Clean

Tech News: Power-Hungry AI and Malicious Toothbrushes

Published Feb 9, 2024, 10:45 PM

Sam Altman of OpenAI wants to see a huge investment in new semiconductor fabrication plants to support AI implementations. The FTC hopes to reverse the Microsoft-Activision Blizzard merger. And did hackers really take over millions of smart toothbrushes?

Welcome to Tech Stuff, a production from iHeartRadio. Hey there, and welcome to tech Stuff. I'm your host, Jonathan Strickland. I'm an executive producer with iHeart Podcasts and How the tech are you. It's time for the tech news for the week ending on February ninth, twenty twenty four So. The Guardian reports that a Chinese backed hacker group going by the name volt Typhoon has spent the last five years at the very least, infiltrating various critical computer systems here in the United States in our infrastructure, and they have compromised systems connected to transportation, shipping, and utilities like water and sewage. The US National Security Agency says that this hacker group has created footholds within these systems, so they didn't just infiltrate and spy on stuff and get out. They made sort of a cozy little nest for themselves that they could easily return to. Now to me, this sounds awfully familiar to stories that I've been hearing for many years now. In fact, I even did a quick search on Google and I just picked the year twenty fourteen at random. It was actually my first choice. I mean, it's ten years ago. I thought, let's see if there's been any stories in twenty fourteen about Chinese hackers compromising critical systems within US infrastructure, And sure enough, the top result was from CNN, and there was an article that had the headline, the US government thinks China could take down the power grid because, just as we're hearing now, Chinese hackers had managed to infiltrate these various computer systems and to kind of leave for themselves the capacity to return to it, so that maybe in the future, in an act of cyber warfare, they could sabotage the system. So, on one hand, this news story is alarming, but it's not actually new, right because we've had issues with Chinese backed hackers compromising systems here in the United States for ages. But on the other hand, it's still a very alarming story to read about. I would argue it might be more alarming to me that we freaking knew about this problem a decade ago and seem to have learned nothing since then. That to me is really alarming. It's not just that Chinese hackers did this, it's that we saw essentially the same story play out a decade ago, and yet haven't done the proper steps to protect ourselves against future attacks. So the Guardian points out that the big concerning element here, besides the fact that Chinese hackers have infiltrated all these different computer systems, is that it does not look like it's an act of espionage at all. This isn't a case where Chinese hackers are trying to get secrets and then use those back in China. This appears to be a preparation for sabotage and cyber warfare. So really this is preparations for an attack, should that ever come to pass. Now, it may be that an attack never would come to pass, but it's more like the Chinese hackers were setting things up in the event that there was a cyber attack to be conducted by China against the United States, they would have already laid the groundwork. So yeah, it's one thing to steal secrets, but it's another thing to shut stuff off on a massive scale. I mean, you can think of how disruptive that would be. So presumably the affected targets are going to seek ways to expel the hackers and harden security against future intrusions. But then you would have figured that would have been the case a decade ago, and yet it's still happening. I guess you can argue that no security system is perfect, and that there are always going to be vulnerabilities, whether that's in the code or more likely due to someone failing to practice could security measures, and that you'll never have a perfect system. So you could argue that, but at the same time, when you find out that it's this apparently widespread, it raises some very serious questions. Anyway, now let's move on to our traditional glut of AI related news. So first up, the Verge reports that Google has dropped the names Barred and Duet from its various AI products. So the new AI Google strategy is called Gemini, or if you're an astronaut from the nineteen sixties, maybe Geminy. But Google is streamlining its approach. So previously, it's aipowered chatbot had the name Barred, which was a nod to William Shakespeare, a nod that I appreciated, and it had some AI enhanced components in the Google Workspace product that had the name Duet. But now all of that is under the name Gemini. So on top of that, Google is going to release a new and larger than ever large language model to the public soon, the Gemini language model. Android users will get a chance to test out Gemini and all its glory. It looks like a lot of different apps are going to get a Gemini upgrade. iOS will be a little bit different, but if you're an Android owner, you'll have access to this pretty soon. And it looks like Gemini may quietly take the place of the humble Google Assistant over time. It might be a long, phasing kind of process, but that's what looks like. So will Jim and I be able to challenge open a eyes domination with chat, GPT and the GPT language model. Will I even notice that it's happened, or by that time will I be hiding out in an off the grid shack in the woods. Find out next week at the same BAT time, same BAT channel. Al Jazeera reports that Sam Altman, the CEO of open Ai, is looking for cash, like a whole buttload of it. So how how over much you're thinking? It's more than that? So, according to the news site, Altman's goal is to raise quote trillions of dollars from investors, including the United Arab Emirates government to boost the world's capacity to produce advanced chips and power artificial intelligence end quote. And further in the article they specify that they're talking around seven trillion dollars good gravy. That's a lot of money that goes beyond a princely sum. Princess don't have access to seven trillion dollars. Now, we know that AI requires a lot of computational power to work, and open AI's business is completely dependent upon customers demanding more AI capabilities. So in order to deliver those capabilities, you've got to have the hardware that can run the AI software on top of it. So it makes sense that open ai needs all these these investments to go into making the hardware that powers everything. So this money would ultimately go toward building more fabrication facilities, and those facilities would be run by established semiconductor fabrication companies, not open Ai. So it's not that open ai would get into the semiconductor business. They're saying, we need this money raised so that we can build these sorts of fabrication plants all around the world and have these established companies handle them so that we have the equipment we need in order to deliver the AI experiences that we want to, so the argument is that we're not funneling the money to open AI itself. I'm not exactly surprised by all this news, but it does make me wonder if we're going to actually see AI replace cryptocurrency mining as a computational application that just demands an insane amount of power. Right? Is it possible that in a year or two we'll be talking about how AI as an industry requires more power than most countries do for a full year the same way we talk about cryptocurrency. We'll actually circle back to talk about cryptocurrency and power demands a little bit later in this episode. But yeah, I just wonder about that. This week, Clint Watts, the general manager of Microsoft's Threat Analysis Center, published a blog post that made me take notice. He was focusing on how Iran's strategy regarding Israel is leaning more and more on cyber operations, and one of those cyber ops apparently involved using an AI generated news anchor to deliver fake news to audiences in different parts of the world. So apparently this happened late last year in December, and an Iran aligned hacker group called Cotton Sandstorm interrupted some streaming TV services in different countries and inserted this AI generated news program with an AI generated news anchor and that quote. The disruption reached audiences in the UAE, UK and Canada end quote. The Guardian further reported that the AI anchor presented quote unfit verified images that claim to show Palestinians injured and killed from Israeli military operations in Gaza. Uote Watts express concern about how this instance is an indicator of what we should expect moving forward, particularly in really eventful years like an election year, which hey, here in the United States we happen to be in an election year. So keep an eye out for those robots on your screens and on your phones and really everywhere. Why that's shack in the middle of the woods is looking better and better. All right, you know what, We're going to take a quick break to thank our sponsors. I'm going to gather myself and try to get rid of this sort of luddite tendency that's overtaking me, and we'll be back right after this. We're back. So, the US Centers for Medicare and Medicaid Services sent out a memo to all insurers that offer Medicare Advantage with a very clear message those insurers are not allowed to use AI to determine if someone on a Medicare advantage plan merits coverage or should be denied coverage. This is really good news because we've already seen some insurance companies do that very thing, use AI to determine if they should grant or deny coverage to a patient. In fact, right now, there are a couple of massive lawsuits against Humana and United Health about that very thing. Patients claim that these companies used an AI tool to decide whether or not those patients should receive coverage for various medical processes, procedures, and prescriptions, typically things like whether or not they should be allowed to stay in a medical facility or if they should be told to leave, even if it's before the doctors said that they should leave because otherwise insurance was going to cut off the support and they'd have to cover everything out of pocket. So, according to the claim, many of those decisions that were made for these patients via AI were wrong. The AI reportedly had a dismal accuracy rating, but the insurance companies were still depending upon it, and the argument is that the companies were following an incentive to use this AI tool because by denying claims, they didn't have to pay out to the insured. So yeah, the tool's broken. But the tool also says I get to keep more of my money if I listen to it, So I'm listening to the tool. In other words, Anyway, now, the CMS is telling insurers that while they can use AI assisted tools to do something like predictive person's length of stay in a medical facility, any decisions on actual coverage have to be done solely by taking that individual patient's circumstances into account, which means you can't just use aggregated data from a larger population and then make a determination for that patient, which is good news for people who are part of these programs. To learn more about this, I recommend Beth Mole's article in Ours Technica. It's titled AI cannot be used to deny health coverage FEDS clarify to insurers. It's a good read. So a bit earlier, I mentioned that AI is placing an increasingly large demand on resources, similar to what we're seeing in crypto mining. Well, according to Tom's Hardware, which is a great website by the way, the US Energy Information Administration released an analysis that says one hundred and thirty seven crypto mining operations in the US demand two point three percent of the US's power demands overall, which is astounding, Right, You're talking about fewer than one hundred and fifty organizations that require that much of the nation's power that we generate. We've heard in the past about how crypto mining, specifically proof of work systems like Bitcoin, have created this sort of runaway demand on power, but boy howdy, I mean, this is just a lot anyway. The implication here is that the US government is starting to get a little concerned about how much power these organizations demand, and that maybe we should expect US agencies to take a bit of a closer look at crypto mining, perhaps with regard to stuff like I don't know, environmental impact, and this in turn could eventually have a big effect on the crypto space, at least here in the United States. Like it's possible that we could see some regulations and restrictions that really hamper crypto mining efforts in the US, which would just open them up in other nations. Obviously. The article in Tom's Hardware includes a map that shows where the administration says these organizations are located. I was surprised to see that a whole bunch of them are in my home state of Georgia. Fun times. Microsoft and Activision Blizzard completed their acquisition deal last year where the two companies merged, But that doesn't mean the US Federal Trade Commission is real happy about it. You know. If you might remember, the FTC opposed that merger, and their concerns were largely dismissed in various courts. But the FTC is still interested in perhaps reversing that deal, and Microsoft is not actually making it easy on themselves to avoid that because one thing Microsoft agreed to when it was trying to convince the US government that it should be allowed to acquire Activision Blizzard was that the company wasn't going to hold massive layoffs in the wake of these two companies coming together, because the intention was Activision Blizzard would still operate kind of as an independent entity, it would not be fully integrated into Microsoft. But after the merger happened, Microsoft laid off nearly two thousand employees I think oney nine hundred or so, and we don't know exactly how many of those were Activision Blizzard employees, but it could have been as many as half of them, and the FTC says that this could be a violation of the agreement Microsoft presented when they were first trying to acquire Activision Blizzard. Furthermore, Phil Spencer, the CEO of the Gaming division at Microsoft, said that the companies have quote set priorities, identified areas of overlap, and ensured that we're all aligned on the best opportunities for growth end quote. But then the FTC says, well, why are you worried about overlap? Because again, you had said that the intention was Activision Blizzard and Microsoft Gaming would more or less operate independently, that they would be their own things. So overlap shouldn't be an issue because you didn't. You said you weren't going to combine these two, and yet now it sounds like you are combining them and creating a more integrated division between the two. In fact, you had said that you were going to keep them separate so that you could potentially divest some or all of Activision Blizzard if you wanted to. But now it sounds like it's all getting mixed up together. So what's the deal here? So the FTC is now asking courts to force Microsoft and an activision Blizzard to temporarily pause the merger process, like put a pause on any more restructuring or anything like that, so that the FTC can actually litigate the merger with the potential ultimate goal of undoing the deal. So could we see one of the biggest mergers in gaming hit the rewind button. It's possible. It is not likely, but it is possible. Sony has decided to sunset the anime streaming service Fundimation, merging it with crunchy Roll, which Sony acquired in twenty twenty one. However, this also means that some properties aren't necessarily going to make the jump, So Sony's going to wipe those films in series off the digital libraries. But worse than that, which it's already bad, but worse than that. Fundamation offered customers the chance to buy physical media versions of some of the anime properties, and they would include a digital code so that the customer could access a streaming version of that through their Fundamation account. So they had the physical media, but they also had a code for digital streaming version. But this part of their service is not going to make the transition to crunchy role. So those digital codes aren't going to work anymore for folks who purchased the physical media but who subsequently lost access to that physical media for whatever reason, Like maybe they bought it but then they realized they didn't have the space to hold this kind of stuff, so maybe they traded in the physical versions, but they kept the digital code. Well, now their digital copy is going to be inaccessible, and this stings quite a bit. Fundamation at one point claimed that customers would have access to these digital streaming versions quote unquote forever. So I guess forever ain't what it used to be. But it turns out that the terms of service indicated that customers never actually owned the digital streaming copy. This is pretty typical for streaming services. You don't own something. Even if you buy the title, like on Amazon Prime or something, you don't actually own that copy. You actually own access to that copy, and that access at some time can be revoked. It could go bye bye. This is one of the many reasons I have found myself going back to buying physical media for some shows and films, because at least then I will still have access to that media, assuming I still have a machine that can play whatever the format is. Finally, the last story I want to cover is one that has gone back and forth and up and down, kind of like brushing your teeth. And hey, it involves electric toothbrushes, all right, so this story has some weird elements to it. A Swiss German newspaper called argaler Zeitung and I know I butchered the pronunciation. It published an article that apparently claimed that hackers had managed to compromise around three million smart toothbrushes. And you might be thinking, what the heck do you do with a compromise smart toothbrush? Make it brush someone's teeth badly? But no, you can use those compromised toothbrushes as a kind of botnet to send a distributed denial of service attack to a target. So each of those toothbrushes can start pinging a target server in an effort to overwhelm it and shut it down. And because the security on these Internet connected toothbrushes was really poor, it was not hard to accumulate a big old army of them. This is one of the major concerns about the Internet of things. You know, sometimes those Internet connected sensors can really be a threat if the company that makes them doesn't implement good security measures. Anyway, some outlets reported on this, and then the American branch of the cybersecurity company called Fordinet ended up saying, oh, the original article was quoting a Swiss branch of Fordinet, but there was a problem with translation, and that in actuality, what the Swiss cybersecurity experts were saying was this is a hypothetical example, not something that actually happened, so you should really correct the article. So then several outlets reported, oh, wasn't it funny This mistranslation made people think that these electric toothbrushes were being used in attacks. But then the newspaper our Gower Zitong said no, no, no, no, no, this really did happen because we asked about it. The Fortnet branch in Switzerland indicated this was an actual case, not just a hypothetical example, and when we submitted our article to them for their review, they approved it. So clearly this isn't just a mistranslation. So our toothbrush is out to get you, I don't know, but you should still brush your teeth in any case, just you know, maybe go with a toothbrush that doesn't connect to your home network, just to be on the safe side. All right, that's it for the news for the week inning on February ninth, twenty twenty four. I hope you are all well and I'll talk to you again really soon. Text is an iHeartRadio production. For more podcasts from iHeartRadio, visit the iHeartRadio app, Apple Podcasts, or wherever you listen to your favorite shows.

In 1 playlist(s)

  1. TechStuff

    2,451 clip(s)

TechStuff

TechStuff is getting a system update. Everything you love about TechStuff now twice the bandwidth wi 
Social links
Follow podcast
Recent clips
Browse 2,448 clip(s)