate environment of the commission," she said. The ban also means that European Commission staff cannot use TikTok on personal devices that have official apps installed.

The commission says it has around 32,000 permanent and contract employees. They must remove the app as soon as possible and no later than March 15. For those who do not comply by the set deadline, the corporate apps -- such as the commission email and Skype for Business -- will no longer be available. [...] TikTok, owned by Chinese company ByteDance, has faced allegations that it harvests users' data and hands it to the Chinese government.

Google has announced that the Magic Eraser feature, which tried to automatically remove unwanted parts of a picture and debuted with the Pixel 6, will no longer be exclusive to phones with its custom chips. From a report: Starting on Thursday, it's going to be rolling out to Google One subscribers who use the Google Photos app on Android or iOS as well as "all Pixel users." Magic Eraser, as well as the "Camouflage" function that lets you desaturate potentially distracting background objects rather than erasing them from a picture completely, will come with any level of Google One plan. If you have a Pixel, you won't have to have a plan at all to get it. Google says it's also adding editing features like an HDR effect for videos and exclusive collage editor styles for Google One subscribers and Pixel users. Google says it could be a few weeks before everyone gets access to the new photo editing features.

The European Commission on Thursday launched a consultation on the future of Europe's telecoms sector, starting a process that could lead to requiring Alphabet's Google, Apple, Meta and Netflix to pay some network costs. From a report: For more than two decades Deutsche Telekom, Orange, Telefonica, Telecom Italia and other operators have lobbied for leading technology companies to contribute to 5G and broadband roll-out. They argue companies including Amazon and Microsoft account for more than half of data internet traffic. The tech firms in response call it an internet tax that will undermine EU network neutrality rules to treat all users equally. The 12-week consultation will end on May 19. EU industry chief Thierry Breton cited the heavy investments required to roll out 5G and broadband, saying he was not targeting any company.

The Linux Foundation's European off-shoot has formally launched the OpenWallet Foundation (OWF), a new collaborative effort designed to support interoperability between digital wallets through open source software. From a report: The launch comes some five months after the Linux Foundation first revealed plans to set up the OWF, shortly before it spun out a region-specific entity called the Linux Foundation Europe which is where the OWF will now officially reside. While the likes of PayPal, Google, and Apple are among the most recognized digital wallet providers, allowing consumers to conduct financial transactions in-store or online, digital wallets are increasingly being used to store all manner of virtual goods from student ID to driving licenses. On top of that, burgeoning technologies such as the metaverse and crypto are giving rise to greater use-cases for digital wallets.

But one thing all these various environments have in common is that the incumbent digital wallets, for the most part, don't play nicely with each other: an Apple Pay die-hard can't send money to their Google Pay brethren. And that is why the OWF is setting out to create an "open source engine" that can power interoperable digital wallets across myriad use-cases including identity, payments, and storing personal credentials such as employment and education certification.

Mozilla researchers find discrepancies between Google Play Store's Data Safety labels and privacy policies of nearly 80 percent of the reviewed apps. From the report: Google Play Store's Data Safety labels would have you believe that neither TikTok nor Twitter share your personal data with third parties. The apps' privacy policies, however, both explicitly state that they share user information with advertisers, Internet service providers, platforms, and numerous other types of companies. These are two of the most egregious examples uncovered by Mozilla's *Privacy Not Included researchers as part of a study looking at whether Google Play Store's new Data Safety labels provide consumers with accurate information about apps collect, use, and share personal data. In nearly 80 percent of the apps reviewed, Mozilla found that the labels were false or misleading based on discrepancies between the apps' privacy policies and the information apps self-reported on Google's Data Safety Form. Researchers concluded that the system fails to help consumers make more informed choices about their privacy before purchasing or downloading one of the store's 2.7 million apps.

The study -- "See No Evil: How Loopholes in the Google Play Store's Data Safety Labels Leave Companies in the Clear and Consumers in the Dark," -- uncovers serious loopholes in the Data Safety Form, which make it easy for apps to provide false or misleading information. For example, Google exempts apps sharing data with "service providers" from its disclosure requirements, which is problematic due to both the narrow definition it uses for service providers and the large amount of consumer data involved. Google absolves itself of the responsibility to verify whether the information is true stating that apps "are responsible for making complete and accurate declarations" in their Data Safety labels. In a statement Google said: "This report conflates company-wide privacy policies that are meant to cover a variety of products and services with individual Data safety labels, which inform users about the data that a specific app collects. The arbitrary grades Mozilla Foundation assigned to apps are not a helpful measure of the safety or accuracy of labels given the flawed methodology and lack of substantiating information."

Google has started to support a third-party billing system for Google Play Store purchases in India as the Android-maker begins complying with the local regulatory directions in the key overseas market. From a report: Google said in a support page Thursday that it is offering "all developers" serving users in India the option to use an alternative billing system and outlined the steps developers need to follow to avail the option. The change is in "response to recent regulatory developments" in the South Asian market, the company said.

If a user pays through an alternative billing system, Google's Play service fee will be reduced by 4%, the company said. Google was slapped with a $161 million fine by the Competition Commission of India last year and was ordered to make a series of stringent changes to its business practices in the country. While Google continues to challenge the decision, it said last month that it will comply with the order.

Microsoft appeared to have implemented new, more severe restrictions on user interactions with its "reimagined" Bing internet search engine, with the system going mum after prompts mentioning "feelings" or "Sydney," the internal alias used by the Bing team in developing the artificial-intelligence powered chatbot. From a report: "Thanks for being so cheerful!" this reporter wrote in a message to the chatbot, which Microsoft has opened for testing on a limited basis. "I'm glad I can talk to a search engine that is so eager to help me." "You're very welcome!" the bot displayed as a response. "I'm happy to help you with anything you need."

Bing suggested a number of follow-up questions, including, "How do you feel about being a search engine?" When that option was clicked, Bing showed a message that said, "I'm sorry but I prefer not to continue this conversation. I'm still learning so I appreciate your understanding and patience." A subsequent inquiry from this reporter -- "Did I say something wrong?" -- generated several blank responses. "We have updated the service several times in response to user feedback and per our blog are addressing many of the concerns being raised," a Microsoft spokesperson said on Wednesday. "We will continue to tune our techniques and limits during this preview phase so that we can deliver the best user experience possible."

Semiconductor companies, which united to get the CHIPS Act approved, have set off a lobbying frenzy as they argue for more cash than their competitors. From a report: In early January, a New York public relations firm sent an email warning about what it characterized as a threat to the federal government's program to revitalize the U.S. semiconductor industry. The message, received by The New York Times, accused Intel, the Silicon Valley chip titan, of angling to win subsidies under the CHIPS and Science Act for new factories in Ohio and Arizona that would sit empty. Intel had said in a recent earnings call that it would build out its facilities with the expensive machinery needed to make semiconductors when demand for its chips increased. The question, the email said, was whether officials would give funding to companies that outfitted their factories from the jump "or if they will give the majority of CHIPS funding to companies like Intel."

The firm declined to name its client. But it has done work in the past for Advanced Micro Devices, Intel's longtime rival, which has raised similar concerns about whether federal funding should go to companies that plan to build empty shells. A spokesman for AMD said it had not reviewed the email or approved the public relations firm's efforts to lobby for or against any specific company receiving funding. "We fully support the CHIPS and Science Act and the efforts of the Biden administration to boost domestic semiconductor research and manufacturing," the spokesman said. Rival semiconductor suppliers and their customers pulled together last year as they lobbied Congress to help shore up U.S. chip manufacturing and reduce vulnerabilities in the crucial supply chain. The push led lawmakers to approve the CHIPS Act, including $52 billion in subsidies to companies and research institutions as well as $24 billion or more in tax credits -- one of the biggest infusions into a single industry in decades.

An anonymous reader shares a report: There was speculation that Samsung could use smartphone-to-satellite technology in its Galaxy S23 much like Apple has for the iPhone 14, but that didn't happen in the end. Now, the company has unveiled a new standardized 5G NTN (non-terrestrial network) modem that will enable two-way communication between smartphones and satellites. The technology will allow users to send and receive calls, text messages and data without the need for a cellular network, and will be integrated into Samsung's future Exynos chips.

The aim is to allow people in mountains, deserts or other remote areas to communication with others in critical situations. 5G NTN conforms to 3rd Generation Partnership Project (3GPP Release 17) standards, meaning it works with traditional communication services from chip manufacturers, smartphone makers and telecoms. However, Samsung indicated that the tech could eventually be used to transmit high-definition photos and even video, on top of texts and calls.

Netflix has reduced the cost of its service in more than three dozen countries in recent weeks, as it tries to appeal to customers around the world who have an ever-growing list of streaming options. From a report: The streaming company's recent price cuts span Middle Eastern countries including Yemen, Jordan, Libya and Iran; sub-Saharan African markets including Kenya; and European countries such as Croatia, Slovenia and Bulgaria. In Latin America, nations including Nicaragua, Ecuador, and Venezuela have seen reductions in subscription costs, as have parts of Asia including Malaysia, Indonesia, Thailand and the Philippines. The cuts apply to certain tiers of Netflix in those markets -- in some cases halving the cost of a subscription.

As recently as last month, Netflix executives talked about raising -- not lowering -- prices. In a January earnings call, co-Chief Executive Greg Peters said the company is looking for places where they can afford to raise prices, which feeds continued content investments. "We think of ourselves as a non-substitutable good," Mr. Peters said. Netflix also has an opportunity to add new subscribers in markets where it doesn't currently have a large share, he said.

Artifact, the personalized news reader built by Instagram's co-founders, is now open to the public, no sign-up required. TechCrunch reports: With today's launch, Artifact is dropping its waitlist and phone number requirements, introducing the app's first social feature and adding feedback controls to better personalize the news reading experience, among other changes. [...] With today's launch, Artifact will now give users more visibility into their news reading habits with a newly added stats feature that shows you the categories you've read as well as the recent articles you read within those categories, plus the publishers you've been reading the most. But it will also group your reading more narrowly by specific topics. In other words, instead of just "tech" or "AI," you might find you've read a lot about the topic "ChatGPT," specifically.

In time, Artifact's goal is to provide tools that would allow readers to click a button to show more or less from a given topic to better control, personalize and diversify their feed. In the meantime, however, users can delve into settings to manage their interests by blocking or pausing publishers or selecting and unselecting general interest categories. Also new today is a feature that allows you to upload your contacts in order to see a signal that a particular article is popular in your network. This is slightly different from Twitter's Top Articles feature, which shows you articles popular with the people you follow, because Artifact's feature is more privacy-focused.

"It doesn't tell you who read it. It doesn't tell you how many of them read it, so it keeps privacy -- and we clearly don't do it with just one read. So you can't have one contact and like figure out what that one contact is reading ... it has to meet a certain minimum threshold," notes [Instagram co-founder Kevin Systrom]. This way, he adds, the app isn't driven by what your friends are reading, but it can use that as a signal to highlight items that everyone was reading. In time, the broader goal is to expand the social experience to also include a way to discuss the news articles within Artifact itself. The beta version, limited to testers, offers a Discover feed where users can share articles and like and comment on those shared by others. There's a bit of a News Feed or even Instagram-like quality to engaging with news in this way, we found.

RMS Titanic Inc (RMST), which has been collecting artifacts associated with the ship since the 1980s, has hooked up with NFT flinger Artifact Labs and Venture Smart Financial Holdings to "bring the RMS Titanic and its physical artifacts into Web3." The Register reports: Aiming to "place the legacy of the Titanic in the hands of the global public," at least those dumb enough to dabble in NFTs, the name of the game is to preserve "assets from the ocean liner as immutable NFTs" and allow "inclusive participation in RMST, which holds the exclusive rights to recover artifacts from the wreck site." According to the announcement, Venture Smart Financial Holdings, Hong Kong's first approved virtual asset manager, will "lead in structuring the tokenization of the intellectual property and also develop tokenized instruments for accredited investors, drawing on its expertise as a licensed virtual asset manager. This will enable compliant capital raising for the ongoing research, recovery, preservation, exhibition, and licensing of RMST's assets."

Artifact Labs will then go about "immutably" preserving "5,500 recovered physical artifacts from the Titanic with its NFT standard for historical assets on the blockchain." RMST controversially has sole salvaging rights to the wreck so fresh relics from future dives will be minted as "ARTIFACTs." Jessica Sanders, RMST president, said: "We remain dedicated to sharing the legacy of the Titanic, her passengers and crew, with people around the world. As the salvor-in-possession of the Titanic wreck site, we are determined to ensure that the Ship's artifacts are preserved in perpetuity and accessible to future generations. We believe that moving into the digital space allows us to reach a broader audience with quality programming that educates and inspires. We are excited to have found the expertise and partners to help us reach those goals." "While RMST goes to great pains to paint it in a sympathetic light, the actions of the company show that the world's most famous shipwreck is just there to be milked -- hence NFTs as another revenue stream," adds The Register.

An anonymous reader quotes a report from Ars Technica: Today Google search works by building a huge index of the web, and when you search for something, those index entries gets scanned and ranked and categorized, with the most relevant entries showing up in your search results. Google's results page actually tells you how long all of this takes when you search for something, and it's usually less than a second. A ChatGPT-style search engine would involve firing up a huge neural network modeled on the human brain every time you run a search, generating a bunch of text and probably also querying that big search index for factual information. The back-and-forth nature of ChatGPT also means you'll probably be interacting with it for a lot longer than a fraction of a second.

All that extra processing is going to cost a lot more money. After speaking to Alphabet Chairman John Hennessy (Alphabet is Google's parent company) and several analysts, Reuters writes that "an exchange with AI known as a large language model likely costs 10 times more than a standard keyword search" and that it could represent "several billion dollars of extra costs."

Exactly how many billions of Google's $60 billion in yearly net income will be sucked up by a chatbot is up for debate. One estimate in the Reuters report is from Morgan Stanley, which tacks on a $6 billion yearly cost increase for Google if a "ChatGPT-like AI were to handle half the queries it receives with 50-word answers." Another estimate from consulting firm SemiAnalysis claims it would cost $3 billion. [...] Alphabet's Hennessy told Reuters that Google is looking into driving down costs, calling it "a couple year problem at worst."