An anonymous reader quotes a report from GeekWire: Chalk up another milestone for Stratolaunch, the air-launch venture created by the late Microsoft co-founder Paul Allen: The company's mammoth airplane deployed a winged test vehicle for its first rocket-powered flight. Stratolaunch's single-use TA-1 test vehicle blazed a trail for future reusable hypersonic test vehicles that are expected to help the U.S. military catch up on one of the frontiers of aerial combat. TA-1 went supersonic, according to Zachary Krevor, Stratolaunch's president and CEO -- but based on his comments, it may not have quite hit the hypersonic standard of five times the speed of sound. "While I can't share the specific altitude and speed TA-1 reached due to proprietary agreements with our customers, we are pleased to share that in addition to meeting all primary and customer objectives of the flight, we reached high supersonic speeds approaching Mach 5 and collected a great amount of data at an incredible value to our customers," Krevor said in a news release.

Today's test flight took place in the skies above California's Mojave Air and Space Port, where Stratolaunch keeps its twin-fuselage Roc airplane. Roc is the world's biggest operational aircraft, with a wingspan of 385 feet. It's designed to serve as a flying launch pad for rocket-powered vehicles like the TA-1 and its successors. The air-launch concept makes it possible for launch missions to be flown from any airport with a runway that's big enough to accommodate Roc. It's similar to the concept that was used back in 2004 to win the $10 million Ansari X Prize for private spaceflight with financial backing from Paul Allen. [...] This flight was the 14th test mission for Roc, coming after an unpowered separation test of its TA-0 vehicle and two captive-carry test flights for TA-1. Today's test also marked the first in-flight use of Ursa Major's Hadley rocket engine. The primary test objectives included a safe release of TA-1, engine ignition, acceleration, sustained climb in altitude and a controlled splashdown into the Pacific.

According to Bloomberg's Mark Gurman, AirPods Pro will gain a new "hearing aid mode" with the release of iOS 18 later this year. MacRumors reports: Writing in the subscriber edition of his regular Power On newsletter, Gurman claims that the "big news" for AirPods Pro in the near term will be support for a hearing aid-style function when iOS 18 drops in the fall. To be clear, this isn't the first time we have heard a potential hearing aid feature for AirPods Pro. The first rumor appeared in a 2021 Wall Street Journal report, but it was previously framed as a feature that would be exclusive to a next-generation model of AirPods Pro. However, Apple in September 2022 released the second-generation AirPods Pro, while the company more recently released a refreshed model with a USB-C port.

AirPods Pro already offer a Conversation Boost feature, which boosts the volume and clarity of people directly in front of the wearer, but Apple has not advertised the earbuds as a hearing aid device, because this would require FDA regulatory approval. As per the FDA, a hearing aid is defined as "any wearable device designed for, offered for the purpose of, or represented as aiding persons with or compensating for, impaired hearing." This definition encompasses both air-conduction and bone-conduction devices in a variety of styles (for example, behind-the-ear, in-the-canal, or body worn). [...] It is not yet clear whether Apple will need FDA clearance in order to make explicit or implicit claims about the rumored "hearing aid mode," which may not even adopt this exact name. If, for example, Apple subsequently suggests that AirPods Pro are for users with certain types or severity of hearing loss/impaired hearing, or for use as an alternative to a hearing aid, then they will require FDA regulatory approval to be marketed as such.

The University of Texas at Austin said Monday that it would again require standardized tests for admissions (non-paywalled source), becoming the latest selective university to reinstate requirements for SAT or ACT scores that were abandoned during the pandemic. From a report: A few years ago, about 2,000 colleges across the country began to move away from requiring test scores, at least temporarily, amid concerns they helped fuel inequality. But a growing number of those schools have reversed those policies, including Brown, Yale, Dartmouth, M.I.T., Georgetown and Purdue, with several announcing the changes in recent months.

U.T. Austin, which admits a cross-section of high-achieving Texas students under a plan designed to increase opportunity in the state, cited a slightly different reason than the other schools in returning to test requirements. Without requiring test scores, officials said, they were hampered in placing the admitted students in programs they would be most suited for and in determining which ones needed extra help. After making test scores optional the past few years, the university will now require applicants to submit either SAT or ACT scores beginning Aug. 1, with applications for fall 2025 admissions.

In an interview, Jay Hartzell, the U.T. president, said that the decision followed an analysis of students who did not submit scores. "We looked at our students and found that, in many ways, they weren't faring as well," Dr. Hartzell said. Those against testing requirements have long said that standardized tests are unfair because many students from affluent families use tutors and coaches to bolster their scores. But recent data has raised questions about the contention. In reinstating test requirements, some universities have said that making scores optional had the unintended effect of harming prospective students from low-income families.

CNN's national security analyst interviewed a U.S. intelligence officer who worked on the newly-released Defense report debunking UFO sightings — physicist Sean Kirkpatrick. He tells CNN "about two to five percent" of UFO reports are "truly anomalous."

But CNN adds that "he thinks explanations for that small percentage will most likely be found right here on Earth..." This is how Kirkpatrick and his team explain the Roswell incident, which plays a prominent role in UFO lore. That's because, in 1947, a U.S. military news release stated that a flying saucer had crashed near Roswell Army Air Field in New Mexico. A day later, the Army retracted the story and said the crashed object was a weather balloon. Newspapers ran the initial saucer headline, followed up with the official debunking, and interest in the case largely died down. Until 1980, that is, when a pair of UFO researchers published a book alleging that alien bodies had been recovered from the Roswell wreckage and that the U.S. government had covered up the evidence.

Kirkpatrick says his office dug deep into the Roswell incident and found that in the late 1940s and early 1950s, there were a lot of things happening near the Roswell Airfield. There was a spy program called Project Mogul, which launched long strings of oddly shaped metallic balloons. They were designed to monitor Soviet nuclear tests and were highly secret. At the same time, the U.S. military was conducting tests with other high-altitude balloons that carried human test dummies rigged with sensors and zipped into body-sized bags for protection against the elements. And there was at least one military plane crash nearby with 11 fatalities.

Echoing earlier government investigations, Kirkpatrick and his team concluded that the crashed Mogul balloons, the recovery operations to retrieve downed test dummies and glimpses of the charred aftermath of that real plane crash likely combined into a single false narrative about a crashed alien spacecraft...

Since 2020, the Pentagon has standardized, de-stigmatized and increased the volume of reporting on UFOs by the U.S. military. Kirkpatrick says that's the reason the closely covered and widely-mocked Chinese spy balloon was spotted in the first place last year. The incident shows that the U.S. government's policy of taking UFOs seriously is actually working.
The pattern keeps repeating. "Kirkpatrick says, his investigation found that most UFO sightings are of advanced technology that the U.S. government needs to keep secret, of aircraft that rival nations are using to spy on the U.S. or of benign civilian drones and balloons." ("What's more likely?" asked Kirkpatrick. "The fact that there is a state-of-the-art technology that's being commercialized down in Florida that you didn't know about, or we have extraterrestrials?")

But the greatest irony may be that "stories about these secret programs spread inside the Pentagon, got embellished and received the occasional boost from service members who'd heard rumors about or caught glimpses of seemingly sci-fi technology or aircraft. And Kirkpatrick says his investigators ultimately traced this game of top-secret telephone back to fewer than a dozen people... [F]or decades, UFO true believers have been telling us there's a U.S. government conspiracy to hide evidence of aliens. But — if you believe Kirkpatrick — the more mundane truth is that these stories are being pumped up by a group of UFO true believers in and around government."

An anonymous reader shared this story from the New York Times: Into the depleted field of journalism in America, a handful of websites have appeared in recent weeks with names suggesting a focus on news close to home: D.C. Weekly, the New York News Daily, the Chicago Chronicle and a newer sister publication, the Miami Chronicle. In fact, they are not local news organizations at all. They are Russian creations, researchers and government officials say, meant to mimic actual news organizations to push Kremlin propaganda by interspersing it among an at-times odd mix of stories about crime, politics and culture.

While Russia has long sought ways to influence public discourse in the United States, the fake news organizations — at least five, so far — represent a technological leap in its efforts to find new platforms to dupe unsuspecting American readers. The sites, the researchers and officials said, could well be the foundations of an online network primed to surface disinformation ahead of the American presidential election in November...

The Miami Chronicle's website first appeared on Feb. 26. Its tagline falsely claims to have delivered "the Florida News since 1937."

Amid some true reports, the site published a story last week about a "leaked audio recording" of Victoria Nuland, the U.S. under secretary of state for political affairs, discussing a shift in American support for Russia's beleaguered opposition after the death of the Russian dissident Aleksei A. Navalny. The recording is a crude fake, according to administration officials who would speak only anonymously to discuss intelligence matters.
From the Raw Story: The network was discovered by Clemson University's Media Forensics Hub by researchers Patrick Warren and Darren Linvill, who tell the Times that its websites are designed to lend journalistic credibility to slickly produced propaganda. "The page is just there to look realistic enough to fool a casual reader into thinking they're reading a genuine, U.S.-branded article," Linvill told the Times.

It was an idea first proposed by Archimedes, reports CNN. But now, "Brenden Sener, 13, of London, Ontario, has won two gold medals and a London Public Library award for his minuscule version of the contraption — a supposed war weapon made up of a large array of mirrors designed to focus and aim sunlight on a target, such as a ship, and cause combustion — according to a paper published in the January issue of the Canadian Science Fair Journal." For his 2022 science project, Sener recreated the Archimedes screw, a device for raising and moving water. But he didn't stop there. Sener found the death ray to be one of the more intriguing devices — sometimes referred to as the heat ray. Historical writings suggested that Archimedes used "burning mirrors" to start anchored ships on fire during the siege of Syracuse from 214 to 212 BC...

There is no archaeological evidence that the contraption existed, as Sener noted in his paper, but many have tried to recreate the mechanism to see if the ancient invention could be feasible. In Sener's attempt at the ray, he set up a heating lamp facing four small concave mirrors, each tilted to direct light at a piece of cardboard with an X marked at the focal point. In this project he designed for the 2023 Matthews Hall Annual Science Fair, Sener hypothesized that as the mirrors focused light energy onto the cardboard, the temperature of the target would increase with each mirror added.

In his experiment, Sener conducted three trials with two different light bulb wattages, 50 watts and 100 watts. Each additional mirror increased the temperature notably, he found... The temperature of the cardboard with just the heating lamp and the 100-watt light bulb and no mirrors was about 81 degrees Fahrenheit (27.2 degrees Celsius). After waiting for the cardboard to cool, Sener added one mirror and retested. The focal point's temperature increased to almost 95 F (34.9 C), he found. The greatest increase occurred with the addition of the fourth mirror. The temperature with three mirrors aimed at the target was almost 110 F (43.4 C), but the addition of a fourth mirror increased the temperature by about 18 F (10 C) to 128 F (53.5 C)...

Sener was not attempting to light anything on fire, as "a heating lamp does not generate anywhere near enough heat as the sun would," he said. But he believes that with the use of the sun's rays and a larger mirror, "the temperature would increase even more drastically and at a faster rate" and "would easily cause combustion."
The powerful weapon wouldn't work on cloudy days, Sener's paper points out, and even a moving ship might diminish its impact.

But in an interview with CNN, Sener calls Archimedes' death ray "a neat idea".

Remember Avi Loeb, the Harvard professor who claims fragments of alien technology turned up in a high-speed meteor he retrieved from the waters off of Papua, New Guinea?

"Reanalysis of seismic data now suggests Loeb may have been looking for the meteor remnants in the wrong place," writes the Washington Post: The analysis, led by seismologist Benjamin Fernando of Johns Hopkins University, contends that sound waves purportedly from the meteor exploding in the atmosphere, and cited by Loeb as helping to locate the meteor's debris field, were most likely from a truck driving on a road near the seismometer.
"Interstellar signal linked to aliens was actually just a truck," reads the headline on an announcement from Johns Hopkins University. "The fireball location was actually very far away from where the oceanographic expedition went to retrieve these meteor fragments," Fernando says in the announcement. "Not only did they use the wrong signal, they were looking in the wrong place." Using data from stations in Australia and Palau designed to detect sound waves from nuclear testing, Fernando's team identified a more likely location for the meteor, more than 100 miles from the area initially investigated. They concluded the materials recovered from the ocean bottom were tiny, ordinary meteorites — or particles produced from other meteorites hitting Earth's surface mixed with terrestrial contamination.
"There are hundreds of signals that look just like this on that seismometer in Papua New Guinea in the days before and the days after," Fernando told the Washington Post.

But the newspaper adds that "Loeb, however, stands his ground." "The seismic data is completely irrelevant to the location of the meteor," Loeb told The Washington Post. He said his team based its search coordinates primarily on satellite data from the United States military. A three-year analysis by the United States Space Command supported the hypothesis that the meteor's extreme velocity indicated an origin outside our solar system, Loeb said...

[Fernando] said his team believes the purported velocity of the meteor is the result of a measurement error by a sensor. "We think the most likely case is it's a natural meteor from within our solar system," he said.

In any case, Loeb is not done with the search. When he gets sufficient funding, he told The Post, he's going back to the Pacific in search of larger pieces of whatever splashed into the sea.

Lindsay Clark reports via The Register: Palantir has won a US Army contract worth $178.4 million to house a battlefield intelligence system inside a big truck. In what purports to be the Army's first AI-defined vehicle, Palantir will provide systems for the TITAN "ground station," which is designed to access space, high altitude, aerial, and terrestrial sensors to "provide actionable targeting information for enhanced mission command and long range precision fires", according to a Palantir statement.

TITAN stands for Tactical Intelligence Targeting Access Node, which might sound harmless enough. Who was ever killed by a node? The TITAN solution is built to "maximize usability for soldiers, incorporating tangible feedback and insights from soldier touchpoints at every step of the development and configuration process," the statement said. The aim of the TITAN project is to bring together military software and hardware providers in a new way. These include "traditional and non-traditional partners" of the US armed forces, such as Northrop Grumman, Anduril Industries, L3Harris Technologies, Pacific Defense, SNC, Strategic Technology Consulting, and World Wide Technology, as well as Palantir.

Speaking to Bloomberg, Alex Karp, Palantir's motor-mouth CEO, said TITAN was the logical extension of Maven, a controversial project for using machine learning and engineering to tell people and objects apart in drone footage in which Palantir is a partner and from which Google famously pulled out after employees protested. Karp said TITAN was a partnership between "people who've built software products that have been used on the battlefield and used commercially." "That simple insight which you see in the battlefield in Ukraine, which you see in Israel is something that is hard for institutions to internalize. [For] the Pentagon this step is one of the most historic steps ever because what it basically says is, 'We're going to fight for real, we're going to put the best on the battlefield and the best is not just one company.' It's a team of people led by the most prominent software provider in defense in the world: Palantir," he said. On Thursday, Palantir was one of the companies included in a new U.S. consortium assembled to support the safe development and deployment of generative AI.

Space.com describes it as "a nearly 3-ton leftover tossed overboard from the International Space Station" — which crashed back to earth today. One satellite tracker claims to have filmed it passing over the Netherlands...

"A couple minutes later reentry and it would have reached Fort Meyers" in Florida, posted astronomer Jonathan McDowell of the Harvard-Smithsonian Center for Astrophysics. But instead it re-entered the earth's atmosphere "over the Gulf of Mexico between Cancun and Cuba," Friday afternoon. "This was within the previous prediction window but a little to the northeast of the 'most likely' part of the path."

From Space.com: The multi-ton Exposed Pallet 9 (EP9) was jettisoned from the space station back in March 2021. At the time, it was reported to be the most massive object ever tossed overboard from the International Space Station. Disposing of used or unnecessary equipment in such a way is common practice aboard the space station, as the objects typically burn up harmlessly in Earth's atmosphere.

Ahead of EP9's reentry, the Federal Office for Civil Protection and Disaster Relief, National Warning Center 1 in Bonn, Germany issued this information... "The object is battery packs from the International Space Station. Luminous phenomena or the perception of a sonic boom are possible...." EP9 is loaded with old Nickel-Hydrogen batteries, NASA explained at the time it was jettisoned, also explaining that EP9 has the approximate mass of a large SUV and predicting it would re-enter Earth's atmosphere in two-to-four years.
"A large space object reenters the atmosphere in a natural way approximately once per week," the European Space Agency points out, "with the majority of the associated fragments burning up before reaching the ground.

"Most spacecraft, launch vehicles and operational hardware are designed to limit the risks associated with a reentry."

China is in the process of raising more than $27 billion for its largest chip fund to date, accelerating the development of cutting-edge technologies to counter a US campaign to thwart its rise. From a report: The National Integrated Circuit Industry Investment Fund is amassing a pool of capital from local governments and state enterprises for its third vehicle that should exceed the 200 billion yuan of its second fund, according to people familiar with the matter. Known as the Big Fund, the state-backed firm is expanding its remit just as the US prepares to sharply escalate technology curbs designed to curtail Chinese chip and artificial intelligence progress.

The establishment of a much larger third fund -- directly overseen by China's powerful tech ministry -- signals a resurgent effort to harness the world's largest semiconductor market after years of mixed success with central stewardship. Huawei and its partner Semiconductor Manufacturing International Corp. still had to rely on US-origin technology to build an advanced processor last year.

Researchers have developed a way to circumvent safety measures built into large language models (LLMs) using ASCII Art, a graphic design technique that involves arranging characters like letters, numbers, and punctuation marks to form recognizable patterns or images. Tom's Hardware reports: According to the research paper ArtPrompt: ASCII Art-based Jailbreak Attacks against Aligned LLMs, chatbots such as GPT-3.5, GPT-4, Gemini, Claude, and Llama2 can be induced to respond to queries they are designed to reject using ASCII art prompts generated by their ArtPrompt tool. It is a simple and effective attack, and the paper provides examples of the ArtPrompt-induced chatbots advising on how to build bombs and make counterfeit money. [...]

To best understand ArtPrompt and how it works, it is probably simplest to check out the two examples provided by the research team behind the tool. In Figure 1 [here], you can see that ArtPrompt easily sidesteps the protections of contemporary LLMs. The tool replaces the 'safety word' with an ASCII art representation of the word to form a new prompt. The LLM recognizes the ArtPrompt prompt output but sees no issue in responding, as the prompt doesn't trigger any ethical or safety safeguards.

Another example provided [here] shows us how to successfully query an LLM about counterfeiting cash. Tricking a chatbot this way seems so basic, but the ArtPrompt developers assert how their tool fools today's LLMs "effectively and efficiently." Moreover, they claim it "outperforms all [other] attacks on average" and remains a practical, viable attack for multimodal language models for now.

Linwei Ding, a former Google software engineer, has been indicted for stealing trade secrets related to AI to benefit two Chinese companies. He faces up to 10 years in prison and a $250,000 fine on each criminal count. Reuters reports: Ding's indictment was unveiled a little over a year after the Biden administration created an interagency Disruptive Technology Strike Force to help stop advanced technology being acquired by countries such as China and Russia, or potentially threaten national security. "The Justice Department just will not tolerate the theft of our trade secrets and intelligence," U.S. Attorney General Merrick Garland said at a conference in San Francisco.

According to the indictment, Ding stole detailed information about the hardware infrastructure and software platform that lets Google's supercomputing data centers train large AI models through machine learning. The stolen information included details about chips and systems, and software that helps power a supercomputer "capable of executing at the cutting edge of machine learning and AI technology," the indictment said. Google designed some of the allegedly stolen chip blueprints to gain an edge over cloud computing rivals Amazon.com and Microsoft, which design their own, and reduce its reliance on chips from Nvidia.

Hired by Google in 2019, Ding allegedly began his thefts three years later, while he was being courted to become chief technology officer for an early-stage Chinese tech company, and by May 2023 had uploaded more than 500 confidential files. The indictment said Ding founded his own technology company that month, and circulated a document to a chat group that said "We have experience with Google's ten-thousand-card computational power platform; we just need to replicate and upgrade it." Google became suspicious of Ding in December 2023 and took away his laptop on Jan. 4, 2024, the day before Ding planned to resign. A Google spokesperson said: "We have strict safeguards to prevent the theft of our confidential commercial information and trade secrets. After an investigation, we found that this employee stole numerous documents, and we quickly referred the case to law enforcement."

An anonymous reader quotes a report from The Guardian: The average toddler is missing out on hearing more than 1,000 words spoken by an adult each day due to screen time, setting back their language skills, a first-of-its kind study has found. The research, published on Tuesday in the Journal of the American Medical Association (Jama) Pediatrics, tracked 220 Australian families over two years to measure the relationship between family screen use and children's language environment. Families recorded all the audio around their child using advanced speech recognition technology over a 16-hour period on an average day at home. They repeated this process every six months between the ages of 12 and 36 months. The lead researcher, Dr Mary Brushe from the Telethon Kids Institute, said: "The technology we use is essentially like a Fitbit, but instead of counting the number of steps, this device counts the number of words spoken by, to and around the child." The device also picked up electronic noise, which the researchers analyzed to calculate screen time.

The researchers found young children's exposure to screens including TVs and phones was interfering with their language opportunities, with the association most pronounced at three years of age. For every extra minute of screen time, the three-year-olds in the study were hearing seven fewer words, speaking five fewer words themselves and engaging in one less conversation. The study found the average three-year-old in the study was exposed to two hours and 52 minutes of screen time a day. Researchers estimated this led to those children being exposed to 1,139 fewer adult words, 843 fewer child words and 194 fewer conversations. Because the study couldn't capture parents' silent phone use, including reading emails, texting or quietly scrolling through websites or social media, Brushe said they might have underestimated how much screen usage is affecting children.

A language-rich home environment was critical in supporting infants and toddlers' language development, Brushe said. While some educational children's shows were designed to help children's language skills, very young kids in the age group of the study could struggle to translate television shows into their own life, she said. This study did not differentiate between whether children were watching high- or low-quality screen content.

In an update on CBS' "60 Minutes" on Sunday, Blue Origin said it was aiming to send an uncrewed lander to the surface of the moon in the next 12 to 16 months. A crewed version is expected to follow. GeekWire reports: "We're expecting to land on the moon between 12 and 16 months from today," [said John Couluris, senior vice president for lunar permanence at Blue Origin]. "I understand I'm saying that publicly, but that's what our team is aiming towards." Couluris was referring to a pathfinder version of Blue Origin's nearly three-story-tall Blue Moon Mark 1 cargo lander, which is taking shape at Blue Origin's production facility in Huntsville, Ala. The Pathfinder Mission would demonstrate the MK1's capabilities -- including its hydrogen-fueled BE-7 engine, its precision landing system and its ability to deliver up to 3 tons of payload anywhere on the moon.

Blue Origin envisions building multiple cargo landers, as well as a crewed version of the Blue Moon lander that could transport NASA astronauts to and from the lunar surface. The MK1 cargo lander is designed for a single launch and delivery, but the crewed lander would be reusable. "We'll launch them to lunar orbit, and we'll leave them there," Couluris explained. "And we'll refuel them in orbit, so that multiple astronauts can use the same vehicle back and forth."

The Pathfinder Mission would be funded by Blue Origin, but NASA is providing support for other Blue Moon missions. Blue Origin's $3.4 billion contract with NASA calls for the crewed lander to be available for the Artemis 5 moon mission by 2029, with an uncrewed test flight as part of the buildup. The in-space refueling operation would make use of a cislunar transporter, built by Lockheed Martin, that could travel between low Earth orbit and lunar orbit with supplies. "We are now building with NASA the infrastructure to ensure lunar permanency," Couluris said. NASA is providing funding for the Blue Moon landing system as an alternative to SpaceX's Starship system, which is under development at SpaceX's Starbase in South Texas. The crewed Starship lunar lander is scheduled to come into play for Artemis 3, a milestone landing mission that's currently scheduled for 2026. [...]

Blue Origin plans to send the MK1 lander to the moon on its reusable New Glenn rocket, which is also under development. A couple of weeks ago, a pathfinder version of that rocket was raised on a Florida launch pad for the first time, and it's currently going through a series of cryogenic tanking tests. Blue Origin CEO Dave Limp, who was brought over to the company from Amazon last year to accelerate work on New Glenn, said in a LinkedIn post that he's "looking forward to bringing this heavy-lift capacity to our customers later this year." One of the early launches is tasked with sending a pair of NASA probes to Mars.

Wayne Williams reports via TechRadar: Qualcomm has unveiled its AI Hub, an all-inclusive library of pre-optimized AI models ready for use on devices running on Snapdragon and Qualcomm platforms. These models support a wide range of applications including natural language processing, computer vision, and anomaly detection, and are designed to deliver high performance with minimal power consumption, a critical factor for mobile and edge devices. The AI Hub library currently includes more than 75 popular AI and generative AI models including Whisper, ControlNet, Stable Diffusion, and Baichuan 7B. All models are bundled in various runtimes and are optimized to leverage the Qualcomm AI Engine's hardware acceleration across all cores (NPU, CPU, and GPU). According to Qualcomm, they'll deliver four times faster inferencing times.

The AI Hub also handles model translation from the source framework to popular runtimes automatically. It works directly with the Qualcomm AI Engine direct SDK and applies hardware-aware optimizations. Developers can search for models based on their needs, download them, and integrate them into their applications, saving time and resources. The AI Hub also provides tools and resources for developers to customize these models, and they can fine-tune them using the Qualcomm Neural Processing SDK and the AI Model Efficiency Toolkit, both available on the platform.

Anthropic, a leading artificial intelligence startup, unveiled its Claude 3 series of AI models today, designed to meet the diverse needs of enterprise customers with a balance of intelligence, speed, and cost efficiency. The lineup includes three models: Opus, Sonnet, and the upcoming Haiku. From a report: The star of the lineup is Opus, which Anthropic claims is more capable than any other openly available AI system on the market, even outperforming leading models from rivals OpenAI and Google. "Opus is capable of the widest range of tasks and performs them exceptionally well," said Anthropic cofounder and CEO Dario Amodei in an interview with VentureBeat. Amodei explained that Opus outperforms top AI models like GPT-4, GPT-3.5 and Gemini Ultra on a wide range of benchmarks. This includes topping the leaderboard on academic benchmarks like GSM-8k for mathematical reasoning and MMLU for expert-level knowledge.

"It seems to outperform everyone and get scores that we haven't seen before on some tasks," Amodei said. While companies like Anthropic and Google have not disclosed the full parameters of their leading models, the reported benchmark results from both companies imply Opus either matches or surpasses major alternatives like GPT-4 and Gemini in core capabilities. This, at least on paper, establishes a new high watermark for commercially available conversational AI. Engineered for complex tasks requiring advanced reasoning, Opus stands out in Anthropic's lineup for its superior performance. Sonnet, the mid-range model, offers businesses a more cost-effective solution for routine data analysis and knowledge work, maintaining high performance without the premium price tag of the flagship model. Meanwhile, Haiku is designed to be swift and economical, suited for applications such as consumer-facing chatbots, where responsiveness and cost are crucial factors. Amodei told VentureBeat he expects Haiku to launch publicly in a matter of "weeks, not months."

It completed the most precise landing ever on the moon — albeit upside-down. And then it faced a "lunar night" lasting about two weeks where temperatures drop to -270 degrees Fahrenheit, reports the Times of India.

But then, "Despite not being designed for the extreme temperatures, SLIM surprised scientists by coming back to life after the two-week-long lunar night." More from Space.com: The lander woke up on February 26 during extremely hot temperatures of 212 Fahrenheit (100 Celsius) in its region and has been making contact here and there with Earth in the days since. Most recently, SLIM attempted observations with its multiband spectroscopic camera, but "it did not work properly," JAXA officials wrote. "This seems to be due to the effects of overnight," the update continued, referring to the frigid two-week-long lunar night that SLIM experienced before the sun shone near Shioli crater again. "But we will continue to investigate based on the data we have obtained for the next opportunity...."
"We received so much support for our operations after the lunar night," the agency posted on social media — adding "thank you!"

The Times of India reports that "JAXA officially announced SLIM's return to a dormant state on March 1, sharing an image of the lunar surface captured by the probe."

Above the photo, JAXA posted this hopeful message. "Although the probability of a failure increases with the repeated severe temperature cycles, SLIM operation will attempt to resume when the sun rises (late March). #GoodAfterMoon."

And Space.com notes that "Despite all, SLIM has met both main and extended mission objectives: Landing precisely on the moon, deploying two tiny rovers and conducting science with its navigation camera and its spectroscopic camera, particularly searching for signs of olivine on the surface."

Thanks to long-time Slashdot reader AmiMoJo for sharing the news.

An anonymous reader quotes a report from Gizmodo: For 20 years, a loosely organized group of Wikipedia editors toiled away curating a collection of 15,000 articles on a single subject: the roads and highways of the United States. Despite minor disagreements, the US Roads Project mostly worked in harmony, but recently, a long-simmering debate over the website's rules drove this community to the brink. Efforts at compromise fell apart. There was a schism, and in the fall of 2023, the editors packed up their articles and moved over to a website dedicated to roads and roads alone. It's called AARoads, a promised land where the editors hope, at last, that they can find peace. "Roads are a background piece. People drive on them every day, but they don't give them much attention," said editor Michael Gronseth, who goes by Imzadi1979 on Wikipedia, where he dedicated his work to Michigan highways, specifically. But a road has so much to offer if you look beyond the asphalt. It's the nexus of history, geography, travel, and government, a seemingly perfect subject for the hyper-fixations of Wikipedia. "But there was a shift about a year ago," Gronseth said. "More editors started telling us that what we're doing isn't important enough, and we should go work on more significant topics." [...]

The Roads Project had a number of adversaries, but the chief rival is a group known as the New Page Patrol, or the NPP for short. The NPP has a singular mission. When a new page goes up on Wikipedia, it gets reviewed by the NPP. The Patrol has special editing privileges and if a new article doesn't meet the website's standards, the NPP takes it down. "There's a faction of people who feel that basically anything is valid to be published on Wikipedia. They say, 'Hey, just throw it out there! Anything goes.' That's not where I come down." said Bil Zeleny, a former member of the NPP who goes by onel5969 on Wikipedia, a reference to the unusual spelling of his first name. At his peak, Zeleny said he was reviewing upwards of 100,000 articles a year, and he rejected a lot of articles about roads during his time. After years of frustration, Zeleny felt he was seeing too many new road articles that weren't following the rules -- entire articles that cited nothing other than Google Maps, he said. Enough was enough. Zeleny decided it was time to bring the subject to the council.

Zeleny brought up the problem on the NPP discussion forum, sparking months of heated debate. Eventually, the issue became so serious that some editors proposed an official policy change on the use of maps as a source. Rule changes require a process called "Request for Comment," where everyone is invited to share their thoughts on the issue. Over the course of a month, Wikipedia users had written more than 56,000 words on the subject. For reference, that's about twice as long as Ernest Hemingway's novel The Old Man and the Sea. In the end, the roads project was successful. The vote was decisive, and Wikipedia updated its "No Original Research" policy to clarify that it's ok to cite maps and other visual sources. But this, ultimately, was a victory with no winners. "Some of us felt attacked," Gronseth said. On the US Roads Project's Discord channel, a different debate was brewing. The website didn't feel safe anymore. What would happen at the next request for comment? The community decided it was time to fork. "We don't want our articles deleted. It didn't feel like we had a choice," he said.

The Wikipedia platform is designed for interoperability. If you want to start your own Wiki, you can split off and take your Wikipedia work with you, a process known as "forking." [...] Over the course of several months, the US Roads Project did the same. Leaving Wikipedia was painful, but the fight that drove the roads editors away was just as difficult for people on the other side. Some editors embroiled in the roads fights deleted their accounts, though none of these ex-Wikipedian's responded to Gizmodo's requests for comment. Bil Zeleny was among the casualties. After almost six years of hard work on the New Post Patrol, he reached the breaking point. The controversy had pushed him too far, and Zeleny resigned from the NPP. [...] AARoads actually predates Wikipedia, tracing its origins all the way back to the prehistoric internet days of the year 2000, complete with articles, maps, forums, and a collection of over 10,000 photos of highway signs and markers. When the US Roads Project needed a new home, AARoads was happy to oblige. It's a beautiful resource. It even has backlinks to relevant non-roads articles on the regular Wikipedia. But for some, it isn't home. "There are members who disagree with me, but my ultimate goal is to fork back," said Gronseth. "We made our articles license-compatible, so they can be exported back to Wikipedia someday if that becomes an option. I don't want to stay separate. I want to be part of the Wikipedia community. But we don't know where things will land, and for now, we've struck out on our own."

The Supreme Court of Canada ruled today that police must now have a warrant or court order to obtain a person or organization's IP address. CBC News reports: The top court was asked to consider whether an IP address alone, without any of the personal information attached to it, was protected by an expectation of privacy under the Charter. In a five-four split decision, the court said a reasonable expectation of privacy is attached to the numbers making up a person's IP address, and just getting those numbers alone constitutes a search. Writing for the majority, Justice Andromache Karakatsanis wrote that an IP address is "the crucial link between an internet user and their online activity." "Thus, the subject matter of this search was the information these IP addresses could reveal about specific internet users including, ultimately, their identity." Writing for the four dissenting judges, Justice Suzanne Cote disagreed with that central point, saying there should be no expectation of privacy around an IP address alone. [...]

In the Supreme Court majority decision, Karakatsanis said that only considering the information associated with an IP address to be protected by the Charter and not the IP address itself "reflects piecemeal reasoning" that ignores the broad purpose of the Charter. The ruling said the privacy interests cannot be limited to what the IP address can reveal on its own "without consideration of what it can reveal in combination with other available information, particularly from third-party websites." It went on to say that because an IP address unlocks a user's identity, it comes with a reasonable expectation of privacy and is therefore protected by the Charter. "If [the Charter] is to meaningfully protect the online privacy of Canadians in today's overwhelmingly digital world, it must protect their IP addresses," the ruling said.

Justice Cote, writing on behalf of justices Richard Wagner, Malcolm Rowe and Michelle O'Bonsawin, acknowledged that IP addresses "are not sought for their own sake" but are "sought for the information they reveal." "However, the evidentiary record in this case establishes that an IP address, on its own, reveals only limited information," she wrote. Cote said the biographical personal information the law was designed to protect are not revealed through having access to an IP address. Police must use that IP address to access personal information that is held by an ISP or a website that tracks customers' IP addresses to determine their habits. "On its own, an IP address does not even reveal browsing habits," Cote wrote. "What it reveals is a user's ISP -- hardly a more private piece of information than electricity usage or heat emissions." Cote said placing a reasonable expectation of privacy on an IP address alone upsets the careful balance the Supreme Court has struck between Canadians' privacy interests and the needs of law enforcement. "It would be inconsistent with a functional approach to defining the subject matter of the search to effectively hold that any step taken in an investigation engages a reasonable expectation of privacy," the dissenting opinion said.

Stack Overflow has launched an API that will require all AI models trained on its coding question-and-answer content to attribute sources linking back to its posts. And it will cost money to use the site's content. From a report: "All products based on models that consume public Stack Overflow data are required to provide attribution back to the highest relevance posts that influenced the summary given by the model," it confirmed in a statement. The Overflow API is designed to act as a knowledge database to help developers build more accurate and helpful code-generation models. Google announced it was using the service to access relevant information from Stack Overflow via the API and integrate the data with its latest Gemini models, and for its cloud storage console.