The most telling element of the video, Wikoff says, is the speed the hacker demonstrates in exfiltrating the accounts’ information in real-time. The Google account’s data is stolen in around four minutes. The Yahoo account takes less than three minutes. In both cases, of course, a real account populated with tens or hundreds of gigabytes of data would take far longer to download. But the clips demonstrate how quickly that download process is set up, Wikoff says, and suggest that the hackers are likely carrying out this sort of personal data theft on a mass scale. “To see how adept they are at going in and out of all these different webmail accounts and setting them up to exfiltrate, it is just amazing,” says Wikoff. “It’s a well-oiled machine.”
In some cases, IBM’s researchers could see in the video that the same dummy accounts were also themselves being used to send phishing emails, with bounced emails to invalid addresses appearing in the accounts’ inboxes. The researchers say those bounced emails revealed some of the APT35 hackers’ targeting, including American State Department staff as well as an Iranian-American philanthropist. It’s not clear if either target was successfully phished. The dummy Yahoo account also briefly shows the phone number linked with it, which begins with Iran’s +98 country code.
In other videos the IBM researchers declined to show to WIRED, the researchers say the hackers appeared to be combing through and exfiltrating data from real victims’ accounts, rather than ones they created for training purposes. One victim was a member of the US Navy, and another was a two-decade veteran of the Greek Navy. The researchers say the APT35 hackers appear to have stolen photos, emails, tax records, and other personal information from both targeted individuals.
Screenshot: IBM
In some clips, the researchers say they observed the hackers working through a text document full of usernames and passwords for a long list of non-email accounts, from phone carrier to bank accounts, as well as some as trivial as pizza delivery and music streaming services. “Nothing was off-limits,” Wikoff says. The researchers note that they didn’t see any evidence that the hackers were able to bypass two-factor authentication, however. When an account was secured with any second form of authentication, the hackers simply moved on to the next one on their list.
The sort of targeting that IBM’s findings reveal fits with previous known operations tied to APT35, which has carried out espionage on behalf of Iran for years, most often with phishing attacks as its first point of intrusion. The group has focused on government and military targets that represent a direct challenge to Iran, such as nuclear regulators and sanctions bodies. More recently it has aimed its phishing emails at pharmaceutical companies involved in Covid-19 research and President Donald Trump’s reelection campaign.
In February of last year, the San Francisco-based research lab OpenAI announced that its AI system could now write convincing passages of English. Feed the beginning of a sentence or paragraph into GPT-2, as it was called, and it could continue the thought for as long as an essay with almost human-like coherence.
Now, the lab is exploring what would happen if the same algorithm were instead fed part of an image. The results, which were given an honorable mention for best paper award at this week’s International Conference on Machine Learning, open up a new avenue for image generation, ripe with opportunity and consequences.
At its core, GPT-2 is a powerful prediction engine. It learned to grasp the structure of the English language by looking at billions of examples of words, sentences, and paragraphs, scraped from the corners of the internet. With that structure, it could then manipulate words into new sentences by statistically predicting the order in which they should appear.
So researchers at OpenAI decided to swap the words for pixels and train the same algorithm on images in ImageNet, the most popular image bank for deep learning. Because the algorithm was designed to work with one-dimensional data, i.e.: strings of text, they unfurled the images into a single sequence of pixels. They found that the new model, named iGPT, was still able to grasp the two-dimensional structures of the visual world. Given the sequence of pixels for the first half of an image, it could predict the second half in ways that a human would deem sensible.
Below, you can see a few examples. The left-most column is the input, the right-most column is the original, and the middle columns are iGPT’s predicted completions. (See more examples here.)
OPENAI
The results are startlingly impressive and demonstrate a new path for using unsupervised learning, which trains on unlabeled data, in the development of computer vision systems. While early computer vision systems in the mid-2000s trialed such techniques before, they fell out of favor as supervised learning, which uses labeled data, proved far more successful. The benefit of unsupervised learning, however, is that it allows an AI system to learn about the world without a human filter, and significantly reduces the manual labor of labeling data.
The fact that iGPT uses the same algorithm as GPT-2 also shows its promising adaptability across domains. This is in line with OpenAI’s ultimate ambition to achieve more generalizable machine intelligence.
At the same time, the method presents a concerning new way to create deepfake images. Generative adversarial networks, the most common category of algorithms used to create deepfakes in the past, must be trained on highly curated data. To get a GAN to generate a face, for example, its training data should only include faces. iGPT, by contrast, simply learns enough of the structure of the visual world across millions and billions of examples to spit out images that could feasibly exist within it. While training the model is still computationally expensive, offering a natural barrier to its access, that may not be the case for long.
OpenAI did not grant an interview request, and therefore did not provide additional context for future plans with regards to its research. But in an internal policy team meeting that MIT Technology Review attended last year, its policy director Jack Clark mused about the future risks of GPT-style generation, including what would happen if it were applied to images. “Video is coming,” he said, projecting where he saw the field’s research trajectory going. “In probably five years, you’ll have conditional video generation over a five to ten second horizon. The sort of thing I’m imagining is eventually you’ll be able to put a photo of Angela Merkel as the condition, with an explosion next to her, and it will generate a likely output, which will be Angela Merkel getting killed.”
via Technology Review Feed – Tech Review Top Stories https://ift.tt/1XdUwhl
Amazon today introduced the Amazon Dash Cart, a "smart shopping cart" that uses weight sensors and cameras to let you scan your items as you place them in the cart so you can skip the checkout line. According to CNET:
"Our primary motivation for building this was to be able to save customers time," said Dilip Kumar, vice president of Amazon’s physical retail and technology. "The alternative solutions are standing in the express checkout lanes or fumbling through self-checkout stations."
Dash Carts will debut at Amazon’s Woodland Hills, California, grocery store, when the location opens later this year. The company last November unveiled plans for the Woodland Hills store as the first location for a new supermarket chain that will be separate from its Whole Foods chain. The store will include conventional checkout lanes, too.
The new shopping cart, which Kumar says is built sturdy enough to prevent those annoying shaky and off-balance wheels, is part of Amazon’s continued work to put its techie signature on the $1.2 trillion US grocery market. It introduced Amazon Go in 2016 and Amazon Go Grocery, a larger store format that includes fresh produce, in February. There’s been plenty of speculation that Amazon will add similar hardware into Whole Foods, which it purchased in 2017, but that has yet to happen.
It’s clear Amazon is trying to eliminate as much friction as possible between them and your money. I didn’t even know checkout lines were causing me to spend less, but Amazon’s data says they do so they must. It’s probably just a matter of time before Amazon skips the middleman that is your brain completely and starts randomly sending you things you might want. I’m kind of joking, but with their return policy, this actually seems like a viable business model for them.
via Geekologie – Gadgets, Gizmos, and Awesome https://geekologie.com/
Transcript: This scooter looks like a suitcase when folded. ATTO from moving life is a full-sized scooter with a unique foldable design. There is trolley mode, which resembles a suitcase, and split mode, where the scooter separates into two parts for easy lifting. You can pack ATTO in the trunk of a vehicle or take it onto a plane or train according to moving life. ATTO comes with a 48-Volt lithium-ion battery with 12 miles of range per charge. The brushless DC motor propels to a top speed of 4 mph. ATTO currently retails for $2,799 on Amazon.
Google is going to invest up to $10 billion in India over the next five to seven years, CEO Sundar Pichai announced today. The news came as part of Google’s annual Google for India event in which the search giant makes announcements specific to that market. This year’s headline is the launch of the Google for India Digitization Fund, which will distribute the $10 billion in investments of the next near-decade.
Pichai said that the fund will make money available for equity investment in local tech businesses, partnerships and infrastructure spending. They will be focused around four broad categories, including language-services (India has 22 officially-recognized local languages as well as English), helping Indian businesses embrace digitization, and tailoring products to suit India’s specific needs. The fourth involves the use of AI for “social good” in areas like “health, education and agriculture.”
Google for India has been a yearly event in the country since 2015, where the company often announces initiatives designed for the subcontinent. Last year, Google announced it would launch a new AI lab in Bengaluru as well as the launch of mobile payments there. The company has also rolled out free public WiFi at over 400 train stations (a program which has since been axed) and adding more Indian languages to its voice and search products.
We’ve seen a spate of big companies making big in-roads into the Indian market, which has been a long sought-after prize for western tech firms. Facebook invested close to $6 billion to buy a stake in Jio, the country’s biggest mobile network, Apple is opening its first store there in 2021 and even Samsung is offering digital tech support in the country. India certainly looks as if it’s going to be a big growth incubator in the future, with the World Economic Forum claiming that India could become the “next Silicon Valley.”
The investment comes at a fortuitous time for Google since India recently cracked down on technology services from China. India’s government has said it will invoke stringent rules on how stores and platforms operate (potentially to Amazon’s chagrin) in the country and recently banned TikTok and WeChat. There may also be greater conflict between Google and the local government after it began work on warrantless searches for user data in the country.
For years, the cycling industry has used 3D printing to make metal bike frames and smaller road-worthy parts. A 3D-printed carbon fibre frame, though? That’s harder to come by, even for elite riders.
Arevo wants to change that. The little-known manufacturer in California is launching a new bike brand today called Superstrata. Unlike most of its competition, which builds carbon fibre frames with multiple parts or pieces, the company has developed “a true unibody construction” that has no visible seams or welding marks. “This piece comes out as a single piece from our machines,” Sonny Vu, CEO of Arevo and the former CEO of wearables maker Misfit told Engadget. “So there’s no glue, no joins, no bolts, no laser welds, none of that stuff. If anything, things are shaved off in the post-processing [phase].”
Superstrata frames won’t be a single piece, though. The team is planning to print the main portion of the frame — which doesn’t have a traditional seat tube, creating an open diamond shape — and front forks separately. Still, it’s an impressive design feat that is catch the attention of other cyclists and road users.
Superstrata
The company will use additive manufacturing to build two similarly-shaped bicycles: a standard $2,799 ‘Terra’ model, and an electrified $3,999 ‘Ion’ version. That might sound expensive, however a carbon road bike can cost anywhere between $1,250 and $12,500. (And even higher, if you want something truly rare.) A top-end Riese & Müller e-bike, meanwhile, will set you back almost $9,000. Superstrata isn’t making a bike for the masses, but the company isn’t operating in the upper echelons of bespoke bicycle building, either.
“It’s less than your high-end carbon fiber bikes, but it’s not affordable in the common sense,” Vu accepted.
According to Superstrata, 3D printing offers a couple of crucial benefits. For starters, each bicycle can be printed to suit the exact dimensions of the rider. The company is promising “over 500,000” possible setups that account for the customer’s height, weight, arm and leg length, as well as their preferred riding position and level of frame stiffness. Most bicycle manufacturers, for comparison, offer a handful of frame sizes that account for most rider heights. And many e-bike startups, such as Cowboy, have a single specification to simplify manufacturing and repairs.
3D printing doesn’t require any expensive molds, either. That means the company isn’t tied down or incentivised to stick with the same frame design for a long period of time. The process requires fewer humans, too, than a traditional carbon fiber frame. “The materials for carbon composites are really expensive, but it’s really the labor that makes it expensive,” Vu said. “That was astounding to me. I always thought it was the material, but then I realized, ‘Oh, it’s actually the labor.’”
Finally, Superstata is promising “seamless strength” that trumps traditional monocoque carbon fiber frames. Vu told Engadget: “We say, ‘Well, your monocoque frame didn’t start as one piece. it was 20 or 30 pieces that are glued together, laser-welded maybe even together, wrapped. But it started as many pieces. And if it hits a tree, it’s going to end up being many pieces.” Superstrata’s bikes won’t be indestructible, but they should fare better if you take a tumble or stand it up haphazardly outside a cafe.
Superstrata
The Ion will ship with a 252Wh battery tucked into its svelte down tube. By e-bike standards, that’s small. VanMoof and Cowboy’s latest wares offer 504Wh and 360Wh batteries respectively, for instance. Superstrata is promising a 250-watt motor, which is effectively the industry standard for pedal-assisted e-bikes, that can rise to 350 watts in a pinch. It will deliver 40Nm of torque, which is well below the 75Nm that Bosch — the supplier for countless bike brands including Trek, Moustache and Riese & Müller — currently offers with its flagship generation four motors. (Bosh is also planning a software update that will increase the torque to 85Nm later this summer.)
Superstrata’s Ion will be lighter than most of its e-bike competition, though. The entire contraption should weigh 11KG, which is 1KG less than Gogoro’s Eeyo 1, 8KG less than VanMoof’s S3 and only 700 grams heavier than the folding Hummingbird. Like the Eeyo 1, the weight and “open-frame” design should make it easy to carry the bike on one shoulder. In theory, the the weight should counterbalance the smaller battery size, too. Superstrata is promising 96KM (60 miles) of range on a single charge and an assisted top speed of 32KMH (20MPH), which will be software restricted to meet Europe’s lower 25KMH (15.5MPH) speed limit.
The company has sourced some of its basic components — the tyres, saddle and groupset, for instance — from third-party manufacturers. We don’t know the exact model numbers, but Vu confirmed they would be fairly “standard” choices. Superstrata couldn’t pick top of the line components, he explained, because the team is ordering them in smaller volumes than traditional bike manufacturers. “The average person, I think they’ll enjoy it,” Vu said. “Pro bikers, they’ll probably scoff at it, the sets. But they can put in their own sets.”
Arevo has been working on its 3D-printed bike formula for some time. Back in 2018, the company unveiled a head-turning proof-of-concept with a blue frame that extended beyond the head tube fork and seat tube. Twelve months later, it unveiled the electric Emery One with Franco Bicycles, a premium road and gravel bike manufacturer based in California. Arevo has since announced that it will help Pilot Distribution Group, a company in the Netherlands, develop a new line off e-bikes that have 3D-printed carbon fiber frames.
With Superstrata, the company is finally striking out on its own. “I said, ‘You know what? Why are we doing this for other people? Let’s do it for ourselves and if other people want it, they can come to us,’” Vu said. “So that’s the idea. Let’s make a nice bike, end to end. Let’s build a whole brand around it and just have fun.”
To deliver its dream bicycles, though, the company is turning to Indiegogo. As with all crowdfunding campaigns, there’s no guarantee that the Terra or Ion will ever materialize. Arevo has been around since 2013, though, which is longer than many electric bike startups. Vu also hinted that the company is using the platform more as a marketing “launch pad” than a finance-raising tool. Still, any reservation is a gamble.
If everything goes to plan, Arevo will be delivering its first bikes in December. The company will make 500 two-wheelers in its initial run, according to Vu, through a combination of printing in the US and Vietnam. He hopes the Superstrata business will increase to “thousands, eventually,” but is keenly aware of how competitive the bike industry can be. If the brand doesn’t catch on, Arevo might have to pivot.
“As a 3D printing company, we can do that,” he explained. “We can say, ‘Nah, okay. We’re done.’ If you’ve invested into a whole line with carbon fiber molds and a complete production setup, you better sell tens of thousands of bikes or else you’re pretty screwed. But if 437 people are hating [Superstrata bikes] and we can’t get anyone’s attention, it’s okay. Let’s do baby strollers or whatever. Let’s do something else. I just love this world of additive manufacturing because you can do that kind of thing. It’s really upending the supply chain structure that’s out there.”
Transcript: A car parking robot. Stan is an autonomous bot that will find a spot and park your car for you. 100% electric, Stan uses storage optimization software and autonomous car organization to take your vehicle and park it with other cars, optimizing space in parking garages and eliminating the hassle of having to do it yourself. When you initiate the valet service, Stan comes to tag and pick up your vehicle, then neatly parks it. Sensors scan the environment and Stan adjusts as needed.
Want to make sure your parking is always on point? Park Right from Maxsa Innovations makes sure you land that perfect spot every time so you don’t cause accidental damage in your garage.
