It looks like you're having connection troubles. Reconnect now.
01/12/2021 14:40:51
32c3-talk-7435 Latest text of pad 32c3-talk-7435 Saved Jan 12, 2021 |
Hallo Du!
Bevor du loslegst den Talk zu transkribieren, sieh dir bitte noch einmal unseren Style Guide an: https://wiki.c3subtitles.de/de:styleguide. Solltest du Fragen haben, dann kannst du uns gerne direkt fragen oder unter https://webirc.hackint.org/#irc://hackint.org/#subtitles oder https://rocket.events.ccc.de/channel/subtitles erreichen.
Bitte vergiss nicht deinen Fortschritt im Fortschrittsbalken auf der Seite des Talks einzutragen.
Vielen Dank für dein Engagement!
Hey you!
Prior to transcribing, please look at your style guide: https://wiki.c3subtitles.de/en:styleguide. If you have some questions you can either ask us personally or write us at https://webirc.hackint.org/#irc://hackint.org/#subtitles or https://rocket.events.ccc.de/channel/subtitles .
Please don't forget to mark your progress in the progress bar at the talk's website.
Thank you very much for your commitment!
======================================================================
So welcome, everybody, for the talk replication prohibited, this talk covers the subject of 3D printing and the state of physical security. Our lovely speaker today is Eric was true. He just flew in from the States four days ago, he told me, and he still jet lagging. He's a student of computer science engineering in Michigan, you say. And please give him a warm round of applause and enjoy the talk. Thank you very much. Thanks, everyone, for coming so early on the last day. I'm Eric Wistrom, presenting research on Replication Prohibited, which is our research on 3D printed keys and its impact on high security locks. This is joint work with my coauthors, Ben Burgess and J.L Telamon from the University of Michigan. So we're talking about pin tumbler locks, which I think is appropriate for this Congress, given that the the theme is gated communities. So take take that what you will. But I see locks as sort of the the gatekeepers for the sneaker net of things. Firewalls are the gatekeepers for the Internet of things. And that's what these are for, for this. So just a brief overview for those that don't know pin tumbler locks. Have, you know, these pins in them that only open the lock if the correct key is inserted, if the wrong key is inserted, then the cuts on that key only lift the pins to different heights and one or more of the pins will block the inner plug from rotating and the lock will not open. But if you put it in the correct corresponding key, then you will be able to open the lock. The the pins will be lined up along the shoreline and broken along the shoreline. And this allows the inner plug to turn. Now, of course, it's important to remember that that is the secret information is encoded in these cuts. If you can ever take a picture or observe someone having these these cuts on a key, then you can figure out how to open the lock. If that goes to and there was an attack in 2008 that sort of generalized this and went a bit further and said that, you know, you
don't have to just worry about people next to you and very close to you that can see your locks. You have to worry about sometimes people very far away that have telephoto lenses and cameras that can take pictures of high resolution from across from across the entire courtyards where, you know, you can sort of see from from some roof level, there's some keys sitting on a book that's barely visible by the naked eye. But you can actually make out the cuts on this key from this image. There's some, you know, tricks to transforming the image so that you can actually see it straight on rather than at this weird angle, but nonetheless, this is possible to do and they demonstrated this in this work. Of course, there's also bumpkins, which is a common attack against pin tumbler locks that have you cut a key to the lowest level and then put it in the lock, give it a sharp tap. And this bounces many of the pins up. And if you turn the key at just the right time, then maybe the plug will rotate as the other pins clear the shoreline just for a split moment. And finally, I think one of the most interesting attacks for pin tumbler locks comes out of MasterCard systems is unique to MasterCard systems and the MasterCard system. You have multiple keys that can open a single lock. So say I have my key that opens my office door at the University of Michigan and it only opens that one. But the building manager has a key that opens every door in in the building. And how this is accomplished is there are actually multiple pins, more than two pins in each stack. And and you have different pins that correspond to each of the different keys. So my low level change key that opens just my office door can can have cuts that correspond to the red and the green pins, whereas the master key that opens all the doors has cuts that correspond to just the red pins. And in 2002, Matt Blaze published this this privileged escalation or this rice amplification attack that's unique to Master Kaede System
s, which I call the blaze attack. This is an attack that allows a a an attacker to that has just changed key, just a low level change key to query their lock and learn what the master key to the entire system is. And to give you a sense of how this works, I'll walk through this illustration. So again, I have my low level change key that opens just my office door and it operates these the red and green pins at the shoreline. And so what the attacker does is it takes this key and makes a similar copy to it that has all of the same cuts except for one pin. One pin is different. It raises it to the to the highest level and the attacker puts this in the lock and tries to open it and if and sees if it opens it. In this case, it wouldn't open the lock because this red pin is blocking the plug from rotating. So the attacker removes it, files it down and tries it again and continues doing this until the lock opens. And because all of the other pins are still kept at their old position, that would normally open the lock. Once the attacker is able to open the lock. They have learned what the master cut is for that PIN position so they can repeat this for each then iteratively of the pins in this lock with different blanks and different different key cuts, ultimately revealing what the master key is to this entire system. One more thing to note about all of these attacks is that you have to find some piece of metal or some piece of material that can fit into the keyway of this lock. Key ways of locks are designed to be a little bit difficult to get into, both for to prevent lock picking, make it harder to get tools into it, and also to to make it a little bit more difficult to to get blanks that can actually fit in these locks. So for a lot of these attacks, for bumping, for making unauthorized duplications and for escalation, attackers really want to have key blanks that can fit into these locks. And for most of the locks that that we we encounter, this is very easy. You can j
ust go to, say, a hardware store and buy one of these blanks and actually get it copied from from the dealer. And there's no control or anything like that that's going on in in these open court systems. One layer above that, there's duplication, prohibited keys where you can still buy them online, things like best keys and so forth. And there are manufacturers that sell them. But you may have a little bit harder time finding locksmiths that will duplicate to them or cut to them. But nonetheless, it's still possible to find them. They're just more rare. Finally, there is a step above duplication prohibited, which is restricted Kiwi's where the keyway itself is often patented or controlled or somehow specifically and custom designed for each system by the lock manufacturer. And this can be system specific in a way that the lock manufacturer and the locksmith that deploys it have some kind of contract where only that locksmith can buy blanks that fit that particular keyway and they have some keycard that authenticates it to to buying that. And so you wouldn't be able to buy these blanks online even if you, you know, knew what they were. To to further make this more more difficult for attackers to find these blanks, these key designs are often patented so that if someone actually was able to manufacture some of these key highways for popular restricted highways, then they wouldn't be able to legally sell them without infringing on the patent. So if you're an attacker and you want to still get access to these blanks, you have a couple of options at your disposal, you could try to custom manufacture it yourself. You could go to a CMC mill and, you know, measure out the key that you're trying to copy and, you know, just drill this down from from some stock metal or something like that. And that certainly is possible for most Kiwis, although some key ways try to make this more difficult by having crazy undercuts and things like that that are difficult to replicate on CMC Mi
lls. But in general, this is going to cost, you know, some amount of money and take a fair amount of skill to do in practice. There's actually a machine that will do this for you called the Chemex Easy entry. And this is a pretty cool device. You just put in a key that you want to copy at the top. It has a little probe that comes out and measures different parts of the of the keys thickness and then has a second part that has a S.A.M. that cuts the smiley face, Kilink, into whatever key that you put in the top. Now, this machine is not cheap. Last I checked, I think it's about 7000 euro, but it's, you know, something that could be useful for attackers if they were, you know, breaking into banks, I guess. So finally, there's 3D printing, this is something that's becoming much more consumer available. I've seen several 3-D printers here at Congress and and 3-D printers are really sort of a fast, you know, innovating thing that's going on and rapidly improving the state and the strength of the materials that they can print. 3-D printing still requires some amount of skill, but the tools there are also improving that. You can find these these free tools to model things. And that makes it much easier than than it used to be. So one question is, if we print keys in plastic or in whatever material, are these printed keys actually strong enough to use in practice? To answer this question, our group performed an experiment on a number of different ways. We we modeled several different key ways of keys in plastic, in acrylic, in nylon, in several different materials, including metal. And we took them all and put them into locks and tried to break them off with the most expensive screwdriver that I have ever purchased in my life. It has it has a USB output and it tells you how many pounds of torque is being applied to the screwdriver. And the one redeeming quality of this screwdriver besides its price is it had a generous return policy, so. So we use this tool to measure how h
ow how much talk we could take before it snapped off in the lock and the results surprised us. We printed these on icemaker bought using clay, and this would cost pretty cheap for for each key. Now, the maker bot itself is not eight cents, but it's a couple thousand dollars. But nonetheless, each individual key that you print off of that is going to be fairly cheap. And to our surprise, this key was strong enough to open most of the applications that you would come across if you're using these keys in practice. So we rank these four things, opening a door latch, which is just a very sort of light spring at a couple of inch pounds of torque all the way down to a crash bar where you're actually physically pulling in one of those sort of massive crash bars from the outside with just the key. And to our surprise, the plane was actually able to do most of these. We marked passive. All of the keys were stronger than the talk it took to open one of these. So all of the all of the keys that we tested were stronger than, say, the torque it takes to open various padlocks. We marked it fail if none of them were strong enough to open any of the tests of of opening them and we marked it may fail if some of them passed and some of them didn't. So we also tested nylon and acrylic and these are two different materials, but they had basically the same results despite being sort of having different failure modes and properties. Nylon is a very stretchy material and acrylic sort of just snaps off and lock. And despite being more expensive and coming from actually a 3D printing service, these were much weaker than than the play originally. And this probably would not be used in any kind of attack that you would want to do, we tried Ali Meid, which is a slightly stronger plastic that has some sort of aluminum filings mixed into it. And and this was a little bit better, you know, three dollars per key. So it's still pretty cheap and it's able to open some of the weaker components, but it
won't open a crack bar or some of the padlocks that we had testing on this. But now you might also had this bad property that it had the very rough surface to it. It almost had a sandpaper like feel to it. So when you were putting it in and out of a lock, it felt like you were grinding down the brass of the lock. So you might you might actually damage it ultimately if you use lots of these in locks and then it might also break off. Finally, we tested a couple of medals, several medals, actually, stainless steel and brass and bronze as well. But but it was the same as brass, which is something that you can surprisingly print in three dimensions with with these services. You can go to these services and and give them a CAD model and they'll ship you that same object printed in whatever metal you want. You can even printed in gold, I guess, if you want to match your Apple Watch or something. So when we tested this in practice, the metal does cost a little bit more. It's about ten dollars for the stainless steel or thirty dollars or so, 25 dollars for the breast of that amount of metal in in breast. And this works flawlessly. This opened everything that we had access to and then some. It was about half as strong as the real blanks that you would buy from, say, a hardware store. But that's still an order of magnitude stronger than you need to to open many of these locks in practice. And as an anecdote, I've actually been using a brass printed key for all of my opening my office door for the last few years and haven't had really any problems with the brass one. Though I will note that the stainless steel one is a little bit rough on the lock, in part because it's actually stronger than the lock, which is made out of brass. So if it's not quite perfectly aligned, you may actually, you know, break some of the lock in some ways. And I believe all of the locks that had regular stainless steel key use had to be replaced ultimately. So. But what happens when these keys break o
ff in a lock is kind of important and how these keys fail. So I will just, you know, note that the things like acrylic are things that are they're very brittle, have these really bad failure cases that you're turning it, you're turning it. And then all of a sudden it breaks. And now you have this piece of plastic stuck in this lock and it can be very difficult to get out if it's play. I'll note the best thing to do is to remove the lock from the door, soak it in acetone for about 12 hours, then take some rubbing alcohol, mix it out, put it back in the door and hope that no one notices your door smells like paint stripper anymore and move on with your life. But things like brass fell much more gracefully. They you can actually turn this and feel as it's as it's bending and, you know, in many cases we were able to turn these keys more than about 90 degrees before they failed completely. So it's very obvious that this thing is breaking and not going to open this lock. And you can feel much before you would, you know, need to get out your broken key extractors if this were to snap off in the lock. So how do you make these models in the first place? I mean, have we just punted the skill to making these models at all? Well, in some ways, yes. But the modeling tools are fairly good. Autodesk Inventor is one of the software packages we use, has a very fast learning curve. It took us about a night to learn and print working keys despite not having any background in this kind of in this area. SolidWorks is another tool for this, but those both cost money. Autodesk has this nice feature that you can actually put visual basic scripts in it so you can just put the key cuts that you want to do and say, well, you know, this program will execute and cut the key down to whatever combination you want, which is handy for testing on the free side of things. There's a 3-D modeling tool called openside, which is pretty cool. It's entirely sort of scripting based. You write out a program
that essentially describes the object that you're trying to build. So you build, you know, spheres and rectangular solids and you take differences and unions of these until you get the desired shape that you want. So this is pretty cool for for people that are used to programing, but it does take some skill to, you know, write off this code in the first place. So what we did to sort of demonstrate that this is even easier than using those tools is we made an automatically generating 3-D model program. This program takes a single picture of the keyway that you're trying to make and model a model for and produces that model in in in CAD, basically from this image. You can then take that blank and printed on a 3D printer or ship it off to Shapeways or rematerialize or whatever your 3D printing service that you want to uses and have it made for you pretty cheap. So how this tool works, first you take an image that can just be taken from a smartphone that's straight out of the lock, that has a good view of the keyway itself that's, you know, sort of the darkest part of this image. And the next thing that the tool does is it tries to threshold this image into various places so it tries a threshold it into a black and white image entirely. So if things are more than 25 in magnitude out of 255, then that pixel is black and otherwise it's white. And same for all of the other ones. And so you can see that depending on what threshold value we choose, we get different sort of goodness of our of finding the keyway mask out of this image. And 35 seems to be the optimal here. And once we find what the keyway mask is, then we we actually once we find the optimal threshold, rather, we pull out this keyway by looking for the largest blob in this image. So if we do that for all three of these, we find that again in 35, we find the best image here. Whereas in the other ones we find either not enough of the keyway or something that's actually not even related to the way at all. But how
do we determine that 35 value automatically, how do we make sure that that's not, you know, image dependent and so forth? Well, one thing that we noticed that that was true across many different locks and pictures that we took was that if you looked at the area of the largest blob area that you pulled out from an image after thresholding, there was this large jump in area right after you had sort of the optimal keyway mask. And so you can see that here, that if you look in the normalizes blob area, it jumps right after 35, which is the optimal one for this value. So that's how we determine what the what the keyway mask is, we just go and look for that jump and say the value before that is the optimal value. This allows us to pull out the keyway mask. And we found this to be fairly robust. We tried some more complicated things that I won't get into the detail of for for computer vision sake. But we found that this was this was surprisingly effective in practice. So once you have this this keyway mask, we have a program that then takes this mask and generates some open CAD code that will essentially extrude this into a 3D model and places a bow on it that you can then, you know, put on a keychain and carry around with you. So finally, you get your 3D model of your key and optionally you can provide cuts to this this program so that it can cut it down to whatever, you know, key that you want to open, whatever door that you're trying to do. And if you're trying to do, say, a privileged escalation attack, maybe you make, you know, seven copies of this with different cuts on each one. So we released this tool as open source and. It's available on Kei's Forbes.com as a demonstration, you can actually go and try this out, you just upload a picture and optionally provide the key cuts that you want on on whatever key you have. And it will produce this steel file that you can download. And then three different. So what can you do with these 3D printed keys? As we said before,
there's these these three main attacks, still a duplication in bumpkins and privileged escalation. But what have people been doing with 3D printed keys in practice? Well, there's a number of people that have have printed keys. We're not the only ones that do this. And we're not even the first people to have publicly done this. In 2013, there was a couple of MIT students at DEFCON that printed Prima's key, which is a moderately high security key that has actually to sort of cut short lines on it that they replicated using openness can actually and and publish this code so that other people could make Prima's keys as well. That was, again, a custom modeling of just that one keyway. And they printed it using materials which after this this all went public in materials, released the statement that so that it did not support 3D printing of high security keys in particular. They said that it materialized, rejects any use of its services to promote activities or to create products which pose a safety or security risk to others. Had they known at the time, they would not have printed these keys. Which is funny, given that if you look at Dematerialize website right now under their platinum or sorry, their titanium page, they actually use this image as on their marketing page to show you what you can print with with their service. So. More recently, there was this picture published of the TSA master keys in some article, for some reason, I guess the reporter or whoever was being interviewed thought it would be cool to flash them keys. And this picture was published and then later taken down. But it's the Internet. So it's, you know, floating around and here it is here. So there's that this this does allow you to you know, it's high enough resolution that you can actually figure out the cuts of each of these keys. And someone did this and actually modeled all of the keys and and modeled this and published this. Now, I will note that TSA, Masterji, it sounds pretty bad, but at
the end of the day, the TSA locks were probably not that high security to begin with. There were probably ways around them already. You could probably pick them in a matter of seconds or just bypass the lock entirely by opening the zipper with a big pen or something. But nonetheless, it's an interesting sort of experiment and lesson in when not to show your keys on camera. And finally, the most recent study here is a tool called Photobomb, which is very similar to ours, which from a single image is able to produce bump keys of of that can be printed in plastic. And this was done late last year and the tool was never actually published. That was just talked about at Lacon, I believe. And I don't think they've released any of the code or anything opensource, but nonetheless, a very cool tool that could make bunkie's that that worked on some of these pretty high security locks that were fairly difficult to bump, even with metal keys, but sufficient to work with with plastic. So how can we defend against these attacks besides just shoving superglue in our locks and giving up? I think there's a number of different directions that we can go in with this with with trying to defend against 3D printed attacks. One is to look at non mechanical locks. So electronic locks are sort of growing in popularity these days. No, actually, that does bring in other kinds of vulnerabilities. And, you know, now you have now you have other problems to worry about, like replay attacks or or any kind of software vulnerabilities that might exist in whatever protocols you're using. But this this will scale pretty well and it won't have the problem that a lot of the mechanical locks will have when it comes to 3D printing attacks. Sort of slightly related, but a little bit more on the mechanical side of things. There are a number of high security keys that use active keys and key ways to authenticate that the right key is in the lock. So in addition to, say, having very high tolerances and pin tu
mblers and things like that, this these keys, like the multi lock, can have actual spring components in the key that sort of, you know, fold in and then come back out at different parts in the insertion process. And this might be difficult to replicate with 3D printing, given that most most of 3D printing at this scale right now is limited to entirely solid fillers and can't have these very intricate, fine features in them. And finally, I think magnetic locks are kind of an interesting sort of cool gadget. But Max is an example of this. This has magnetic HP-UX inside of it that actually rotate physical, mechanical things inside the lock that have to line up in order for the lock to open. This would again be a little bit tricky to replicate with with 3D printing alone, although you might be able to do some kind of inlay thing where you put magnetic things and be able to copy a key remotely. If you had, say, you know, a 3D printer and a good compass or something, but. I think one interesting idea that I haven't really seen proposed for for defending against 3D printing is trackways. So trapped highways are pretty unknown sort of thing in the world, but they can be used to essentially trap a key in the lock when it's inserted. So if you have someone, let's say a contractor or something that goes away and takes their key with them or something, and you don't want them to come back at some point and be able to access whatever door that they used to have access to, you can install a different keyway that's essentially configured to trap that key when it's inserted and turned in lock. And if that happens, you won't be able to pull the lock the key out of the lock and the lock will actually have to be drilled. But nonetheless, the person will be stopped from entering that facility. And this is useful for for unauthorized old keys like that. But it could also prevent privileged escalation and other attacks that are enabled by 3D printing, just by being able to trap all of th
e different sort of iterations of the blaze attack that that might happen during that process. So looking forward into the future, I think there's it's a pretty exciting time for 3D printing in locks. We're really just starting to be able to make these keys that are viable in practice. And 3D printing is only getting better. It's only getting higher resolution and cheaper and more ubiquitous. And there's more materials every day, I think. Recently, I saw that you can now 3D print in wood, which is pretty cool. But all of this sort of improvement of technology is happening on the 3D printing side. I don't see an analog happening in the lock innovation. And to be fair, locks are fairly old and have, you know, done fairly well in in recent years in history for the most part. But I am a little bit concerned because locks sort of have this intrinsic property that they're limited in sort of how small they can get because the humans that operate them aren't getting any smaller. So on the other hand, the 3D printing tools that we have are able to get smaller and smaller. And so at some point I think we'll be able to do more of these active keyway attacks and things like that with 3D printing and moving. Even beyond that, I think that there's a lot of room for looking at different ways that 3D printed materials could actually actively interact with the pins in the lock. Maybe that you could actually make a lock picking device that actually did a single pin picking inside of the device using just something that was manufactured on a 3D printer. That's probably a ways out. But I think that that that sort of, you know, combination of 3D printing and meme's like devices would would be a really interesting future for the combination of 3D printing and locks. So with that, I'll wrap up and be happy to answer any questions that you have. Thanks. Well, thank you for this interesting talk, Eric, I'm fairly sure that this was the magic moment for all the lock pickers present and the h
ighlight of that Congress. So if you have any questions, please move to one of the microphones we have here in the room. I was told that we are only using the four microphones that are down here with the numbers one, two, three and four. If you have questions, please move to the mikes. Even the Internet is not awake yet. We don't have any questions from the Internet. All right. But we have a question at microphone number one, please, to ask you a question. Does the tool include taking a picture from a key and then make it or only the keyway? So is it able that I can photograph your key from here? It only does it from the lock. So because it's using the outline of the keyway, if you were able to take a picture of the end on view of the key, then perhaps you could, you know, Photoshop that to make that the darkest part of the image and then use that to upload and maybe you have to mirror it at some point, but that would also work. OK, thank you. Next question from Michael on the floor, please. Thank you for the. American locks are notorious for having bad tolerances. Did you try your findings with European locks, for example, with equal locks, which are pretty much better manufactured than Chicago or some other junk? So we did not try it on on those particular locks. We did try it on a number of different ways. But the latest ones that we tried it on are not the easy ones or something like that. They were Everist and high security. They were often so small format, interchangeable cores, which can be harder to pick because they have the control shear and everything. But in general, it's all in the system. So any system key, the 3D printers have enough tolerance and resolution to print at that sort of twelve point five miles per. All right, next question for Mike, number two, please. OK, thank you for the interesting talk. I was noticing that the key ways of the criminal system seem pretty overcomplicated. Is it possible to actually simplify them by replicating the keys
? So is the question, can you can you print simpler key ways and not the entire the whole thing? Yeah, right. I mean, yeah, they have details that don't seem to be necessary. Yeah. So the short answer is yes. And people have discovered this before, especially with Medco. There's this surprising property that there's that actually a straight line fits through the keyway just fine. So you can just, you know, stamp out sheet metal or a credit card or something like that. You don't have to actually replicate any of the the squiggles or anything like that. But on most high security locks, that's not true. You will have to get at least, you know, some corner or something that that actually comes. And the key market is a great example of that. It has this very wide sort of leg that comes off to the side. But in general, yes, you could make something that smaller for 3D printed things you might not want to because it's stronger if it's thicker in different areas. And sometimes that's part of the the keyway design is to make sort of these these these ridges and runners that make the key stronger. Thank you. All right. Thanks. After I bashed the Internet, woke up and decided that it actually does have a question. Yes. Thank you, Al. Straight on just the picture has to be of the of the keyboard. How sensitive is it? Yeah, I don't know exactly how many degrees, but it's not too sensitive as far as sort of, you know, you can just walk up. You don't have to stand there with a protractor or anything or any kind of guide. It's a little bit tolerant because it will actually make a key slightly smaller for tolerance reasons to fit in the lock. So you have some leeway there to sort of figure out if you're at some SKU or something like that. But in general, you know, just taking it straight on and eyeballing it seems to be good enough. Thank you. Next question from Mike Namath, right? Yeah. Did you notice that any lock manufacturer is for traditional mechanical locks, so no active comp
onents trying to do something to prevent 3D printing? Of course, I haven't seen anything concrete from the manufacturers. I know they have mentioned that they're aware of these attacks, that they're interested in looking into defenses. But I haven't seen anyone that made any specific changes that are locks yet. But given that, it probably takes some time to actually, you know, make these changes in practice, I don't really fault them for, you know, not coming out right away. But nonetheless, these these printed 3D printed keys have sort of been a long time coming, I think, in the last few years, and they're just getting more popular. So I think it would be kind of foolish for for manufacturers to ignore that. But I haven't seen anything yet. All right. I see someone at Mike No.7. I'm not 100 percent sure it's open, but you can try. All right. Have you tried to make a negative form and fill this up with some kind of resin with carbon or fiber in it? Are you make a 3D printing of the negative form and then build the key as the positive while filling up with some fluids? Yes, so that's actually how the metal 3D printed keys are manufactured by the service that that makes them. Brath is done with essentially a lost wax casting where the wax is printed, a plaster mold is put around it and then melted, the wax melted out and then molten brass poured in. Stainless steel is printed in a slightly different process with you have this sort of sintered stainless steel and glue resin that you print the positive of. And they say it has the consistency and structural integrity of a sandcastle at that point, that they then put in an oven and fill the rest with bronze, which then replaces the glue so that you have an all metal key at the object or whatever you're printing at the end of that process. We haven't tried doing that ourselves, but that's certainly what the services are doing for some of the more complicated materials. And it seems to work quite well. Brass has a really hi
gh resolution, as you can see here. This is a 3D printed brass key and the replication prohibited is part of the 3D model and shows up when you print this. All right, we have another question from the Internet, thank you, have two brief questions. First of all, as have you also looked at a rotating cylinder logs and have you ever used abscessed material? I printed a key on an ABS commercial machine, and it was, I believe, not as strong as the play, but it was sort of comparable as far as that goes. It was it was still able to open many of the locks. And what was the first question? Have you ever looked at rotating disc cylinder locks? Oh, detaining Justin? No, I haven't looked at that specifically for 3D printing. I think a lot of these would extend to that. But there's probably different tolerances in sort of the Z direction that you can get for some of these printers. Part of the tricks here can be sometimes to figure out what angle or what orientation you want to print the key on, because some of these printers will have better resolution in the X Y direction and the Z direction or vice versa. So you have to sort of play around and find what the what the actual tolerances there. But I haven't I haven't personally played around with that. Kind of like another question from Mike number before ask away. Okay. Um, you've been tried you've been testing a few locks. Right. So do you think that in the future will we'll see more like locks that with hardening steel or that have challenge response authentication systems? Because from an emergency services point of view, it's quite important to be able to just drill them open. Yeah. I mean, a huge important part of of of of LOCKSS that they have to always work, right? If you run of a battery or something like that, then that's a problem. A lot of electronic locks have solved that by having the key provide the battery so that, you know, you know, you now have a battery that's external and you can replace that. But still, th
at's kind of finicky because maybe your battery runs out, you don't have an extra one and so forth. I think that it's likely in the short term that people will go more toward the active ways for at least the high security locks. It's still expensive to do that. It's expensive to manufacture that and expensive to make in practice. So it will probably be reserved only for the high security locks. And most of the homeless will sort of have the same same problems that they always did. It will be interesting to see which institutions adopt these higher security locks. If universities say, for example, that traditionally heavily used, say, MasterCard systems will move toward higher security locks or moved completely away from MasterCard systems to avoid some of these these problems, but I don't know have what will happen in the future. OK. Another question from the Internet, is it possible to use a metallic core and print the key around it? Sorry, can you repeat to use a metallic core and print a plastic key around this core? A metallic core and a key around the core, it would be difficult to do that. It is so I have seen people on 3D printers sort of print some base layer of plastic, then print and play something else, say even a PCB or some kind of electronic or a battery or something or a magnet even, and then print the rest of the player. Maybe on top of that, I don't know that you would be able to do that with the thickness of these keys. They're fairly thin. I mean, the actual especially with all of the key ways, sort of wiggles and and so forth, it's kind of hard to get an actual very thick part that you could put something metal in. One thing that you can do, however, is you can sort of leave out a gap in the back of the key and then insert a tension wrench there instead. And so now you have, you know, all the strength of attention and the plastic key is just operating the PIN tumblers and and raising them to the right height. All right, the brave person waiting f
or quite some time at Magic number eight. How do you actually get the dimensions of the key from the picture? Yeah, so in the tool that we use, the dimensions are assumed to be in sort of a standard small format, interchangeable core, so. From that, most of those are standardized to a specific height, so you can just sort of take the top to the bottom and you know what that height is for more complicated locks you might try. And something that we tried to do early on was detect the circle around the lock out of the mortise. And then if you knew how big that was, either by measuring it or entering it in or knowing, you know, the standard sizes of Martz's, then you would be able to scale that image based off of that. All right, Mike, number four, please. I'm quite excited. I haven't been thinking that line that it's possible to do that with a 3D printing, and I want to look from a different angle, like the one who wants to protect himself against any threat with reasonable means financially. What would you do? In my case, I think you'd have to define reasonable, you know, financial means. But, you know, if you're if you're running a master kid system, then I think you would probably want to either upgrade to a higher security lock or a different kind of master key system. I know Yale has a, I think, Biaggio lock that has actually two different ways. One that one, that's for the master key and one that's for the change key. And so then these privileged escalation attacks are not a problem for that kind of system. I don't know the relative cost of that offhand, but I know that MIT has moved to those locks almost exclusively. I guess they maybe had a problem with students being part of the escalation or something like that. But I don't know many other universities that have done that in recent years. I think that's a good compromise for that particular attack. But again, it depends on your threat model, right? If you're concerned about people copying your keys, then mayb
e you want to go with something that's just restricted and so has active components to it, like a multi lock or something like that. Or if you're concerned about bumping or picking, then you can find bump resistant locks or something like that as well. So. I may add one more question. I've seen some people using no locks on these not so good or so, no pads. I mean, they're kind of an electronic lock, right? It's it has many of the problems, but it has a limited interface. So it's a little bit better than say something that's, you know, talking to your smartphone or something like that as far as the attack surface goes. But it still has the problems of, you know, if it runs out of a battery, there's no backup or something like that. It can be difficult to to get in and reliability problems there and can also be hard to weatherproof them in some locations. And no pads in particular have this problem that if you only ever enter correct combinations on the number pad, the correct buttons were down more than the incorrect ones. And so you don't have this side channel for learning on. What the correct combination is so. All right, the Internet is curious and has many questions this morning. Yes, yes, yes. Have you looked at Kramnik or EPOXI like materials for keys, sorry, Carmac or EPOXI like materials for kids. I have not I don't know that you can print any epoxy, but maybe you could do some kind of, you know, mold based thing on and use that instead. Play, I think was fairly good in part because it has this sort of sweet spot between being flexible and rigid. Nylon was was way too flexible and just sort of like a rubber rubber band inside the lock. And acrylic was way too brittle and sort of didn't have any give to it and just snapped off. So I'm finding that sweet spot is, I think, pretty important. All right. Any other questions? You have another 10 to 15 minutes to ask the hell out of the speaker. Now, that is here and you can get your hands on him. So any other ques
tions? It does not seem to be the case, and even the Internet is satisfied, at least for now. So please give another warm round of applause to speak, Eric. Thank you.
