IoT news of the week for March 22, 2019

Facial recognition needs some form of control: One of my biggest fears with the collection and analysis of data enabled by the internet of things is that it will be used in ways that are harmful to the consumer but at the same time are totally opaque to said consumer. For example, corporations might use postings on social networks to generate credit scores, thus raising the cost of a person’s borrowing by mechanisms that consumer can’t fathom or, by extension, control. In retail, the ubiquity of connected cameras and pretty good facial recognition algorithms has turned my fear into reality. This article discusses how retailers and others are now able to use facial recognition software in local stores to identify shoplifters and then share that data across a nationwide network of stores. Imagine your worst day and then imagine if whatever you did or whatever happened to you on that day was recorded forever and you were judged against it. That’s essentially what ubiquitous facial recognition (and frankly what most data collection) can now do. The question is, if we don’t think it’s fair, how do we regulate it? (CNET)

IoT will kick off a new era of policing: In the second example of computer-assisted policing gone awry, Slate has a story about a program called Graffiti Tracker, which a few cities have implemented to track graffiti artists’ tags and then attribute the individual acts of vandalism back to each of them. It used to be that if someone was caught in the act, their punishment was related to that one specific incident. But in cities using this software, police can attribute every piece of graffiti to the person they have caught. In one case documented in the story, a teen was caught and charged with tagging 218 different walls. Instead of the one crime, he went to jail for all of them. That may not seem unreasonable to everyone, but ask yourself how you would feel if a cop who pulled you over for speeding was able to write you a ticket for every single instance of speeding you had ever committed. With sensors embedded in roads and cameras on every light post, that would be a possibility. Technology is about to give us the ability to monitor everyone all the time. As such, we need to think hard about how our laws — and the penalties for breaking them — are structured.  (Slate)

Oh noes! Google is killing some IFTTT integrations: As someone who has an IFTTT applet set up to turn a Hue light bulb red upon receipt of an email or text from my editor, I received the news that Google was changing its Gmail IFTTT support with dismay. This news came out in October, but I only learned of it this week when IFTTT send out an email. Basically, unless a developer is using a new Add-On program for APIs, some prior API integrations won’t work. In this case, it means that the only things you can now do with IFTTT is send yourself an email or send one to someone else. Google changed its program to promote better security, but I’m not sure how IFTTT can respond or whether it will choose to invest in the new program. For more on the coming wave of breakages and frustration associated with APIs, check out this story I wrote last year. (Hacker News)

A fight over mobility data access in LA is a good test case: The city of Los Angeles wants to keep track of where scooters are on its city streets. But Uber doesn’t want to share real-time location data on its scooters, claiming that the data is private. While I trust Uber’s willingness to look after my privacy about as much as I’d trust a cat to babysit a canary, the fight does illustrate a real tension with today’s access to real-time data. For example, I have no problem with a city’s department of transportation asking for data to make sure scooter providers follow laws and are available to low-income residents, but I wouldn’t want that data to hop the fence to the Los Angeles Police Department. Privacy experts would like to see safeguards around the data as well. In an ideal world, anytime a city collects data like this it should outline exactly how and where it can and will use it. (Motherboard)

AWS CEO Andy Jassy shared his thoughts on edge computing: Unsurprisingly, Andy Jassy, who leads AWS’ businesses, believes that the next-generation hybrid cloud infrastructure will involve the cloud and billions of low-powered devices gathering data and then sending that data up to the cloud. Amazon has clearly built out AWS services such as Greengrass (which easily links dumb edge devices back to the AWS cloud) to meet this need. However, I’m not sure I entirely agree with him. There are several use cases that will rely on powerful edge computing on premise, with limited data being sent back up the cloud. In that model, some gateway computing will process data at the edge and send decisions back to sensors and actuators. The cloud will be reserved for storage, non-latency-sensitive analytics, and training machine learning models. That’s still a huge chunk of work, but I think the margin growth will be at the edge gateways. (CRN)

You can’t turn redundancy and safety into an upgrade: More drama from the Boeing crashes that makes me question how we can trust automation without proper regulatory oversight.  (NYT)

Falkonry brings ML to the edge: An update to Falkonry’s analytics software lets customers run it at the edge and have those nodes predict trends or anomalies without needing to send the data to the cloud. This is essential for low-latency environments. Falkonry executives also say the software and models can be implemented without a data scientist. It’s part of a no-code approach to data science that most industrial platforms are pushing. For example, last week C3 made a similar claim. (CRN)

Mirai is back: This time it is infecting more brands of smart TVs and aimed at businesses. Remember, if you have an infected smart device that can’t be or isn’t patched, your best bet is to physically disconnect it from the network. (Ars Technica)

This is a good technical dive in NXP’s latest embedded processors: After a wave of consolidation in the embedded chip world NXP is certainly the biggest provider of embedded silicon, and it’s constantly pushing the envelope on security and size. The latest i.MX 8M Nano chip is small and features ARM’s new PSA security architecture. I think I’ll do a deep dive on various security architectures in the next week or two. Anyhow, if you’re selecting silicon, give this a read. (Forbes)