 So this is my thanks slide to everyone that helped me with this project. I had a ton of help with this. You'll see from what I'll show you with the photography, visuals, getting all the licenses and permits and the substances required for this. So big shout out to all those other hackers up there that helped me with this. But here's the motivation for this work. The aspects of the things that we throw away, right? The stuff we throw on the trash. The things that have been around for a while in school are industrial secrets that people didn't want their competitors or criminals or the general public to know. We used to go through the trash, right? There's just like a lot of things that people throw away. So, you know, old phone system manuals, right? Like the brickers used to go and get. Looking for passwords and stuff in the dumpster. But now the stuff that gets thrown away is increasing exponentially, right? So we're not just looking for printouts anymore. We can hold in our hands a device that contains terabytes of information, right? And that stuff does get thrown out. So here's some examples of the way we see that security risk happening. Printers and copiers, right? They're all digital now. They used to be fully optical. But now they scan because you can email the PDF to yourself later. You can go to these copiers and stuff. There's a lot of research out there on recovering old copiers and printers and then recovering interesting information from the hard disks. There's actually a big report out in the United States news group that went out and saw that this is still a problem, right? They recovered all these confidential medical reports, sensitive information that were still stored. Some printers don't even erase the documents that are stored on the disk until they need it or the rest of the recoverable. Another example that's coming increasingly into prominence, Google in circles is sensitive information in devices that don't stay put, right? Your drone aircraft might contain all kinds of private information or information that a company considers that proprietary information, right? So like Google's maps or whatever might be in their driverless car if you can get access to that hard disk. So we have to care about this media that goes for a ride somewhere. But the point of this research really is what if someone tries to take that data from you when you're still using it? You're still using it. And so this was originally inspired by some work that was presented on 19. Machine Lawson, Bruce Potter and D.B. Mollum. Great presentation. So that's how I lost my eye. They were working with some data center scale stuff and they had this thought experiment. What if we needed to get rid of this really fast? So here are the goals of the research. We want to be able to flip a switch and have our drives gone, not a single bit left standing. Now you might say, well, hey, all my drives are encrypted, right? So I just don't care. The crypto doesn't necessarily save you because crypto is really as strong as the security of your key, right? And the key has to be stored somewhere. And you may be surprised by where your key ends up. In particular, it can be unexpectedly recoverable for your physical media. So, you know, we just need to be careful about this. And the other thing, of course, data center scale that I mentioned, we want to be able to destroy a lot of drives quickly. And we're going to also consider as our threat model the ultimate adversary, right? Extremely high capability, criminal adversary, like a bad nation-state intelligence service, right? So it has the maximum forensic capabilities. So let's take a look at what's done industrially first, right? This is what we're competing with, if you work for a company and you need to get rid of your data securely. So if you want to decommission your disk, then the standard method is you decouple the drive if it's a magnetic storage system and then you throw it into some giant industrial shredder like that that reduces it to a lot of little component parts. The good thing here is also they're all mixed together. So once again, if you're destroying something, you have to predict your adversary. So what we know is that forensic organizations with unlimited funding, I think you probably know who I'm talking about, they are able to exploit physically correct, physically destroyed drives that have been collected. So as part of this research, they talked to military intelligence officials from the United States who had been in Iraq collecting data and they were under instructions that if they found any physical media, no matter its physical condition, no matter how it had been physically destroyed, they would have collected that and sent it in as long as it had not also been burned. So physically destroyed drives. So here's your take away at home if you want to destroy your own stuff, if you want to really do good from orbit, you should degauss your drive, you should then crush and shred it, and then you should burn the remnant remnants, and then if you're really paranoid, you should throw those remnants away in separate places. Just split them up, right? Really make things difficult. But here are the rules for the research I'm going to present today. And these are adapted pretty much straight up from the F119 presentation. We have 1U for our server containing all our drives. We have 1U above and below that in Iraq whatever we want to put in there for protection. So that means we can have hot gas extraction, we can have just slabs of concrete, anything we want to contain what we're doing here. We have 60 seconds to get to completion. We want to not set up our fire system, right? Because we don't want sprinklers going off, we don't want hay lawn systems flooding the room. We want to contain our damage within our storage equipment because we don't want to damage neighboring equipment and we also don't want to damage neighboring humans whoever's in the data center. So I'll keep them safe. So those are the rules. Let's take a look at what we're trying to destroy here. Quite hard, this is a made out of. Most of the drives you'll find still in the data center are made up of spinning platters. And those platters are made usually of aluminum. And increasingly now, especially in laptop computers, they're made of glass. Glass is very easy to destroy. The reason they're made of glass, by the way, is that it's really easy to make it very flat, flat that you can spin at a high rate of speed. That sort of machining problem can be easier with glass. But mostly in the data center we'll see aluminum and that's what I'm going to concentrate on because glass smashes very easily. The surface coating, the aluminum doesn't store anything, it's just a substrate. The coating, we have underlayers of cobalt nickel and iron. And then we have the magnetic alloy, which is cobalt chromium and platinum. So we actually have some valuable metals inside a hard disk. And those layers are separated by four atoms of ruthenium. Very, very thin coating. But nevertheless, that coating makes them extremely chemically resistant. Rethenium is pretty unreactive. So that's why I haven't covered chemical attacks here. Then also a little bit at the end on solid-state drives. A solid-state drive is nothing special in terms of its physical design. It's just a bunch of high-performance flash chips soldered to a circuit board. So any method that works on circuitry will work on an SSD. So let's look at the methods I'm going to cover here. I'm going to look at thermal attacks using heat. I'm going to look at kinetic attacks using physical force and electric attacks. First up, thermal. So our goal here is that we want to exceed the temperature at which the stabilized spins of our magnetic material align and store our magnetic data become disorganized. That's called the curate point. So cobalt, which is what we care about here, that's a little bit above a thousand C. That's pretty hot. So for each of these sections, I'm going to tell you what I didn't try. So in this case, I didn't try flame torches because Bruce and Shane and Divi had covered that at F-19. Flameless chemical reactions I was pretty excited for. I did some research there. I couldn't find anything that got hot enough. An electric oven. We've got time constraints for a minute, and then also it's just not that interesting. And I think Dr. Foundry works when you're melting down your metal to cast it. I didn't look at this because we know it works. And then secondly, there's a lot of infrastructure that's required for those things. Your little inductive furnace has a giant power supply. So first of all, went into my workshop and said, well, let's start off with thermal. What's the hottest thing I have in the workshop? Let's get a shot with a plasma gun. Temperature, jet of electrified plasma coming out of it. Again, a little flame out to that side to know that it's done. You'll see that coming out of the bottom just there. Some gloves with molten metal coming out as well. So just like the timer on the microwave. It's all like the thing. When you're done, 45 seconds less than a minute. So looking pretty good. So let's open up the drive and take a look. You can see here the screws have already been taken out. And I'm wearing gloves now because I burned myself several times trying to get this hot visc open which takes a long time to cool down. It got really hot. You can see here the drive was spinning when I started this up. And you can see that ring is another look at that showing damage to the surface of the drive but it stopped spinning pretty quickly. So as it heated up that spindle seized and we didn't have any spin anymore but we cut a nice hole in the disc and here we go looking at the plate underneath and you can see that hole goes all the way through and there is the whole package. What we can see from this is that although we couldn't rely on the spin of the disc to destroy the whole thing we definitely destroyed that plate nicely so what we would need is a ring of plasma ejectors very easy to make. We could totally fabricate something like that so this is an extremely feasible map. Great. So I'm feeling pretty excited here. I feel like my research is off to a good start. But I got to thinking well these platas are made of aluminum. Aluminum is an extremely reactive metal actually you just can't really tell because it's normally covered in a protective oxide coating. Oh yeah. So what if we could use the drive itself as the fuel and all we would need is to push in the loxodizer and have everything take place so I thought well we could think about a nozzle that pierces the drive and pump oxygen in we just need to get it going so I built up a little loxody injector rig here and you can see the wires off to the side which has a little tiny igniter that is to get things going and here we go. I think you can see that the oxygen tube is melting and we were worried about starting a fire we couldn't control so we thought well we better start the safety first. So here we are with some high feed footage of this you can see the little igniter wire goes off and immediately things go getting busted inside that drive right oxygen is one of the most reactive things you can pump into a system and plenty of heat coming out here there is a lot of combustion going on this drive so this looks pretty good let's take a look inside it and while here is the outside you can see once again there is no fuel in there except for the drive itself so everything is being provided by the unit and open up and you can see a lot of combustion has happened in there and if we had been less experimental about this if we had used the metal tube for our oxygen this thing would have consumed the entire drive so there we are with the drive cleaned and you can see we've really started to combust that blood so despite the fact that the coating is chemically unreactive under normal circumstances pretty much everything oxidizes and so we're just going to eat away that drive no problem at all so this method is potentially feasible I think it shows what we need to do there wasn't a need for further experimentation there but you know what I really was getting at here when I started the thermal attacks is because when you talk to anyone about how to destroy hard disk immediately pretty much everyone has a one word response to that they're like oh, thermite so I was like yeah alright we thought we think this is going to work but who's really tried it who out there that says thermite has really tried it I certainly haven't so let's take a look but we know that the military uses thermite as a data destruction method they make these thermite in senior hand grenades they're actually issued for the purposes of getting rid of sensitive stuff and equipment and what I really wanted to do here was I wanted to make thermite into some kind of a paste or slurry that I could inject into the drive when necessary to fill it up in like a few seconds and set it off, boom right so it's going to be fairly awesome unfortunately this failed I don't have time to show you those failures but they're not very exciting because they're failures but anyway here's what thermite is anyone that's taken high school chemistry probably remembers it's a standard oxygen swap reaction between aluminum, metal and iron oxide that oxygen just changes partners like it's at a swing dance and the aluminum becomes oxidized you get molten iron out of it and a lot of energy so 2500 C is the theoretical temperature you can reach with a block of thermite so if you want to make this at home that's the ratio of 3 to 1 by weight and let's give it a shot so my first experiment here was while looking inside a hand disk there's sort of quite a lot of empty space that's not needed right there behind those read heads so even though my slurry injection is not going to work maybe I can just modify my drive to put stuff in there and the other good thing is there's some of the pins on the drive bus that aren't used so I can use some of those pins to make a little electric igniter and put in my 15 gram baggie of thermite seal the drive up and it actually still works if you're careful about it you don't get things too messy in there great so here we are with our 15 grams of thermite real time a little bit of flame there we can deal with that no problem so here's what it looked like after that definitely things got hot in there this looks pretty melty but as soon as we start taking a close look we can see a lot of this is just dust and debris there's nothing that a forensic lab couldn't deal with pretty easily here we are with a fully cleaned disk and there's basically no damage to it at all sure the electronics is cactus but the platters themselves no problem at all I wouldn't trust that wouldn't trust my life to the data being un-extractable from that so that's a fail but I'm not ready to give up on thermite yet because everyone knows that's the way to do it so taking a close look at those military grenades they actually don't use a regular straight up thermite they actually use a variant called thermite which is a regular thermite 30-70% and then 30% of variant nitrate the variant nitrate actually increases the burning temperature and it also produces gas and the gas throws everything around so it just makes everything a bit more violent that's what we want here so here's another 15 grams of thermite here inside of your eyes and we can see just from that real-time footage that we have a lot more violent of a reaction going on here here we are with a high-speed shot this is just taken off the shelf like so many VES cameras or something like that it's about 240 frames per second but we can see we're getting a nice little pyrotechnic shutter here squeezing out from under the top of the drive so certainly we're going to have to think about our pyrotechnic containment here but the notch with the kind of violence that we want here from our substance for time constraints I'm going to click ahead so here we are roping it up and we can immediately see that something a bit more violent has gone on inside that view actually did pretty good quite impressive compared to straight-up thermite I guess there's a reason why the nutrients is just falling off you can see I was impressed at the time when I saw that so let's take a closer look we can see here that molten iron that's been produced by the thermite has really been spread out everywhere all those little beads we cleaned it off we can see a few interesting things here we can see that we gloved molten iron onto the reed heads we can see we've actually delaminated the platter here that's great we know we're getting up to the temperatures we need when we're doing that and we can see we've welded the platters together too so things look pretty good but the rest of that platter still looks in pretty good shape so I think even though we're getting one-sided but hot enough not confident enough about the other side so we're being like honest scientists too we're going to call this a fail well there are other types of thermite I'm not ready to give up on thermite yet because everyone says that's the way to go and so one such variant that I've read about is copper thermite this similar reaction we use copper oxide instead of iron oxide and once again for those following along at home if you want to make it 4.41 copper oxide to aluminum by weight now copper thermite is a very aggressive thermite so it's a little unconfined we're almost at gunpowder speed there so hey let's do it let's put it in there see how it looks even though there's this giant part of smoke keep watching down in the lower right of the screen here there's the top of the drive that flew off so I just busted those screws right out but keep watching there's more to come and there it is there's the drive itself so from a standing start on top of that cinder block it rebounded up several feet in the air also produced a giant cloud of toxic smoke and Miles and I had to run out of his laboratory into the street and explain ourselves to his neighbors so a little bit embarrassing but let's take a look at how it looked here is the top plate you can see where the screws ripped through on the corners and beautiful patina on the plate itself so totally a valid method for people interested in doing pyrotechnic art here's the disc itself also it looks kind of you know like a steampunk kind of disc now it's very Victorian close to look there we can see lots of stuff has gone we've certainly used material all through the drive let's take a closer look over here we can see we've made beads of elemental copper that have stuck all over the place we can take a look here and see all these little bits of copper and you know you kind of feel intuitively like covering electronics in copper probably isn't good for them all the other some more like nice beads of copper there in the corner so once again we open it up we use our extremely scientific forensic clean up lab to the side of Miles' photo lab so we can see you know there's definitely some copper that's stuck to things but even though we might feel intuitively that copper is not good for electronics we're not really talking about electronics here inside the drive it's not the bits that we care about we're talking about magnetic storage but it's not magnetic so you're a sufficiently motivated forensic organization could easily polish that copper off again and I don't think this is going to work for you so I cast this as invisible but a lot of fun but I'm still not ready to give up on thermal because when you ask someone how to destroy drives they always say thermal so I'm going to make a case and I'm going to pack as much thermite in as I can fit our one use with the drive so this stack here that's exactly what you would like I'm going to put 250 grams now of my aluminum thermite here and see if I can just melt through the top of the drive so what I've done is I've cast the top of the drive out of a refractory surrounded material with a cavity to pack our thermite in it can survive the heat and hopefully we're just going to totally toast this drive I've got a clamp down here to make sure we don't have any hotness to jumping around the place this time so the first thing you notice here at this high speed is that containment is going to remain a challenge it's not the neatest and cleanest thing that's happening here in fact this shot set the carpet on fire in my friend's house and a large chunk of the corner of that missing now sorry Miles but look at what's going on here even after that initial spray we've got just blow torch happening here inside this box so it's looking pretty good I'm feeling confident about this one now let's see how it looks photographic lights set up there because we're pros we can see again some interesting things here we can see that we've actually melted a small hole in the top of the drive enclosure and we can see big beads everywhere of that molten iron that we produced or formerly molten iron once again you open it up and what we see is the electronics again it's just melted into slag but the disc itself we still got that shiny surface maybe we reached the temperature we needed maybe not it's hard to say there's the molten slag but here it is after we cleaned it up and once again we just don't know how much damage we caused to those pliers maybe they got to the temperature we needed maybe not would you trust your life to it I would so I would say unreliable so that's it for thermal the thing is that the bottom of the disc is a giant slag of aluminum it's a massive heat sink and so you just have to be aware of how much heat you need to dump into a drive in order to burn it so sure you can just throw a drive with thermite but just be aware of how much that you need and now it's time to move on to kinetic methods so our goal here with the kinetic that's physical destruction is we want to deform spindle, mutilate our drive preferably separate into pieces and make it so it's going to be really hard for a mechanical scanning to read the surface of that drive so what that means is if you're going out against the most powerful adversary then you should also dig out the drive too redundancy be safe so here's some ideas I didn't explore here I didn't look at if they make disc crushers operate vertically because of our 1U constraint you could make one that went horizontally and you could do it so it wasn't too interesting and then also using various high pressure cutting tools like a water jet cutter or something like that we totally could do that but you need a lot of excess infrastructure the same as the inductive methods so that wasn't worth exploring within the constraints but my first idea was what about puncturing because I have a nail gun at home that's designed to fire nails into concrete so if you could fire a nail into a concrete you could fire it through a drive here's a video of that nail gun it basically uses a 22 caliber cartridge so it's like a 22 caliber bullet and you fire it into some concrete this is the 480 frames per second and still it's so fast you can't see the nail and you can see it makes sense to me that that's cinder block so if we can punch a cinder block I feel like we can probably punch through a hard disk so here we are we built the end off the disk so that it's still spinning you can still see it but I have to do it upside down because you actually have to hit this thing with a hammer and you can see where I already tried it on the top and I bent the top so we had to go from the base but here we are at 480 we're going to straight through those drives nails the flighters, cracks the moulding at the base no problem at all so I feel like this is good and at the same time this I shot out in LA with Miles Ledlett and so I had to borrow a nail gun from someone because I couldn't take it with me on the plane and the guy I borrowed the nail gun from also had a pneumatic nail gun and he said hey you want to try this as well I was like of course we're doing science we have to try everything so here's the pneumatic nail gun and I didn't have high hopes for this because whatever it's the pneumatic nail gun designed for nailing would not conquer you but as you can see it goes straight through so here's what it looks like afterwards you can see those platters have squished together they've warped if we had a whole bunch like we could make some kind of a pancake cylinder that fire like a whole bunch of nails right around the drive no problem at all especially if we were to get off things as well so I say this method but what I'm really interested here in the kinetic method is high explosives because this is some work that I've wanted to do for a long time in general not just painting the high explosives so for a start there's no doubt we know that we can destroy a drive with high explosives that's not in question what we're interested in here also we get thermal factors for free with high explosives because they get hot too so we can do explosive welding so it's a great method it's very violent very destructive but we know it can work what we're interested here at number one is can we confine that explosion can we do something that keeps that in the rack equipment and then also this is a bit that I really wanted to do for years and years was experiment with these new techniques in particular I wanted to 3D print high explosives penetrating which is work that hasn't been done before so let me speak a little bit about that when you have an explosive it directs its energy all over the place in every direction what we care about is directing that energy to one place so here's a little illustration of that as an aside I have to say there's interesting cultural notion China invented fireworks but Americans turns it into like a hobby that people do in their homes America is a very weird place so this is at the annual convention of the American amateur pyrotechnics and what these guys are trying to do here is they kind of knock down the rack that this explosive is suspended on so this is just happening at a county fair grant they have a huge explosion at that distance because all that energy is going in every direction it's not concentrated but we can't afford to do that in our rack so what we want to do is direct that explosion and what we're going to make use of is something called the Monroe effect or what is better known in general circles as a shape charge so what that is is a block of explosives that's got a shape cut in it that shape is usually a cone but it can also be a linear kind of like if you stretch the cone into a line so a triangle groove and what happens here is you place that cavity up against the material that you want to damage and you set your high explosive off the cavity sort of counter-intuitively has the effect of concentrating that explosive and it concentrates the shock wave into kind of a jet and it'll cut through and so where that cavity was you get much more damage than everywhere else and so you form jets you can also if you line the explosive you can form what's called an explosive formed penetrator so for example here's if you buy one of these things which you can do if you have licenses you'll see you've got your explosive cone you got a line there and some standoff distance and then the penetrating jet will cut through so having a metal line that greatly increases that penetration the jet is actually composed partially of high-temperature metal and so what you want is a metal that is very dense and very ductile so copper or tantalum it's one place you'll see tantalum outside of a capacitor so here's an example of some commercial penetrators these are something that's used in the oil and gas industry so when you drill your oil well you don't want the oil to come up while you're still drilling it you want to see everything up and then when you're ready to get the oil you sand these things down the hole you use them to punch holes in the metal and out comes the oil and you slip it all up so a few things to note here you can see at the top there's a little foil coating that exposes the high explosive to whatever you're going to set it off where the blasting cap or some detonation cord and then also sometimes you have a built-in little cup that gives you the standoff distance for optimum formation of that jet so when we did this work it was out in Colorado with a local bond squad we needed a place to do it they had an explosives range and they said okay because they like research too and they also said hey we got some stuff we need to get rid of can you guys blow it up for us when you do this research we said of course we can we'll problem it off so they gave us some of these oil well-perforated and so at the end of the day we've got a couple left over and so this is just to show you guys what happens here we can put them in the ground in a hole with the jet pointing up so that we didn't have a lot of things spraying all over the place and here we are setting it off and here's a still from that video where you can see those two jets forming and going straight up this is an amazing photograph because not only do we have this great still of those jets you can see these things are this big right there in the palm of my hand and it's massive jet we can also see the spray from the plastic cap that we set the depth cord up with and then this has all happened so fast that that cord that they're needing cord which has high explosives inside a plastic tube the explosives are fine because we find both of our appropriators but the plastic hasn't shattered yet and gone away and as far as this all happens so here just proof of concept right in POCOGTFO we're going to strap an old appropriator to a hard disk and see what happens we don't care about that yet so here's some slow-mo footage and see a chunk of the disk fly off there to camera left through recovery we can see we've sliced right through the bottom molding of that disk and here are the flaps so that jet has just gone right through those aluminum platters like a piece of cheese or something so that looks pretty good here's all the bits that we recovered from that disk so you can see we didn't find one half of the bottom but the platters are what we care about but hey what's this over to the side here that's interesting we didn't notice that that's a hole that those penetrators aren't going through the disk it punched that hole in the metal plate that's the other side of that hole this is the ground underneath it and that piece of wire is the wire that we stuck in that hole to see how deep it went 15 inches so that's 37.5cm roughly in metric that's a deep hole we're going to be using these shift charges to have a containment problem because that's going to get through one U of concrete no problem at all but let's take a look at the stuff that I was really interested in here which is our design of our Monroe effect using 3D printing we have a few parameters we can adjust for our shape charge those are the apex angle of our cone the standoff distance and then the overall height of the charge that's the closest to engineering research people from the 60s that show how one of those parameters affects things that's the apex angle versus the definition velocity by the velocity of the jet and we can see a pretty clear and inverse relationship so the steeper our angle is the faster of a jet we're going to get so we take those parameters we can generate a parameter for our digital high-explosive fabrication with 3D printing we can create a code that turns into 3D models and we can have parametric functions new material is a materials handling problem basically you have to do a lot of material science to figure out how you're going to effectively deposit that substance so things like how are you going to extrude it if you're going to melt at what temperature how are you going to deal with supports if you've got overhangs what about internal voids if you need homogeneity to make sure that we get that these are things that if you had a 3D printing company you could easily spend a couple of years getting right or longer I don't want to do this with this research project so I need some way to get around it so the hacker me is thinking like what am I going to do about this and I've solved that problem that's not the problem I'm interested in so what I decided was what about printing on a traditional printer so that explosive is going to be the shape that I wanted it's going to be void 3 because it's self-compacting in this case I'm going to use a liquid explosive it just might work so again I want to sort of toot my own horn here and say this is the first open literature example of 3D printing on explosive geometries that's why I was really interested in doing this so what to use for an explosive so I did some research I found there are some liquid explosives out there on the market but there in particular the one I'm interested in was expensive and I didn't want to pay for it so I'd rather make it myself another sort of hacker mindset thing so I developed this thing we call fuel expedient liquid explosive or Felix it's similar to these other ones you can get in the United States like I said before you know how I said in the United States everyone has like a fundamental right to shoot at things that explode right there's like part of the things that Americans take around food if they're going to shoot at it it can explode you can't use this stuff for any other use but you can use it for shooting at it and so there are things called tannerite kind pack they're these ones that you can buy they're two part explosives each part individually is okay you mix them you're only allowed to shoot at but they're solids so I didn't want to just buy those they're cheap actually so instead I found this other commercial product and then I reversed engineered it and figured out how to make it myself instead of a monium nitrate it's basic high explosive substrate as nitro methane so you might have heard of this as high test racing fuel or also certain kinds of model aircraft rather than nitro methane what this means is you can just buy this stuff from Amazon and then it's sensitized with an aluminum powder that's a particular size and stearic acid coated and that's very very important getting that stearic acid ratio right one of the good things here is that you can just shift this as hazardous materials it's not explosive until it's mixed here's the chemical reaction with a stoichiometry the nitro methane is actually high explosive by itself but it's very very intensive you'll never get it to go by itself but when it does decompose decomposes into a variety of gases and water that temperature is also a gas the good thing about this is that water goes on to re-enact with the sensitizer so we're actually aluminum reduces more hydrogen more energy turns out the stoichiometry ratio is not what you want it'll work but it'll be way more expensive because we use more aluminum than you need and you'll see that coming up so let's do this my first idea was take that linear sheet chart a triangle groove wrap around into a circle and make an annulus that matches the hard disk so there's my OpenSCAD model there's my 3D printed container you can see the other side showing the triangle groove wrapped around and what we find is that we can fit about 60 grams of Felix into that slot so it's kind of a paste put it in there, shake it around a bit and it makes it give you a nice void free geometry there it is, strapped with a plastic cup so do the standoff distance and ready to insert a blasting cap and test it out so here's our first chart with the annular charge but you'll see on the high speed chart here we're still going to have to think carefully about our containment with the annular charge let's try it so we can see a few interesting things from this result number one, as I mentioned before the stoichiometry the stoichiometric ratio is not what you want we can see a lot of unconsumed aluminum we're wasting money by using that much aluminum we can see that we've stripped those platters off the spindle and we've actually compressed and welded them together and then we can see what we've tried to achieve here what we really wanted was that penetrating jet in a ring shape that was going to cut through all those platters we haven't done that we've only cut through in one place and if we look at how we have actually set up what's going on here, we can see very clearly that that one place we cut through that's where the blasting cap was located so we've done what we wanted to in a small area but the the cutting jet has not propagated around the ring so let's try a different geometry this is a great thing here we can just read whatever we want so let's do a radial shape chart where those cutting grooves radiate around from the center there's our open S-CAD model and what we're going to do here is we're going to run a loop of death cord around the top so that we can initiate everything at the same time from the top and so there's a little top case there where it will hold to feed the death cord through here's our 3D printed model we can fit more Felix in it because there's more empty space so we can get 100 grams in there we're going to put some death cord there around the top and there it is all packaged up ready to go so here's that chart things move a little further this time 960 high speed chart we're going to have to work on that it's at that disc go a long way to that what we can see is all the components were stripped off the control board and there's pieces of the platters they're all welded together plus they're shredded we did a good job here we can see there just complete explosive welding at the platters so we're getting somewhere here we're definitely destroying that disc but we're not 3D printing is doing a good job but we're not taking care yet of our other goal suppressing the blast and keeping everything in within one year so let's think about what we need here to do here to suppress that blast we need to direct our energy into the drive not everywhere else to want to couple that from our surrounding equipment so we have our explosives against the disc and we need some kind of damning material in between the explosives and the equipment shell so our damning material what's that going to be let's think about it we want to have some kind of matrix that compresses and absorbs energy and then has alternating incompressible parts to keep the matrix stable we want to just the same as firing a bullet through a lot of layers of fabric we want to absorb that energy as we go along so we need some kind of foam for example maybe a liquid and gas foam and it's going to be inexpensive because once again we're trying to do this at scale and we also want to be able to inject that what might we be able to use what's an inexpensive foam that you can get and we also want to be able to inject that when we need to because we don't want to have our piece of equipment full of foam all the time because we have to replace hard vests so we don't want to have goo constantly in our device so what can we buy that's cheap, foamy comes out when you press the button shaving cream and this is something you know again I was talking to a lot of experts when I did this research and an Explosives Engineering instructor in Colorado said hey you might want to check this out we do demos for this stuff so let's give it a try so here's a big shot this is 41 inches of 100 grand cream deck we don't care about the geometry right now we just care about the blocks this is twice as big as the last blocks plus our shaving cream we put it in a box we can see immediately but that blast is significantly smaller in terms of its envelope than the ones we did before so let's take a look at our video and take single frames this is the first frame of the explosion for each of those videos we did two experiments here with the shaving cream this is with shaving cream that's without on one of those previous hard disc shots with, from a second camera without you can really see the difference there the explosion on the left this one is twice as big in terms of explosive material as the one on the right so that's pretty serious pressure so let's investigate that so what we did we have a 75 pound Felix shot we built a little one u-wrap simulator here and have a steel plate with some angle steel then we are all set up in the shaving cream put our sandbag on top to simulate one u-wrap concrete on top and here we go low shot at 960 that sandbag certainly moves because it's just hitting there nothing's holding it down so if that was a slab of concrete that was bolted down we could definitely suppress that blast take a look at the steel plate there's like a slight dent in it compared to the shattering we've done to our hard disc our bottom plate we can see that we've definitely expanded out that angle line so we'd have to care about the sides if this was in a rack but the plate itself it's got a little impression of the hard disc on it but it's other ways pretty much I'm scared so summary of our kinetic approach with high explosives with an up-engineering effort this might work in fact we have a lot of electricity in the data center we've got the power let's use it against our meteor instead of for it in particular we've been doing spinning discs up until now now I want to take a look at SSDs so again what I didn't look at mass degausing of drives because degausing we know is a thing that we do exotic electromagnetic poles to microwave or RF attacks not time for it this time but first of all let's look at the exploding fridge wire effect so my idea first was like if we just dump a ton of high voltage into a flash chip what's going to happen to it in particular I've done some experiments where if you pass enough current at enough of a rise time that wire will actually explode and that's used in commercial detonators so here's my friend Miles as exploding which fridge wire we've used that many times work together on a lot of these projects looks pretty nasty but it works and here's what I'm going to use because at the time we did this SSDs were very expensive they were still pretty expensive they didn't want to dump hundreds and hundreds of dollars no one would give me any decommissioned ones because they're all still in service so thanks to the math works it gave me a lot of free flash drives I exploded them so here's just what the exploding fridge wire looked like 3, 2, 1 so here we are with a high speed shot at 960 it basically happens instantaneously that wire is gone you can see a lot of energy came out of that little wire so first of all I thought well let's just be fully kinetic here right let's just wrap this around the SSD and see if we can just convert electricity into force smash that drive 3, 2, 1 close the look but that drive has kind of bounced off camera doesn't look like it suffered a lot of damage sure enough when we take a look it's scorched we decamped the chip on the bottom which is the microcontroller but certainly I do not trust that that flash memory is inaccessible you could just remove that from the board and solder it somewhere else so this method is not reliable but really what I thought would work here is to put a huge high voltage power spike inside the chip itself and the best way to do that is to raise the ground pin to a huge voltage because no one designs electronics with protection on the ground side against this kind of thing this power protection is always on the power side so wire it up into the ground here and let's just dump 15,000 volts into this chip 3, 2, 1 we see more pieces there that's a good sign and here we go we're taking a look at what we've done coupled that exploding wire to all of the bond wires inside the chip we've cracked and decapped the chip you can see what's left of the wires are hanging out the traces have blown off the circuit board nice there's the cap of the flash memory chip so another look at it we've also decapped the microcontroller so this looks like a potentially feasible attack this is something I got to confess this is something that I shot much earlier on the show that I was on with Miles this is the same exploding wire rig but we're doing it here for a can crusher this is something I just wanted to show you guys because although we can clearly attack SSDs with high voltage what if we could also attack aluminum flat and dry and sure enough we're going to crush aluminum inductively so this is dumping a bunch of high voltage high current through a coil separated physically through a can and we're going to crush that can without contacting it and this happens very fast so this is a 2000 FPS and you can see that inductive current crushes that can and riffs it in half in a fraction of a second that cuts the can in half with some super high speed shoots this is 100,000 FPS and so you can see this is happening I mean even a fraction of a second doesn't cover it here check out this one on the right a little bit blown out from the flash there but everything that you're seeing in this video right now happened just less than a temp of a second so we could do this to an aluminum case drive I'm confident but we don't know how much power it would take the rig that we have I didn't try that but I wanted you guys to see that because I think it's pretty cool this is a summary of the research and what we learned the most visible attacks in each category in thermal the plasma cutter turns out sometimes simple as best the oxygen injection but you may require a sort of a complex injector to get that to happen in the kinetic side the nail guns definitely work as long as you are not super paranoid if you're going against the ultimate adversary then the nail gun alone will not work that will exploit that drive and the dam type explosive turns out to be super visible but you need a horizontal barrier as well the explosive force is going to go out the side of your rack as well as up and down and on the electric side the hydrogen power spike seems good against ssds the one thing is we don't have a good idea yet about the forensic capabilities on ssds ssds are weird when you overwrite data on an ssd it doesn't necessarily actually overwrite the space that data lived on so there's still a lot to be learned about forensics on ssds there's a bunch of research out there right now already about erasing ssds and then hooking them up and recovering data from the flash memory that assumes that the memory chips haven't been damaged in terms of what can be done to a damaged flash chip that can't still be hooked up to power and read we don't know the answer to that so again if your life depends on any of these methods then you should be conservative about your choices now I just want to leave you with a little fun example we talked right at the beginning about the mobile problem you've got your driverless car you've got your drone you've got your laptop we have some cautionary tales from the last few years about people who allowed their data to be taken from them while they were getting it and it didn't end well for them on the left Ross Albright, founder of the Silk Road darknet market on the right a more recent example Alexander Kazeps an operator of Alpha Bay they were both captured by law enforcement while they were using their laptops so their laptop, even though it was encrypted it was unlocked all of their data was recovered very bad for them Ross is in prison in the United States serving a life sentence Kazeps is dead he died within a few days of his arrest under uncertain circumstances in prison in Bangkok no good so I want to leave you with this whole thought experiment about destroying your mobile data in case you get apprehended