"Amazon is discontinuing its Echo Look camera, a standalone device that gave owners fashion advice using artificial intelligence and machine learning," reports The Verge. The gadget raised eyebrows when it was first announced as it included a virtual assistant with a microphone and a camera specifically designed to go somewhere in your bedroom, bathroom, or wherever the hell you get dressed. From the report: The Look's companion app and the device itself will stop functioning on July 24th. Between now and July 24th, 2021, Look users can back up their images and videos by making a free Amazon Photos account. (People with existing Photos accounts will have their media backed up automatically.) Anyone who wants to delete all their existing photos and videos will have to do so before the July 2020 deadline; otherwise, they'll have to call Amazon's customer service to have them deleted. They can currently delete them through the Look app.

Amazon points out that much of the Echo Look's functionality is now included in the Amazon Shopping app, including Style by Alexa, which involves the AI offering fashion pointers. The company says people should download the app to keep consulting with Amazon, and they should also recycle their Look through Amazon's program.

silverjacket writes: A story in this week's issue of Science. Artificial intelligence (AI) just seems to get smarter and smarter. Each iPhone learns your face, voice, and habits better than the last, and the threats AI poses to privacy and jobs continue to grow. The surge reflects faster chips, more data, and better algorithms. But some of the improvement comes from tweaks rather than the core innovations their inventors claim -- and some of the gains may not exist at all, says Davis Blalock, a computer science graduate student at the Massachusetts Institute of Technology (MIT). Blalock and his colleagues compared dozens of approaches to improving neural networks -- software architectures that loosely mimic the brain. "Fifty papers in," he says, "it became clear that it wasn't obvious what the state of the art even was." The researchers evaluated 81 pruning algorithms, programs that make neural networks more efficient by trimming unneeded connections. All claimed superiority in slightly different ways. But they were rarely compared properly -- and when the researchers tried to evaluate them side by side, there was no clear evidence of performance improvements over a 10-year period. The result [PDF], presented in March at the Machine Learning and Systems conference, surprised Blalock's Ph.D. adviser, MIT computer scientist John Guttag, who says the uneven comparisons themselves may explain the stagnation. "It's the old saw, right?" Guttag said. "If you can't measure something, it's hard to make it better."

Researchers are waking up to the signs of shaky progress across many subfields of AI. A 2019 meta-analysis of information retrieval algorithms used in search engines concluded the "high-water mark ... was actually set in 2009." Another study in 2019 reproduced seven neural network recommendation systems, of the kind used by media streaming services. It found that six failed to outperform much simpler, nonneural algorithms developed years before, when the earlier techniques were fine-tuned, revealing "phantom progress" in the field. In another paper posted on arXiv in March, Kevin Musgrave, a computer scientist at Cornell University, took a look at loss functions, the part of an algorithm that mathematically specifies its objective. Musgrave compared a dozen of them on equal footing, in a task involving image retrieval, and found that, contrary to their developers' claims, accuracy had not improved since 2006. "There's always been these waves of hype," Musgrave says.

Amazon has suspended thousands of third-party sellers for price gouging during the COVID-19 pandemic, but sellers have found a loophole to avoid detection when raising prices: labeling their products as "collectibles," even if there's no conceivable way they could be. The Verge reports: Take this Bowflex dumbbell set, which my colleague Casey Newton encountered while browsing Amazon. Before it sold out in mid-March, Amazon had been selling the weights for $279. This week, the only available weights were from 20 sellers who were offering the dumbbells for between $899 and $1,275 (with free shipping). All had listed the item as "collectible." The automated systems that detect price gouging appear not to monitor products if their condition is listed as "collectible" rather than "new."
[...]
The collectible loophole has existed for some time, sellers say, but before the pandemic, it was rarely used, because sellers only ran into price ceilings when an item was unusually popular and in short supply [...]. But with COVID-19, entire product categories -- cleaning supplies, webcams, home gym equipment -- saw unprecedented demand. Supply-chain disruptions meant some items were already running low, and safety concerns in Amazon's warehouses meant they took longer to restock and ship out. Amazon, and then sellers with normally priced goods, quickly sold out. The remaining sellers raised their prices, sometimes deliberately and sometimes using automated repricing software, and started running into price ceilings. As Amazon's marketplace came under strain, mechanisms that typically work in the background were brought to the fore.

An anonymous reader quotes a report from Axios: Sen. Ted Cruz (R-Texas), in a letter Friday to the Justice and Treasury departments, is calling for a criminal investigation of Twitter over allegations the company is violating U.S. sanctions against Iran. Twitter is already under fire from President Trump for adding fact checks and a warning label, respectively, to misleading and incendiary tweets he made in recent days. Cruz's letter adds another dimension to the tech company's woes in Washington.

Twitter allows Iranian leaders to maintain accounts on its service, and Cruz is asking Attorney General Bill Barr and Treasury Secretary Steven Mnuchin to probe whether that violates U.S. sanctions prohibiting American companies from providing goods or services to the country's top officials. "I believe that the primary goal of (the International Emergency Economic Powers Act) and sanctions law should be to change the behavior of designated individuals and regimes, not American companies," Cruz wrote."But when a company willfully and openly violates the law after receiving formal notice that it is unlawfully supporting designated individuals, the federal government should take action." In February, Cruz led a letter from Republican senators to Twitter CEO Jack Dorsey, calling on the company to ban Iranian leaders from the site for the same reasons. Twitter responded in April, arguing that its service is exempt from the sanctions, and that the public conversation on the platform is critically important during the coronavirus pandemic.

"Fundamental values of openness, free expression, public accountability, and mutual understanding matter now more than ever," Vijaya Gadde, Twitter's legal, public policy & trust and safety lead, wrote. "Regardless of the political agenda of a particular nation state, to deny our service to their leaders at a time like this would be antithetical to the purpose of our company, which is to serve the global public conversation."

Google announced this week plans to enable its new anti-notification spam system in Chrome over the summer, with the release of Chrome 84, on July 12, 2020. From a report: Known internally as the "quieter notification permission UI," this Chrome component works by blocking sites from showing notification requests, which are hidden under an icon in the Chrome URL bar (on desktop) or under a toolbar (on mobile). Google first announced the "quieter notification permission UI" in January, and shipped it in February, in Chrome 80, in a limited, user opt-in fashion. But in a blog post, Google said the new UI and its ability to detect spammy notification popups has been improved and will roll out enabled by default for all users in July, with the release of Chrome 84.

An anonymous reader quotes a report from The Register: A very specific software bug made airliners turn the wrong way if their pilots adjusted a pre-set altitude limit. The bug, discovered on Bombardier CRJ-200 aircraft fitted with Rockwell Collins Aerospace-made flight management systems (FMSes), led to airliners trying to follow certain missed approaches turning right instead of left -- or vice versa.

First discovered in 2017, the flaw was only apparent when pilots manually edited a pre-set "climb to" altitude programmed into a "missed approach" procedure following an Instrument Landing System approach. It also arose if pilots used the FMS's temperature compensation function in extremely cold weather. In theory the bug could have led to airliners crashing into the ground, though the presence of two trained and alert humans in the cockpit monitoring what the aircraft was doing made this a remote possibility. "The bug was first uncovered when a CRJ-200 crew flying into Canada's Fort St John airport used the FMS's temperature correction function," the report adds. "They discovered that the software turned their aeroplane in the wrong direction while it was following the published missed approach, something that generally does not happen. The fault was swiftly reported to the authorities and the relevant manufacturers."

Full details, including the maths, are available here. The U.S. Federal Aviation Authorities also published a Powerpoint presentation (PDF) about the bug.

Twitter and Reddit have filed an amicus brief in support of a lawsuit challenging a U.S. government rule change compelling visa applicants to disclose their social media handles. From a report: The lawsuit, brought by the Knight First Amendment Institute at Columbia University, the Brennan Center for Justice, and law firm Simpson Thacher & Bartlett, seeks to undo both the State Department's requirement that visa applicants must disclose their social media handles prior to obtaining a U.S. visa, as well as related rules over the retention and dissemination of those records. Last year, the State Department began asking visa applicants for their current and former social media usernames, a move that affects millions non-citizens applying to travel to the United States each year. The rule change was part of the Trump administration's effort to expand its "enhanced" screening protocols. At the time, it was reported that the information would be used if the State Department determines that "such information is required to confirm identity or conduct more rigorous national security vetting."

Google, facing an advertising slump caused by the pandemic, has rescinded offers to several thousand people who had agreed to work at the company as temporary and contract workers. From a report: "We're slowing our pace of hiring and investment, and are not bringing on as many new starters as we had planned at the beginning of the year," Google said in an email to contracting agencies last week that was seen by The New York Times. The company told the firms that it "will not be moving forward to onboard" the people that the agencies had recruited to work at Google. The move affected more than 2,000 people globally who had signed offers with the agencies to be a contract or temp worker, according to three people familiar with the decision, who spoke on the condition of anonymity because they were not allowed to speak publicly on the matter. Google employs more than 130,000 contractors and temp workers, a shadow work force that outnumbers its 123,000 full-time employees. Google's full-time staff are rewarded with high salaries and generous perks, but temps and contractors often receive less pay, fewer benefits and do not have the same protections, even though they work alongside full timers.

Cal Newport, writing at the New Yorker: Technological transitions often stumble when we expect them to sprint. In 1989, the Stanford economist Paul David wanted to understand why so many companies were so slow to adopt computer technology; for historical perspective, he turned to the history of the electric dynamo, which had been invented around a hundred years before, and which, before it transformed industrial production, had also been adopted slowly. In his paper "The Dynamo and the Computer: An Historical Perspective on the Modern Productivity Paradox," published in the American Economic Review, David explained that, at the turn of the century, most factories were powered by massive central steam engines. The engines turned overhead shafts, which were connected by an intricate array of belts and pulleys to close-packed machinery. When electric motors were first introduced, factory owners tried to integrate them into their existing setups; often, they'd simply replace the hulking steam engine with a giant electric dynamo. This introduced some conveniences -- no one had to shovel coal -- but also created complexities. It was hard to keep all the electrical components working; many factory owners opted to stay with steam.

It took decades for factory owners to figure out how to make the most of electric power. Eventually, they discovered that the best approach was to put a small motor on each individual piece of machinery. Since a factory no longer needed to draw power from a central engine, its equipment could be spread out. This, in turn, changed the nature of industrial architecture. Buildings that no longer required reinforced ceilings to house shafts, belts, and pulleys could incorporate windows and skylights, of the sort we know today from urban loft buildings. Inertia, David found, had been part of the problem. Factory owners who had spent a lot of money and time building physical plants organized around central-drive trains were reluctant to commit to complex, expensive overhauls. There were imaginative obstacles: powering each machine with its own individual motor may seem like an obvious idea now, but in fact it represented a sharp break from the centralized-power model that had dominated for the previous hundred and fifty years. Finally, technological barriers stood in the way -- small issues, compared to the invention of electricity, but persistent and important ones nonetheless. Someone, for instance, had to figure out how to construct a building-wide power grid capable of handling the massively variable load created by many voltage-hungry mini-motors being turned off and on unpredictably. Until that happened, it was central power or bust.

In some respects, we may be in an electric-dynamo moment for remote work. In theory, we have the technology we need to make remote work workable. And yet most companies that have tried to graft it onto their existing setups have found only mixed success. In response, many have stuck with what they know. Now the coronavirus pandemic has changed the equation. Whole workplaces have gone remote; steam engines have been outlawed. The question is whether, having been forced to embrace this new technology, we can solve the long-standing problems that have thwarted its adoption in the past. Some useful innovation is possible on an individual level. As a newly minted remote worker, you may find that demands on your attention are actually more incessant and intrusive than they used to be -- a natural consequence when a workplace depends more than ever on phone calls, e-mails, and video conferences. You might respond by consolidating all of your appointments into a given half of the day -- say, between 1 p.m. and 5 p.m. -- preserving the other hours, by default, for actually working on the items discussed.

The US National Security Agency (NSA) has published a security alert warning of a new wave of cyberattacks against email servers, attacks conducted by one of Russia's most advanced cyber-espionage units. From a report: The NSA says that members of Unit 74455 of the GRU Main Center for Special Technologies (GTsST), a division of the Russian military intelligence service, have been attacking email servers running the Exim mail transfer agent (MTA). Also known as "Sandworm," this group has been hacking Exim servers since August 2019 by exploiting a critical vulnerability tracked as CVE-2019-10149, the NSA said in a security alert shared today with ZDNet. "When Sandworm exploited CVE-2019-10149, the victim machine would subsequently download and execute a shell script from a Sandworm-controlled domain," the NSA says.

Two years ago, Google open-sourced Plus Codes, a digital addressing system to help billions of people navigate to places that don't have clear addresses. The company said this week it is making it easier for anyone with an Android device to share its rendition of an address -- a six-digit alphanumeric code. From a report: Google Maps users on Android can now tap the blue dot that represents their current location to view and share their unique six-digit coordinate with friends. Anyone with the code can look it up on Google Maps or Google Search to get the precise location of the destination. The codes look like this: G6G4+CJ Delhi, India. Google says it divides the geographical surface of the world into tiled areas and attributes a unique six-letter code and the name of the city and country to each of them. More than 2 billion people on the planet either don't have an address or have an address that isn't easy to locate. This challenge is more prevalent in developed markets such as India where a street address could often be as long as a paragraph, and where people often rely on nearby landmarks to navigate their way.

From a company-wide memo sent by Rony Abovitz, the founder of Magic Leap, which has raised nearly $3 billion from high-profile investors including Andreessen Horowitz, Google, AT&T, NTT DoCoMo, Public Investment Fund of Saudi Arabia, JP Morgan, Kleiner Perkins Caufield & Byers: As we've shared over the last several weeks, in order to set Magic Leap on a course for success, we have pivoted to focus on delivering a spatial computing platform for enterprise. Technology columnist John Gruber, translates: As nearly everyone has finally realized, our actual technology is nothing at all like what we promised, lied about for years, and sold gullible deep-pocketed investors on. Our con is falling apart at the seams, so we'll milk the last few dollars out of the only investors dumb enough to give us even more money, by repeating the word "enterprise" and doing that thing with our fingers like Obi-Wan Kenobi. Magic Leap's memo adds: We have closed significant new funding and have very positive momentum towards closing key strategic enterprise partnerships. Gruber, continues translation: You're not going to believe this but we somehow raised another $350 million. I know, right? Magic Leap: As the board and I planned the changes we made and what Magic Leap needs for this next focused phase, it became clear to us that a change in my role was a natural next step. Gruber: Everyone agrees the jig is up. Magic Leap: I discussed this with the board and we have agreed that now is the time to bring in a new CEO who can help us to commercialize our focused plan for spatial computing in enterprise. We have been actively recruiting candidates for this role and I look forward to sharing more soon. Gruber: Our Craigslist ad: "Florida company seeks Bernie Madoff type." Magic Leap: I have been leading Magic Leap since 2011 (starting in my garage). We have created a new field. A new medium. And together we have defined the future of computing. Gruber: No one will remember us or anything we've doneâ--âunless Netflix makes one of those documentaries like the Fyre Festival one. I love that movie. Which makes me think maybe we should change our Craigslist ad to "Billy McFarland type." Actually, when does he get out of prison?

Twitter placed a notice on a tweet from President Trump, shielding it from view for breaking what the company said are its rules about glorifying violence [Editor's note: the link may be paywalled; alternative source]. From a report: Mr. Trump's tweet was a comment on the violent protests in Minnesota. The post can now only be seen after users click a box with a notice saying it violated Twitter's rules against encouraging violence, but it otherwise remains visible. "We've taken action in the interest of preventing others from being inspired to commit violent acts, but have kept the Tweet on Twitter because it is important that the public still be able to see the Tweet given its relevance to ongoing matters of public importance," Twitter said on its official communications account.

This is the first time such a step has been taken against a head of state for breaking Twitter's rules about glorifying violence, a company spokesman said. The company said users' ability to interact with the tweet will be limited, and that users can retweet it with comment, but not like, reply to, or otherwise retweet it. "...These THUGS are dishonoring the memory of George Floyd, and I won't let that happen. Just spoke to Governor Tim Walz and told him that the Military is with him all the way. Any difficulty and we will assume control but, when the looting starts, the shooting starts. Thank you!," Mr. Trump's tweet said. The official account of the White House, which tweeted Trump's message, has been flagged as well.

An anonymous reader quotes a report from the BBC: Uber is destroying thousands of electric bikes and scooters, after selling its Jump business to Lime. Videos of its red bikes being crushed at a recycling centre were shared on social media, angering cycling advocates. Uber said it had decided to destroy thousands of its older-model vehicles due to maintenance, liability and safety concerns.

In 2018, Uber said it would focus more on its electric bike and scooter business than on cars. But on May 7 this year, Uber announced a deal that saw Lime take over the Jump bike business. As part of the deal, Uber invested $170 million in Lime, while Lime acquired "tens of thousands" of Uber's Jump bikes -- and the associated intellectual property. Lime's chief executive Wayne Ting has said he prefers the design of Uber's bikes and will deploy more of them in the future. However, there were also "tens of thousands" of older-model bikes that Lime did not inherit as part of the deal. Videos shared on Twitter show the bikes arriving at a recycling facility in North Carolina to be destroyed. "We explored donating the remaining, older-model bikes," Uber said in a statement. "But given many significant issues -- including maintenance, liability, safety concerns, and a lack of consumer-grade charging equipment -- we decided the best approach was to responsibly recycle them."

The decision to destroy these bikes comes amid a national bike shortage. "We have never seen anything like this in a very long time," said Dave Nghiem at College Park Bicycles in College Park, MD. "We have never locked down half the planet like this so they can't do their jobs to build bikes. So, no one has been building bikes for three months. If no one is building bikes, there's no bikes on the continent," said Dave.

Kurt of Bike Share Museum seems to think it is all about killing Jump, "destroying every bike they can, and slowly taking Lime down in the process." He adds: "We also can't emphasize enough how disgusting it is for UBER to scrap 20,000 bicycles in the midst of an unprecedented pandemic where bicycles have literally become an object of survival. Heavy as they are, these could be transportation for the many who have been brought to financial ruin during COVID-19."

In response to a cyberattack on the German Parliament in 2015, Germany wants to impose a European travel ban and asset freeze on those responsible. Reuters reports: Russia has rejected allegations that its military intelligence was behind the cyber attack after media reported that data had been stolen, including emails from Chancellor Angela Merkel's constituency office. State Secretary Miguel Berger told the ambassador that the government would call for the EU's cyber sanctions mechanism to be invoked against those responsible for the attack, said the German ministry in a statement. The EU last year approved a system to freeze hackers' assets in the bloc and banning them from entry.

Federal prosecutors issued an arrest warrant on May 5 for Russian national Dmitry Badin over the attack and the German ministry said there was credible evidence that he was part of the GRU military intelligence service at the time of the attack. "The arrest warrant against Mr Badin was issued on the basis of the strong suspicion that the accused conspired with other hitherto anonymous persons to carry out intelligence activities against Germany on behalf of the secret service of a foreign power," said the ministry. In a statement on Wednesday, the Russian embassy in Berlin said German officials so far had not been able to present facts to underpin the accusations against Moscow.

The American Civil Liberties Union on Thursday sued the facial recognition start-up Clearview AI (alternative source), which claims to have helped hundreds of law enforcement agencies use online photos to solve crimes, accusing the company of "unlawful, privacy-destroying surveillance activities." The New York Times reports: In a suit filed in Illinois, the A.C.L.U. said that Clearview violated a state law that forbids companies from using a resident's fingerprints or face scans without consent. Under the law, residents have the right to sue companies for up to $5,000 per privacy violation. "The bottom line is that, if left unchecked, Clearview's product is going to end privacy as we know it," said Nathan Freed Wessler, a lawyer at the A.C.L.U., "and we're taking the company to court to prevent that from happening."

The suit, filed in the Circuit Court of Cook County, adds to the growing backlash against Clearview since January, when The New York Times reported that the company had amassed a database of more than three billion photos across the internet, including from Facebook, YouTube, Twitter and Venmo. This trove of photos enables anyone with the Clearview app to match a person to their online photos and find links back to the sites where the images originated. People in New York and Vermont have also filed suits in against the company in recent months, and the state attorneys general of Vermont and New Jersey have ordered Clearview to stop collecting residents' photos. According to the A.C.L.U. suit, "Clearview has set out to do what many companies have intentionally avoided out of ethical concerns: create a mass database of billions of face prints of people, including millions of Illinoisans, entirely unbeknownst to those people, and offer paid access to that database to private and governmental actors worldwide." The company's business model, the complaint said, "appears to embody the nightmare scenario" of a "private company capturing untold quantities of biometric data for purposes of surveillance and tracking without notice to the individuals affected, much less their consent."

Google has been hit by a lawsuit filed by Arizona Attorney General Mark Brnovich, alleging the search giant deceived its users in order to collect location data from their phones. From a report: The company generates the vast majority of its revenue through its massive advertising operation, which is buttressed by personal information Google collects when people use its products. But users were "lulled into a false sense of security" because Google led users to believe they disabled settings for location data gathering, when they were still turned on, Brnovich wrote on Twitter. "Google collects detailed information about its users, including their physical locations, to target users for advertising," Brnovich wrote. "Often, this is done without the users' consent or knowledge." The lawsuit seeks damages, but the amount is unclear. Brnovich's office didn't respond to a request for comment.

An anonymous reader quotes a report from VentureBeat: Google today launched Android Studio 4.0, the latest version of its integrated development environment (IDE). Android Studio 4.0 is supposed to help developers "code smarter, build faster, and design apps." Version 4.0 includes a new Motion Editor, a Build Analyzer, and Java 8 language APIs. Google also overhauled the CPU Profiler user interface and improved the Layout Inspector. [In the article] you'll find Android Studio 4.0 features broken down by category: design, develop, and build. The new version also includes the usual performance improvements and bug fixes on top of the new features (full release notes). Google didn't share its plans for the next version. Normally we'd get hints at the company's I/O developer conference, but 2020 is a weird year.

YouTube sparked widespread speculation about its moderation policies this week after it admitted to accidentally deleting comments that contained phrases critical of the Chinese Communist Party (CCP). Today, the company told The Verge that the issue was not the result of outside interference -- an explanation for the error floated by many. The Verge reports: The phrases that triggered automatic deletion included "communist bandit" and "50-cent party," a slang term for internet users paid to defend the CCP. Some speculated that an outside group, perhaps connected to the CCP, manipulated YouTube's automated filters by repeatedly reporting these phrases, causing the algorithm to tag them as offensive. Speaking to The Verge, YouTube spokesperson Alex Joseph denied that this happened and said that, contrary to popular belief, YouTube never removes comments only on the basis of user reports.

"This was not the result of outside interference, and we only remove content when our enforcement system determines it violates our Community Guidelines, not solely because it's flagged by users," said Joseph. "This was an error with our enforcement systems and we have rolled out a fix."

An anonymous reader quotes a report from Ars Technica: Believers of 5G conspiracy theories have apparently been buying a $350 anti-5G USB key that -- not surprisingly -- appears to just be a regular USB stick with only 128MB of storage. As noted by the BBC today, the "5GBioShield" USB stick "was recommended by a member of Glastonbury Town Council's 5G Advisory Committee, which has called for an inquiry into 5G." The company that sells 5GBioShield claims it "is the result of the most advanced technology currently available for balancing and prevention of the devastating effects caused by non-natural electric waves, particularly (but not limited to) 5G, for all biological life forms." The product's website charges 283 British pounds for a single 5GBioShield, which converts to nearly $350. That's what it costs to get "protection for your home and family, thanks to the wearable holographic nano-layer catalyser, which can be worn or placed near to a smartphone or any other electrical, radiation or EMF emitting device." The USB stick apparently doesn't need to be plugged in to anything to work its magic. "It is always ON and working -- that's why we used quantum nano-layer technology," the company says in an FAQ.

But what does the 5GBioShield actually consist of? The BBC pointed to a recent teardown by security company Pen Test Partners, which found that the device is just a USB stick with 128MB of storage. The company wrote: "When plugged in to our test machine we may have missed the bubble of 'quantum holographic catalyzer technology' appearing. The stick comes loaded with a 25 page PDF version of the material from 5GBioShield's website. It included a Q&A of distances for the "bubble" and how to know if it is working. It's an "always on" system apparently, is always working, powered or not, so no visual checks needed. A review of the stick's properties revealed nothing more than what you'd expect from a regular 128MB USB key. We weren't even sure that 128s are still in production!" The report says that the London Trading Standards has launched a probe to investigate this product.

How will the company defend itself? BioShield Distribution Director Anna Grochowalska told the BBC, "We are in possession of a great deal of technical information, with plenty of back-up historical research," and "we are not authorized to fully disclose all this sensitive information to third parties, for obvious reasons."