Subscribe
Share
Share
Embed
Chaos Lever examines emerging trends and new technology for the enterprise and beyond. Hosts Ned Bellavance and Chris Hayner examine the tech landscape through a skeptical lens based on over 40 combined years in the industry. Are we all doomed? Yes. Will the apocalypse be streamed on TikTok? Probably. Does Joni still love Chachi? Decidedly not.
[00:00:03.03] - Chris
So quick question. Have you ever noticed that young people don't know anything?
[00:00:12.16] - Ned
Uh, as a former young person, how dare you? And as a now old person, yes.
[00:00:18.03] - Chris
No, you completely misunderstand me. When we were young, we had the whole world understood and we were the future. Now that we're not young, the young people don't know anything. And they have no experience and we shouldn't listen to a word they say.
[00:00:29.15] - Ned
That sounds accurate. Yeah. I don't see any flaws in this reasoning.
[00:00:35.02] - Chris
I said something in a meeting and basically the phrase damning with faint praise came up.
[00:00:41.23] - Ned
Okay.
[00:00:43.21] - Chris
And the young person who shall remain young and nameless, Young Lee, basically was like, so you think the guy's going to hell? Oh, what are you supposed to do with that as a person?
[00:00:56.23] - Ned
Well, I guess they just didn't study the classics.
[00:01:00.03] - Chris
Or anything except, I don't know, TikTok. What do young people even do?
[00:01:06.02] - Ned
Well, it seems to me that you should just make a TikTok that brings back old-timey phrases like damning with faint praise.
[00:01:13.22] - Chris
I mean, you're making a joke, but I'm like, hmm.
[00:01:16.19] - Ned
No, I'm— it started as a joke, but Now that I think about it, I'm like, no, that would probably do really well.
[00:01:25.01] - Chris
How are your TikTok dances these days?
[00:01:28.11] - Ned
Elaborate.
[00:01:30.03] - Chris
Are they on fleek?
[00:01:32.19] - Ned
Stop trying.
[00:01:33.06] - Chris
Cap or no cap? I'm doing great, right?
[00:01:37.23] - Ned
And welcome to the Chaos Lever podcast. My name is Ned, and I'm definitely not a robot. I'm a real human person with neurons, A cerebellum and a dual intake valve. With me is Chris, who is also here. How's your dual intake valve, Chris?
[00:01:56.05] - Chris
If I were to ask you to define dual intake valve—
[00:02:02.01] - Ned
A valve that is able to intake 2 things.
[00:02:05.14] - Chris
It intakes dually.
[00:02:07.08] - Ned
Yes.
[00:02:08.06] - Chris
None of the young people would have been able to figure that out. That's all I'm saying.
[00:02:14.06] - Ned
Out.
[00:02:14.06] - Chris
Facebook's Muse AI already has a serious 0-day vulnerability. From the, oh my goodness, who could possibly have seen this coming department, Mark Zuckerberg's latest billion-dollar boondoggle, the Muse AI assistant, which was rushed out the door and probably vibe-coded by Facebook's increasingly abused, disinterested, cynical, and shrinking team of developers, Has a major security vulnerability. Muse, as you'll remember from last week, is expected to basically be your digital assistant. This means that Muse would be most effective if it has access to everything in your life— email, bank accounts, calendars, etc. How could this possibly go poorly? I wondered quietly to myself and loudly last week. Well, in the words of a great philosopher, God, I hate being right all the time. This week saw the announcement of a serious and easy-to-deploy 0-day vulnerability for Muse. According to security researcher Patrick Wardle, the attack, using a well-known social engineering technique called click fix, allows an attacker to do, quote, whatever we want, unquote, within Muse. He continues, quote, instead of us having to write a very comprehensive Mac malware stealer, we can just leverage the AI assistant itself.
[00:03:40.02] - Ned
It's here to help.
[00:03:42.00] - Chris
The bug let attackers hijack where Muse sends your voice for transcription, steal the login token that controls your account, and then let them append commands to Muse to do anything the attackers want. Wow. It's keen.
[00:03:55.19] - Ned
Hmm.
[00:03:57.14] - Chris
The attack, much like AI assistants themselves, is stupid. Wait, stupidly simple. The attacking server acts as a proxy between the user and Muse. The user enters a prompt and the attacker just kind of, you know, adds to that. Muse is none the wiser because Muse is stupid and was poorly written in a great and unnecessary hurry. So it just does what the attacker's prompt tells it to do on the user's devices, which the user has given almost unlimited permissions to.
[00:04:30.19] - Ned
Mm-hmm.
[00:04:32.15] - Chris
Now, to be fair, Facebook did patch this one pretty quickly. Stands to reason, right? An easy attack had an easy fix. But what about the ones that aren't easy to fix? Patrick Wardle again, quote, when you take a look at Muse, It's like they didn't, in my opinion, think about security, which is really worrisome. At the very least, they should be thinking about security from the very start, and they are just not. The lessons to take from this are simple. 1, never trust Mark Zuckerberg to do the right thing in any situation ever. 2, never trust Facebook with any valuable information. 3, Never give an autonomous system access to your stuff. Just cancel your unused subscriptions by hand. It is annoying. I'll grant you that. But it's not as annoying as a hacker stealing your life while Marky Zuck sits back and just shrugs.
[00:05:32.06] - Ned
Anthropic goes 1-1 with the DOD. A DC federal appeals court has ruled against Anthropic in the ongoing litigation regarding Anthropic's refusal to allow the Department of Defense to use Claude in such a way that violated the restrictions laid out in a contract and their subsequent blacklisting by the U.S. government and labeled— labeling as a supply chain risk. I know that was a lot of words, so let me lay it out a little more clearly. If you don't remember, basically Anthropic had put some restrictions on what Claude was able to assist the DOD with. Specifically, Anthropic refused to have Claude be used for development of surveillance programs or autonomous lethal weapons, which seems like pretty good. Attaboy, Anthropic. You did one good thing. Unfortunately, little boy Pete Hegseth threw a temper tantrum, decried Anthropic as woke AI, and pushed to have them blacklisted by all government agencies. Due to them being a supply chain risk. Blacklisting Anthropic would not only block the use of their products by government personnel, but also have a cascading effect on contractors and vendors that work with the federal government. Anthropic filed suit in the District of North California to block the order and add an injunction.
[00:07:01.01] - Ned
The injunction was added blocking this blacklisting, and the Northern California court ruled in favor of Anthropic. Citing First Amendment protections and a lack of evidence that Anthropic truly represented a supply chain risk. The government appealed, and now in a 2-to-1 decision by the D.C. Appeals Court, the decision has been kind of reversed. The court took a broader reading of what constitutes a supply chain risk within the context of the DOD, citing the fact that Anthropic could, quote, deny, end quote, functionality that would impede military operations as sufficient for the supply chain risk label. So where does this leave Anthropic? The decision in the District of North California applied broadly to the entire government and placed a permanent injunction on the Anthropic ban. The appeals court decision was more narrowly focused on just the DOD And Anthropic cannot be directly used by the DOD. However, the ban does not apply to contractors, which the DOD uses a lot of, a lot of contractors. The next step in this process would be an appeal to the 9th Circuit or Supreme Court, which is not exactly the friendliest court in the land given its conservative majority and extreme deference to the executive branch.
[00:08:28.01] - Ned
Either way, the Claude restrictions are still in place, so the DOD will need to rely on Groq to make its autonomous death machines. Yay.
[00:08:39.13] - Chris
So when you first read that, I thought you said Anthropic goes one-on-one with the Department of Defense, which in my head immediately went to Dario playing Pete Hegseth one-on-one.
[00:08:53.19] - Ned
Oh my God.
[00:08:55.09] - Chris
And honestly, that would probably be a faster way to get these decisions made. It would make just as much sense.
[00:09:02.12] - Ned
I like— as much as I dislike Pete Hegseth, I'm pretty sure he would dominate Dario in a one-on-one basketball game.
[00:09:09.23] - Chris
Oh, you are dreaming. That alcoholic man-child would last 30 seconds before he needed a break. Data center operations in the United States are bad. New Jersey edition. Let the hyperscalers do whatever they want, they said. They'll be responsible with what they build, they said. New data centers means new jobs, they said. Just go ahead and ignore the skyrocketing electrical costs, the insane noise pollution, and oh, here's a new one, actual pollution. And the only town in America so far willing to put a fight up against this bullshittery? Vineland, New Jersey, apparently.
[00:09:49.01] - Ned
Let's go Vineland.
[00:09:50.19] - Chris
So the story comes with the 300-megawatt data center in question, not even completely built yet. The data center is being developed by DataOne on behalf of the Nebius Group.
[00:10:04.20] - Ned
Nebius.
[00:10:06.12] - Chris
Nebius Group?
[00:10:07.11] - Ned
Sure.
[00:10:08.22] - Chris
Nebius Group. Mobius Group.
[00:10:10.22] - Ned
I've heard it always.
[00:10:12.14] - Chris
Morbius Group? On behalf of Microsoft, ultimately for, you guessed it, their artificial intelligence initiatives. Data center construction in 2026 is not without controversy, as evidenced by literally every single data center project in existence. But in this case, the issue is pretty straightforward. In December of 2025, the site was inspected and deemed okay. Outrageous, loud, and actively harmful to taxpayers. Sure. But otherwise, okay. Those same inspectors came back in July and saw natural gas power generating equipment, like 62 installations of them, all installed without proper permitting and without ever even bothering to follow up and file proper permitting.
[00:11:04.12] - Ned
Sounds right.
[00:11:06.06] - Chris
This, according to the jerks— I mean, classy and professional executives in charge— was merely a temporary measure until the permanent fuel cells it planned were finally operational. Now, why not wait until those fuel cells were in place, you might ask?
[00:11:22.21] - Ned
Hmm.
[00:11:23.12] - Chris
Because fuck you, that's why, DataOne seems to say. The temporary power generators are, to the surprise of nobody, giant air polluters. Did DataOne try to at least apply a band-aid of like air scrubbers or otherwise whitewash this notably illegal action? No, of course not.
[00:11:44.19] - Ned
Why?
[00:11:45.11] - Chris
Well, see above. It's the same reason. Vineland, New Jersey saw this and fined them $1.07 million. While this is apparently the largest action ever taken against a data center in New Jersey, it is a drop in the bucket when compared to the overall $17 billion Microsoft expects to spend on these initiatives. DataOne now has 45 days to get proper permits or shut the whole thing down. Considering the Sierra Club is demanding said full shutdown and residents have already sued DataOne over the project, I know what the majority of people want to happen, but I'm just not optimistic enough to say that's what's going to happen.
[00:12:28.03] - Ned
Hmm, at least they stood up and did something. Didn't the Groq data center do the exact same thing, but they didn't apply for permits and they are permanent polluters in their data center?
[00:12:41.21] - Chris
Yep. And they're just going to go ahead and pollute.
[00:12:44.00] - Ned
Yep. No permits. No worries. EVs and charging are better everywhere else. Imagine being able to drive up to an EV charger, plugging your car in and going from 10% to 70% in less than 5 minutes. Sounds amazing, right? Well, if you happen to live in China, the dream of charging an EV in the same amount of time it takes to refuel an ICE vehicle is now a reality. Of course, the actual need to charge that fast is due to our conditioning from years of owning gas-powered machines. The reality of owning an EV requires a subtle shift in how you plan to refuel. Refuel your vehicle. The fact that you basically have a gas pump at home means you can top off anytime, and there's no reason not to plug in your car every night and start the day with a fresh charge. It really is that simple, and a Type 2 or even Type 1 charger is adequate for the vast majority of people. I'm gonna drop a link to a Technology Connections video that makes this point abundantly clear. in exhaustive detail. But despite all this, range anxiety is still a concern for many people, and they worry about how to manage longer distance driving that may require stopping for a charge or two.
[00:14:07.23] - Ned
The current fastest charging in the US is about 600 kilowatts if you happen to own an 800-volt capable vehicle, and 250 kilowatts for those in the 400-volt range. That still drives the charge time down to about 30 minutes or so, but God forbid we stop anywhere for even half an hour. It's important. I have places to be. No, you don't. Over in China, the battery and EV makers are in a battle to maximize both range and minimize charging times. The current champion is CATL, with charging from 10 to 98% in just under 7 minutes, Using a charger in the 2-megawatt range. BYD and Geely— G-E-E-L-Y, we're going with Geely— are not far behind, with all of them boasting charge times under 10 minutes for a full charge and sub-5 minutes for a 10 to 70% bump. Unfortunately for us poor schlubs in the good old US of A, we won't be getting this amazing charge technology anytime soon. China's EV makers cannot sell directly to the US, and even the battery makers have trouble selling due to tariffs and trade restrictions. It's almost like the current regime in DC doesn't like EVs and renewable energy.
[00:15:30.07] - Chris
What?
[00:15:31.10] - Ned
I know! While I find all of this stupid and short-sighted, the fact is the current crop of EVs available to American buyers will need— will meet 90% of their needs for travel, maybe more. Most people just don't drive that much or as far as they might think. And with gas prices being what they are, never having to fill up at the pump again is seeming ever more attractive. I love it.
[00:16:01.03] - Chris
This feels a lot like the AC versus DC war of the currents from the early 1900s.
[00:16:06.23] - Ned
Yeah, DC's better. We all know this. Oh, for fuck's sake.
[00:16:14.02] - Chris
Tom Scott shows Steam Deck being used to test Mars rover called Rosalind Franklin built by Airbus. Yeah, I think that's it. Oh, I need to state up front that this is not a Mad Lib. This is a real thing that is happening. So vocabulary: Airbus is the European Boeing competitor best known for not having the, shall we say, lax quality control that Boeing has taken a spin in recent years. But that is not all they do. In what was breaking news to me, they actually also design helicopters, help with space launch vehicles, and apparently also Mars rovers.
[00:16:55.03] - Ned
Hmm.
[00:16:56.23] - Chris
The long-term goal for this project is a rover that can work autonomously on Mars, but in the short term, for testing, they need manual control. That manual control at present is being provided by a Steam Deck. A Steam Deck is a rather stupidly named handheld built by Valve. It's basically a less popular but more functional Nintendo Switch. It's got a screen in the middle. It's got controls on either side. The main reason it exists is to be a portable way to play Steam games. But the hardware is also pretty open and has been reused for many other creative things in the past. Tom Scott is one of the most famous OG YouTubers probably ever. Like top 5, easy.
[00:17:45.08] - Ned
Easily.
[00:17:47.05] - Chris
He quasi-retired a few years ago, at least from making videos on the weekly basis, literally flying away from the camera, hanging from a helicopter. These days he does occasional videos talking about whatever interests him in the moment. And finally, Rosalind Franklin was the chemist whose frankly underappreciated, even 60 years later, X-ray work was crucial to Watson and Crick's work cracking the DNA double helix structure. Okay, did I miss anything?
[00:18:20.01] - Ned
No, I think you've actually explained the entire headline and we're only 70% of the way through the article.
[00:18:28.23] - Chris
So long story short, they used commodity hardware to do temporary controls while this thing is being tested. There's been some mockery of this online because aside from wantonly stealing your PII, mockery is basically all that the internet is good for. But this is, this is not dumb. This strategy makes a lot of sense. Steam Decks are stupidly named. That's beyond doubt. But they're also very well built with an open design and probably millions of hours of use that have helped smooth out any issues with control. Plenty of other high-value hardware on Earth uses commodity devices to control them. This just makes sense. Why design something that already exists and works fine elsewhere when you can just pull it off the shelf?
[00:19:15.13] - Ned
Mm-hmm.
[00:19:17.11] - Chris
People mock that idiot in his homemade sub from a couple years ago because it used a Logitech controller, but it wasn't the controller that failed. It was literally everything else.
[00:19:26.21] - Ned
Mm-hmm.
[00:19:29.05] - Chris
Using a repurposed Steam Deck would probably be fine for an actual deployment on Mars, but again, that is absolutely not what it's expected to do. The rover is slated for a 2030 mission, and on Mars, the rover will be autonomous. Which probably means AI, which probably means it will flip over and explode 6 seconds after being set loose on the Martian surface, but that will be AI's fault, not some random Steam Deck's fault.
[00:19:54.20] - Ned
Mm-hmm. Google sucks. Heat away from the Earth. There are many sources of energy available on Earth that don't involve burning anything. The most obvious is the sun. And holy shit, does it beam a ridiculous amount of energy at the Earth for free. Also, the wind and water sure seem like good options. One problem with all of them is their cyclical nature. The sun isn't always shining, the wind isn't always blowing, and water isn't— well, well, water is pretty damn consistent. Anyway, you know what's even more consistent? The heat coming up from the Earth's mantle. It is making that heat 24/7, 365, even on leap days. There's a reason Old Faithful has its name. So why not harness that heat and convert it into electricity? That's geothermal power, and it isn't exactly a new idea, but it does require that the heat source is relatively close to the surface of the Earth, at least. It did. Google-backed startup Fervo has successfully set up a plant that creates 33 megawatts of geothermal power. This is the first of 3 geoblocks they are currently constructing to bring that total up to 99 megawatts by the new year. Rather than looking for heat close to the surface, Fervo uses technology boil— boiled?
[00:21:25.01] - Ned
Borrowed from the oil and drilling industry to dig twin shafts down into the Earth's crust, reaching about 2,500 kilometers. That can't be right. 2.5 kilometers. That seems more reasonable. The twin shafts are connected at the bottom and water is cycled through the system, heating up to 190 degrees Celsius. That's like, that's, that's a lot. I didn't know water could do that. I guess when it's pressurized. Because like when you put things under pressure, then they boil at a different temperature. Anyway, the hot water is used to make steam and that drives turbines and yada, yada, power. The company plans to continue expanding, reaching half a gigawatt in the next 2 years. Google will be using the power generated by Fervo to drive, you guessed it, data centers.
[00:22:14.21] - Chris
I was so excited until literally right now.
[00:22:17.09] - Ned
Hey, maybe the solution to the data center problem isn't putting them in space. It's putting them underground. Seems like that would be way easier and not require burning thousands of pounds of rocket fuel to change a bad hard drive. At any rate, it's clear there is no silver bullet when it comes to meeting the power needs of the planet, and our best bet is a blend of approaches, one that hopefully doesn't include burning shit and releasing more CO2 into the environment. I'm looking at you, DataOne. Hey, thanks for listening or something. I guess you found it worthwhile enough if you made it all the way to the end. So congratulations to you, friend. You accomplished something today. Now you can go sit on the couch, fire up a compromised Muse charm, and give all your secrets to Zuckerberg. You've earned it. You can find more about the show by visiting our website, chaoslever.com, where you'll find show notes, blog posts, and general tomfoolery. We'll be back next week to see what fresh hell is upon us. Ta-ta for now.
[00:23:21.23] - Chris
Tom Scott shows Steam Deck being used to test Mars rover called Rosalind Franklin. Oh, I got so close.
[00:23:29.11] - Ned
So close. Almost there. Deep breath. You can do it.