A different way to bring AI to the edge: This article is trying to explain three difficult concepts in one go, starting with how reservoir computing works, moving on to how neural networks work, and then assuming that you understand how micro-electromechanical system (MEMs) work. In other words, there’s a lot going on. The short version — which doesn’t require you to know how this works (although it is really fascinating) — is that researchers in Canada have built a device that is smaller than the width of a human hair and could process a single type of physical input, such as motion or sound, on the chip itself. These MEMs devices would be able to perform such highly specific tasks in an energy-efficient manner, allowing them to be powered by energy harvesting. (IEEE Spectrum) -S.H.
Here’s Sidewalk Labs’ plan for data privacy in smart cities: I’ve written about the challenges of building smart cities, and Toronto’s Quayside water development project, which is done in conjunction with Sidewalk Labs, has exemplified some of those challenges. This week, the company proposed its plans for the area and shared its thinking on data privacy. The main idea is that all urban data should be placed in a Civic Data Trust and be made accessible to all. The other big idea is that any entity that wants to collect or use the data will have to file a Responsible Data Impact Assessment. That assessment will be publicly available and reviewable to anyone. These are reasonable steps, but they do have two implications. The first is that they put the municipality on the same footing as corporations when it comes to accessing the urban data. This gives me pause, because public safety is more important than marketing, and any data I might share with my police department for public safety purposes isn’t something I’d necessarily want to share with a private company. The second implication is that being able to review these assessment plans will require activists, journalists, and the like to bone up on their statistics and data analysis in order to understand how the data could be used and the potential impacts. (Medium) -S.H.
IoT bad citizen alert: One of the challenges of IoT security is that one bad apple can pollute a whole lot of the internet. Mixed metaphor aside, I am in love with this concept from Brian Krebs where he calls out a company that’s producing connected devices with terrible security practices, such as hard-coded passwords and a lack of updates. Here, he’s shaming Xiongmai, a maker of cheap IP video cameras. The company was called out in the wake of the Mirai botnet and promised to issue a recall. It hasn’t. Instead, it has threatened to sue security researchers and journalists. To learn more, read the article, as it lays out a strong case. (Krebs on Security) -S.H
Prepare for the rush of quantum security startups: This week, a startup called Crypto Quantique launched, claiming that its silicon couldn’t be hacked, even by quantum computers. This article clearly wants to take its claim with a grain of salt, but the writer may lack the technical depth. I’m not saying that to be a jerk, but because I, too, have waded into the crypto-quantum security mire before and returned with a bunch of equations that might as well have been written in Aramaic. We’ll see more of these companies emerge in the next few years because quantum computing is finally looking plausible. The concern is that with quantum computing, the current means of encryption that computing uses will become worthless. That’s why companies such as Crypto Quantique are launching and companies like Secure RF are touting their quantum impenetrability. Even the National Institute of Standards (NIST) is trying to solve this problem. I’m talking to the Crypto Quantique folks next week, so I’ll offer my two cents on the tech soon, but I’m worried I may just have to wait for NIST or someone much smarter to weigh in. (TechCrunch) -S.H.
Facial recognition and blockchain for…animals? Yup, it sounds weird, but it’s true. This article provides a dozen examples, ranging from the use of Google Cloud Auto-ML Vision on elephants to protect them from poachers, to using facial recognition on sheep to ascertain how they physically feel. A Chinese company is using similar tech with pigs to help track their age, weight, and diet. Perhaps the most unexpected example is “GoGo Chicken.” Using blockchain, you can learn where the chicken you bought for dinner was born, what it ate and — look out, Fitbit — even how many steps it walked during its lifetime. (NewYork Magazine) – K.C.T.
Alexa, whisper sweet nothings in my ear: Amazon announced Whisper Mode for Alexa last month, and the feature is now rolling out in the U.S. If you missed the news, in Whisper Mode Alexa will whisper back to you if you quietly say a voice command. This is ideal for nighttime use or, in my case, when the dog is napping (Alexa always wakes Norm). To enable the feature, you can say “Alexa, turn on whisper mode” to your Echo device. Alternatively, there’s a setting in the Alexa app, too. If you don’t see it yet, don’t yell. You’re sure to see it soon. (VentureBeat) – K.C.T.
With Smartfrog, Canary could leap into the EU: Since I bought it last year, there hasn’t been much new functionality added to the Canary camera security system in my home. It works fantastic as-is, though. And instead of Canary adding new features, the company is adding more money thanks to Smartfrog, an EU-based IoT company that sells its own webcams and services that turn smartphones into remote security cameras. Smartfrog invested $25 million in Canary and is also gaining a controlling interest in it, although that’s separate from the amount of the investment. Why does this make sense? Here in the U.S., Smartfrog isn’t well known, and you could say the same about Canary in the EU. Perhaps the two will cross the pond and expand their respective reputations. (TechCrunch) – K.C.T.
Google Home integrations improve on Nvidia Shield: I could swear that some of these Google voice commands already work with my Nvidia Shield TV set-top box, but I do see some new ones. Now you can get a little more granular by telling Google to tune in to a particular channel or open a specific app on the TV. Perhaps this means those $50 Nvidia microphones designed to speak to a Google Assistant, the ones the company announced nearly two years ago, will never see the light of day. And why should they, when a Google Mini can often be found on sale for less? (Nvidia) – K.C.T