 Good morning. Thank you for waking up early on a Saturday for caring about public safety. My name is Josh Corman. I don't think I have a title slide, but I'm one of the founders of I am the cavalry.org. And I might touch on what that is a little bit, but mostly I'm going to talk about one year ago here on our first birthday we launched a five-star automotive cyber safety framework. We basically said to the auto industry through an open letter in the mainstream press we said look you're masters of your domain, we're masters of our domain, and now that our domains have collided with computers on wheels we're going to have safer outcomes sooner if we work together. It was a bit controversial at the time because our friends thought cars would never be hacked or even if they could no one would do it and now you fast forward one year and there's been quite a lot more activity on that front. But our basic belief was that if we leave them to their own devices they're going to either make the same mistakes we've made for 15, 20 years in enterprise security, but the difference is the consequences of failure are so much higher, right? It's life and limb, it's your family, it's flesh and blood. So anyhow, over the course of the last year we've been aggressively trying to work across the automotive industry as more of a helping hand than a pointing finger. And I'll give you an update of what's worked, what hasn't, where we're stuck, and maybe how you could help. But the last two weeks alone have had pretty massive breakthroughs, good and bad, in how this is all playing out. So it kind of started two years ago with Charlie and Chris. Now they are not the first to hack. Sorry for the late speak on unsafe at any speed and apologies to Ralph Maynard. But as many of you saw they had Annie Greenberg who was at Forbes at the time driving a car, they did this to a Toyota Prius into a Ford vehicle. And they essentially showed that while connected to the dashboard directly to the can bus, they were able to manipulate the things that the can bus can be manipulated. And it was a real wake up call for several people including a senator, Senator Markey in Massachusetts who was like, I don't like this, what's going on here? So it scared quite a few people and yes they wore track suits and did it for a little bit of glory and ego, but it did capture imagination and attention and it started a lot more people looking into car hacking. Now the kind of things they could do, hopefully you guys know at this point, but they could lie about the speed on the odometer, they could tug on the seat belts, they could disable the brakes which scared the bejesus out of the driver. Luckily it was in a parking lot at the time. But the one that really bothered me is because some of these have parking assist functionality and have a very strong motor that can control the steering wheel, they could trick the car into thinking it was going zero miles an hour and was doing the parallel parking and they could jerk the steering wheel out of the driver's hands and it's much stronger than you'd be able to resist. So one of the things I try to tell, what you're about to see is what I tell Detroit or what I tell the auto industry and since we launched this, Bo Woods myself, Craig Smith, we had about a call a day for the first 45 days after DEF CON last year and it was pretty much everybody. It was people who make the tech packages for automobile makers, it was automakers themselves, it was automotive dealers, it was trial lawyers, insurers, regulators, DHS, so pretty much anyone with a stake in car hacking or in the automotive industry wanted to know more. And at first they kind of hated us and thought we were going to blackmail them and drop O-Day on them because Charlie and Chris are fairly hated by large chunks of the automotive industry despite what they tell you. They might have one or two people in the security team who are really happy to see the external pressure but what they're doing as well and this is one of the reasons we have to have plural voices is they're also scaring people and they're scaring the legal teams and the PR teams to be polarized against the research community. But we had to find ways to fuzz the chain of influence and figure out how to make them engage. So this was a presentation right around April, a couple of chunks where I finally found our stride and we finally got them to realize that we're here to help. One of the things we said is all systems fail, right? Yes, all systems fail. So we had to ask the room, is there anybody in here who thinks a car won't be hacked? Is there anybody left here? And one of the things I like to put this because you have a whole bunch of engineers, physical engineers and systems engineers in the room is that there's a certain number of defects per thousand lines of code no matter who's writing them. There's a range of defects and when you look at a modern vehicle there's over 100 million lines of code in modern vehicles. Some of them are much worse. So you know there's bugs in there and you know it takes one, right? The presence of a defect doesn't mean it's exploitable but clearly there's gold in those hills, right? But that's not really what the issue was because a lot of them knew that and they've had software in cars for a very, very long time what they weren't paying attention to because they were trying to believe their own story is that the number of remote attack surfaces was exploding. So the vulnerability of being there wasn't the issue if you had to plug it into a dashboard. And for a long, long time their attitude was well, if you have physical access to the car, you know, that's not really hacking. You could slash the tires or cut the brake lines. There's all sorts of things you can do and this prevented them from being motivated to do something about it. Now what they weren't paying attention to is in parallel they were adding 4G LTE Wi-Fi hotspots standard to all Chevy vehicles for example. But you have Bluetooth in your field, radio, you have the tire pressure monitors which were mandated by California so you could have fuel economy with having full tires. You have the App Store which could be malicious third party apps and raise your hand if you don't think you can pop malicious third party apps through an App Store. And then other sorts of attack surfaces and even a less hackable car is proving to be hackable because you have these OBD2 dongles from Progressive from your insurance company who if they have a vulnerability they are now the stepping stone onto the wide open current network. So that's really what's changed. It's not so much that they've been vulnerable although they are increasing their complexity attack surface but that the remote access to them is what's going up through the roof. And the App Stores for example. And this doesn't even touch on the vehicle to vehicle and vehicle to infrastructure protocols which we will touch on. So ultimately when you actually push them they say yeah but Josh no one would hurt you there's no money in it. Which is A is a stunning lack of imagination because I can think of several ways to make money off of car hacking. But even if you think about that we don't just have one type of adversary. If you ever done a threat model before the first adversary is Murphy. It's accidents and adversaries. And just plain old straight up glitches. Could have consequences for life and limb. I've never seen how often you have to reboot your laptop. But I'm actually much more concerned about different types of adversaries. It could be script kiddies. I don't think they're going to try to hurt you but they could accidentally hurt you. You could brick a bunch of systems. Russian business network is a traditional one for enterprise security but if people forget Putin jammed a bunch of cyber systems as he drove tanks into Georgia. So this is a form of asserting will or disabling vehicles or emergency response. There's the entire nation of Brazil has a sole source vendor who has vulnerable remote kill switches and all their emergency response vehicles. So cop cars, ambulances, fire trucks, shipping containers. If you wanted to disable the city. If you wanted to cause panic you could do so en masse. And that's not just Brazil. But that feature once added has now made its way in every make and model. So you have to think of every motivation in the human condition. And it could be someone like Jeremy Hammond from Mulsec who is much more aggressive and anti-law enforcement. It could be someone like a nation state. It could be political campaigns. But there's every motivation in the human condition. And what I like to remind people is where was the money in the Charlie Hebdo attacks? There wasn't any. Where was the money in the Boston Marathon bombers? Or what motivated the kid to shoot up his school in Sandy Hook, Connecticut? So I don't want to have to depend on the kindness of every human being on earth. And as Dan Geer likes to say, on the Internet every sociopath is your next door neighbor. So I just don't like the dependence we're doing. We went from a period where they couldn't hurt me to now where one I'm hoping they wouldn't hurt me. Now one of the things that got their attention though is they don't want to be regulated. So one of the responses to Charlie and Chris's research is letters went out from Senator Marking in Massachusetts and Senator Blumenthal in Connecticut. And they basically said we want you to answer these survey questions about your readiness for hacking and tracking. They were concerned about safety and privacy. And after a year of aggregating these and doing analysis on it, they put out a blog post and a report called Hacking and Tracking. And in it they kind of hinted that they might be doing a cyber bill on cars. And they hinted they were doing three things for privacy and three things for security. And a lot of the hackers kind of made fun of some of their points, but there was some good instinct in there. But what this did is it changed the tone of the automakers to stop ignoring Charlie and Chris or stop ignoring the Cavalry's efforts or stop ignoring Craig Smith's open garages and say wait a second, if we don't get our act together they might regulate us and we'll have to spend years undoing bad regulation. The other thing that happened almost the exact same time is there was a class action lawsuit after the 60 minutes piece where DARPA showed they could hack a particular vehicle. In fact, in that one week period in February, you had the 60 minutes piece which told the general public that cars could be hacked. You had the Senator Markey report come out. You had the BMW vehicle compromised if you remember the hack in February. And you had Craig Smith and I working with NBC New York. We showed that the NBC reporter could hack a vehicle in Seattle from 3,000 miles away on the news. So that was a really nice concentrated amount of okay maybe we have to stop ignoring this. So what we had published one year ago, the formal names are we basically said look you have five star crash writing systems let's have some sort of way to instead of just tell you you screwed up on this one feature on this one make and model. Nick and myself the other Calvary folks we said we have no interest in finding and fixing one flaw in one device on one make and model from one vehicle. We want to make the whole industry safer. And to do so you need some sort of common ground and some minimum denominator. And no one wants to create the PCI standard for cars, right? In fact I think I skipped that but I probably shouldn't. Yeah, let me just take a second to say that's here. So we hate hackers obviously are a bit allergic to formal rules and government regulation, right? Now we don't have a political ideology but in general there's very low confidence the government will figure this out. And I completely appreciate that perspective. We know they're going to be acting so what we wanted to do was make sure that we help them act in a more intelligent way. And I spent a few years personally finding the PCI council for credit cards because I actually call it the no child left behind act for information security because it was actually making customers weaker. It was making enterprises do really old ineffective things instead of doing really good things they spent all their time and money on the wrong stuff. So when we knew left to their own devices we're going to have a PCI for cars. And I just want you to think about this for a second. We spend about 80 billion dollars a year protecting credit cards. About 80 billion. And it's an abject failure. Nearly every retailer has had a compromise of their credit cards. It's an abject failure. The only reason it's acceptable is because the banks as long as they only have a 4% fraud rate in total payout it's still okay. So the standard doesn't prevent breaches. It just makes sure that the damage in financial terms is acceptable and they keep it within that 4% range. So I've got to ask you is if we even replicate the 80 billion dollars of spend and the 140 security categories that we bolt on to payment card systems is a 4% additional tax surface and failure rate acceptable for cars is a 1% acceptable for cars. So not only we have to be as good as enterprise security we need to be significantly better and that's why I knew if we didn't try to articulate this to insurers, to regulators, to automakers they're going to come up with a PCI for cars. And it's going to be you need a firewall and an IDS and antivirus and all this terrible stuff, right? And there's going to be a role for bolt-on security and there always will be but we wanted them to take this more seriously and have built-in defensible architectures so we actually have a fighting chance. So what we basically said is you need five postures towards failure. Tell your customers publicly how you do safety by design. How do you seek to protect them from any risk incurred by their computer own meals. The second one's really key for this room which is do you have third party collaborations? Do you have a public attestation of your that you will not sue third party researchers acting good faith following your process which is a huge deal for getting us over the hump here and I'll get in that a little bit more. Do you have evidence capture? Do you cannot decry that you have no evidence of hacking when none of these vehicles have any evidence capture to prove otherwise. It's completely circular. In fact right when we presented this last year somebody came up to me said I know Michael Hastings was assassinated you know you guys need to take this up the hill and I said that kind of talk is going to get you nowhere but what I can tell you is if we can get something like this into place there will always be fingerprints and evidence if someone does try to hurt people in their cars and we can learn from it. Number four security updates so you can't have a hackable vulnerable remote interface on a vehicle and no way to actually fix it and I'm not sure if you're comfortable with sending us b dongles out through the mail and hoping people apply them correctly but that doesn't feel very robust to me if when I was talking to people in the auto industry optional recalls have sometimes less than 10% adoption so everyone's celebrating this thing got patched no this thing got a patch available this thing isn't patched until it's patched if you can trust us with BMW they were able to over the air update every single one of their customers before their customers even knew they were vulnerable I'm not saying BMW is awesome but the distinct advantages of comprehensive and quick response time is a key that's going to be necessary for cars because we're gonna get hacked often and then segmentation isolation so we're going to dive into some of these but then my neighbors that I don't know any of those things mean Josh and I realized the much easier way to say this is there are five postures towards failure if all systems fail tell us how you avoid failure tell us how you take help avoiding failure tell us how you capture and learn from failure tell us how you respond quickly to failure and tell us how you contain an isolate failure so we're going to show you what we told them for a couple of these so do you have a published out of station of your secure thing blah blah blah now these are corporate sounding words but basically a lot of them think if I do a pen test from one pen tester I've done a security program and all that does is measure your mistakes or identify a subset of your mistakes so we kind of pointed them that even though we used to make fun of Microsoft they're very over overtly transparent about what they do and even if you don't want to follow Microsoft's software development life cycle it includes things like paper based threat modeling before you write a single line of code it includes things like adversary resilience testing it includes things like system hardening configuration and attack surface reduction and lease privilege and we wanted to show them that that last little mile of do a pen test at the end you may have thousands of findings whereas if you have defensible resilient designs and people we will help you do them you may only have dozens of findings so hopefully this firm knows this kind of a concept and they can make a really weak at a station and maybe their customers won't trust it or they can make a really robust one like we expect maybe someone like Tesla could do and maybe their customer say hey if I care about safety I'm going to go with this good now at the moment hack ability is not the primary driver car purchases but there was a time when crash testing wasn't either and if you just had a new kid you want the five-star crash resting car not the three-star and you can't even tell me in the room unless you're from the auto industry what the heck the difference is between a two-star three-star four-star five star for crash ratings but the public can make decisions based on their relative symbolic risk level this one's key for this room which is the third party collaboration they are terrified of researchers in part because they get these big surprises and stunt hacks on the news and it puts them in a crisis management mode and other reason is we found out many of them have been distorted someone finds a bug threatens to publish it with full disclosure unless they're given a lot of money so many of them have had you know criminal extortion on a regular basis so they kind of hate researchers I think all hackers are going to extort them and for the last year we've been building the trust with empathy to make them see there's lots of types of researchers not all of them want glory or money some of them want to make the world a safer place so do you have a published coordinated disclosure policy if you know Katie Mazuris who's now at hacker one but she wrote she ran blue hat and the blue hat program and the bug bounties for Microsoft she also one of her claims the famous she helped write the ISO doc and I can't believe hackers actually like an ISO doc but this ISO doc 30111 tells companies how to have a coordinated disclosure policy that promises that if any researcher brings you a bug you have seven days to acknowledge the receipt that bug and what we're encouraging these guys to do is adopt at least the minimum bones of this internationally recognized standard on vulnerability disclosure and if you do this you're basically saying we will not sue you and this is the best way I put it is it a beware of dog sign or is it a welcome mat one of my friends decided to play with his his own car bricked it and was afraid to get in legal trouble just because he was trying to tinker with how does his car work maybe one of you guys in this room knows that person I don't know this one is a really really hard one but it's a really really important one the first two you can get just by making a policy choice you can say we're going to describe our security program the no engineering effort required to do that you just have to have the guts to be transparent to your customers what you are and are not doing the second one doesn't require engineering either in fact bug crowd and hacker one I skipped that for now just in the sake of getting to the flow but in there we tell them how to do these things better or worse and in that we said look if you have a recognition reward system it's even better right telling people you won't see them is step one if you have a recognition reward system is even better and what's even better than that is if you're using a coordinated service like a bug a bug crowd or hacker one then every researcher knows they'll have a consistent interface regardless of car company and every car company will have a consistent interface regardless of how wild the researcher is and it works out for both parties that they can actually get bugs sooner and it also at a minimum this tells your customers that we take your security so seriously that we want to find bugs any way we can and fix them as quickly as possible. The evidence capture is really hard this is the first one with the engineering required is going to be hard and the standards to get there are going to be really hard but you cannot complain that no one's hacked your car if there's no proof to capture if there's any hacking at all so just like we have a black box in an airplane or in a train we have the NTSB internationally studying failure so that anytime a plane goes down we can make sure that those particular conditions never really hurt anybody else and then we can stop this conjecture of no one's hacking or no one's trying to hack because you'll have evidence across a global basis with a consistent amount of security capture and logging to know if there's reconnaissance being done, experimentation being done if you see something might be wormable you'll have concrete data instead of conjecture and belief. Now this one scares the bejesus out of them because previous privacy people have fought this and they should fight it because the original ideas of a black box were tracking the movement and activity of the driver and what we've shown is that's a false coupling you can do this without having any impact on privacy especially in international context Germany and other parts of Europe are way more diligent about privacy than the US is and since these cars are sold in multiple countries this has been stuck for 10 years in debate because they thought you can only do a black box if you have infringed on civil liberties so we said guess what guys you can do something just like we have in the server world then case and court admissible evidence that has nothing to do with privacy just focus on the integrity and operations of the system and any sort of security relevant events. This one's controversial and I get in fights with some of my best friends in the DEF CON community about this but security updates so when something goes wrong remember star number zero is that all systems fail remember we said there's over a hundred million lines of code in these vehicles and remember we said there's a dozen attack surfaces to reach that vulnerable code so they will be hacked as we've seen the last two weeks with Charlie Chris with Sammy with the Tesla guys these things will be compromised the question is what's your best response time and I was around at a government vehicle thing with six different car makers this was last November before they fully trusted us yet I said so guys we all know there's a remote kill switch vulnerability in most of your makes and models it's implemented differently but we all know this could happen my question to you is if you got hacked today and you're on the news tonight what is your best case minimum response time to be able to fix it it's just a simple fair question right if it's good 30 seconds no one answered no one wanted to be the first one to admit it and then one of the automakers from the U.S. said well let's just face it the 2018 models are already done now I got a lump in my stomach because I thought it was going to be measured in months and when he basically admitted in that statement is he can't fix it at all they know it exists they're hoping no one violates it and the best chance is 2018 and then someone else that actually know it's worse because our IT packages are done through about 2020 and unless they're compatible with those 2018 jasses they may not work either now this should bother you if the response time is measured in years so I advocate that while it does add a remote attack surface it's damn worth it so what we're going to do is whether it's an over the air update or a robust update mechanism we do this on our phones we do this on our PCs if you're going to have software that can be hacked invulnerable we need a similar OODA loop and a similar response time we need a similar capacity not just to have a quick response but a comprehensive one those USB keys are not going to reach every vehicle even the Takata airbags that everyone knew about on the mainstream news don't have a hundred percent adoption rate when they tell you repeatedly you need to fix your airbag it doesn't do it high portions for some of these things like thirty percent don't ever do it so if we want to announce this thing's hackable and then assume that that hackability has been removed because a patch has been issued this is a much better way to do so and if you're from the industrial controls world there's a really harsh conditioning that we should be using read only memory and things should not be patched rule and they cannot be patched and this adds attack surfacing complexity of course it adds because the alternative of a multi month response time or really poor uptake is not acceptable and I would love to debate that because we want to make sure our thinking is sound and the last one is probably the most important one if you've been watching the Jeep hack which is do you have logical and physical isolation between physical critical systems and non-critical systems if you're surprised that the infotainment system and the UConnect could be hacked in the Jeep vehicles then I think this must be your first fcon perhaps so that's not surprising what should be surprising though is why is that able to control the power the steering the brakes the answer is they're on the same wide open can bus network as everything else or at least even if they're on plural can buses they can all freely talk to each other because there's really poor segmentation isolation so I fully expect we're going to circle and everyone's going to do the me too it's just like heart bleed right if you remember heart bleed coming out it had a pretty logo but what people didn't see is the other 41 in 2014 that didn't have a logo once that bug was found there were 41 other CVE's published against the same code base so there's going to be a lot more of these hacked but what we aren't thinking about is if we have the segmentation and isolation if we do separate critical from monitoring you can have a flood in several compartments and still not sink the boat and the wide open nature of these is the issue so while the stunt hacking got a lot of attention and while we have a lot of hearts and minds now paying attention what I was really disappointed is very few of the pieces in the media actually covered the fact that a robust fix to something like this is that future we don't want to be sick and think about it but we can't just keep pointing fingers at past failures we have to be a helping hand for future success we can't just look at offensive techniques we have to help them with defensive strategies and we can't keep looking at the current fleet we have to be making sure the future fleet is making smarter design choices so we can use nicks back in 2009 is what we're trying to do is make sure that we're not just focused on mistakes that have already been made the spilt milk and what we saw is in February everybody made fun of BMW because they got hacked but we wanted to say if you take those five stars we just described how is this a success story and we did an analysis number one they got this brought to them in a collaborative way and number two because they had remote over-the-air capabilities they patched every single one of their vehicles before a single adversary or even a single customer knew they were vulnerable and then push was comprehensive but here's the best part without anyone forcing them to do so someone pointed out that their remote over-the-air update was passed in the clear and submitted that and re-corrected it and tightened their systems they went in an iterative continuous process they didn't do the researchers they fixed it over the air they tightened their system design and told others including competitors who now have the opportunity to avoid making that same mistake so am I saying that I love the fact that they were hackable in a bad way no but am I happy that we bring to things to wake up every day and look for what's wrong with something one of the reasons we're building trust with Craig Smith with open garages with us is we're telling them what they're doing right and then we follow it up with how they can do it better and you think about Microsoft I don't know how long you guys been in the industry but I feel like it's been way too long for some of them usually Microsoft sometimes Oracle unless you're Litchfield in which case he has an entire series of them but they went from hating researchers to you fast forward and they had the blue hat conference which was small private and invite only and then they expanded that and then they added a bug bounty and then they added the blue hat prize which is a six figure prize for finding defensive evasion techniques and whatnot that happened but so were several others and I call this the meantime to enlightenment how do they go from cease and assist to a continuous thing where now Microsoft's execs find it to be a critical and necessary part of their SDL and their business value in fact there are parts of Microsoft that are fighting to get into the program because they can't actually service every division companies be as enlightened but what we can do is compress the meantime to enlightenment to maybe three years because if we get too impatient they're going to lose our help so we have to put some pressure on but we don't want to put so much pressure on that we expect them to go straight from infant you know you got to crawl and then walk and then run we can't expect them to be you know Microsoft level a huge mistake as well and if you're all involved in this arena please help me with this is the strategies you need to use for the past fleet are going to be very very different than for the future fleet in fact we might even need three strategies but what I found is you're so concerned about the cars that have already been built so they want to do a bolt on thing like an IDS or an anomaly detection and there's a whole bunch of third party ones and there's some pretty good stuff right they're going to have a surprise for bolt on versus built in we have to favor built in wherever we can we have to favor least privilege and segmentation and isolation and tamper evidence and all these other things but they're so focused on the fact that the other cars that are already shipped could be hacked that they're still doing bolt on for everything and while you're looking over your shoulder you're making the same exact mistakes on a go forward basis so I've encouraged most of the automakers in the building so that we don't just keep doing the band aids forever and have to pick out an antivirus to run on my infotainment system every time I buy a new car so if you're working with them make sure they're looking both at the current problem and the future problem and then the one that really sunk in and hit home anybody know what this picture is say it out loud it's deep water horizon that's right I can't remember how many days but I think the last number I heard was like 38 billion and counting and the cost to BP so this is one of those things where it happens and they can't do anything about it and to the auto industry we've told them it doesn't matter which one of you gets attacked it's going to hurt all of you and part of the reason is if people lose confidence in their vehicles it'll have a material impact on their bottom line on energy commerce and transportation I said you know I'm surprised that you guys are looking at the regulation usually you wait for something bad to happen you don't want to screw with the free market and she said Josh let me be really clear and I want you guys to pay attention to every syllable here she said the automotive industry is a double digit part of the US economy and any loss of faith 99 Americans are employed directly or indirectly by the automotive service industry or parts industry and this could cripple the economy if people lose faith so they're taking it seriously proactively because they see I think it's like 20% or more of the US economy all things considered it's insane so this isn't something that we're just hobbling with and if you haven't been in the car hacking village to play with the faith we're placing on something never designed for security and if you haven't started playing with it then you should be reading the car hackers handbook and you should be taking courses and dabbling with the can tools that Craig and others are building because it's not that hard once you understand how the system works and you can bring your passion and talent to something that avoids something like this now the last thing is the reason I'm bringing this up is autonomous vehicles or semi autonomous vehicles aren't 10 years out they're already driving if you've been to Menlo Park and you've been around the area they're already happening in fact Volvo I think it was Volvo or maybe it was Daimler Chrysler just applied for permission to do autonomous tractor trailer shipping trucks about two weeks ago I saw the article so there's trials and autonomous vehicles already but one of them is and vehicle to infrastructure if you've seen Cesar's research from IOActive where the smart city stuff doesn't even have any authentication they want low grade technicians not to have to remember any passwords so they're passing vehicle to infrastructure information completely naked in the clear on purpose so one of the problems with this is if we can't make a single car secure one of my idols is Dan Geer and he's got this security principle security is not composable the way I put that is if you take a secure thing A and you take another secure thing B and you put them together you can't assume it's secure there's seams and there's cracks so you cannot compose security but the corollary to that is you can never take a secure thing A and combine it with an insecure thing B and expect it to be secure and for security and none of them are tamper evident to know if they are no longer trustworthy all the fruits and benefits we're going to depend upon will be based on it's a castle built on the sand there's no foundation here so I can look at it as a prerequisite for V2V and V2I to work is that we can actually trust the integrity of the individual vehicles participating in it not to mention it will be the largest I know you can do PKI but you're the only one I've ever met so basically what we try to say is the road ahead is up to you we can continue to have an adversarial relationship or we can embrace these five things to at least give you a foundation those five things are not a PCI checklist and it's not the finish line it's a starting line I think the way you put it Craig was no one forced you to put yourself on the internet failure take help avoiding failure notice and learn from failure I respond to failure and contain an isolate failure I don't think that's unreasonable and yet they've told us that they can't do something like the black box until at least 2020 and none of this really changed until you as a researcher help us build the body of knowledge but also you as the car because you like how it looks you like the fuel mileage you shouldn't have to worry if they've got a 4G LTE Wi-Fi hotspot in fact small anecdote my wife went to buy a new car when I was in DC and the guy was trying to sell her how hard really hard on the the 4G LTE and that the kids iPads would work in the back seat she said I don't think I look at this is of course they're going to differentiate themselves in the market and in fact I saw a car commercial whose entire commercial was a bunch of kids at the back seat playing on their iPads and they went in a different car and the iPads didn't work they said who wants this car let's go back to the other one and that was the entire car thing nothing about features nothing about fuel economy nothing about you know the Wi-Fi hotspot and it can't shut off my brakes that elective attack surface is insane like we're going to look back at this point in history we're going to say why the hell did we do this and one of the ways I put it is things like asbestos it was cheap it was lightweight it was fire retardant there was tons of obvious benefits everybody couldn't get their hands on asbestos fast enough it killed a whole lot of people there have been class action lawsuits billions of dollars but moreover those hospitals those schools those manufacturing facilities had to get retrofitted condemned cleaned up it was one of the biggest mistakes in infrastructure history and I think what we're seeing here I know I'm going to say the word cyber again this is our cyber asbestos period and if we want to do something like that go for it just make sure it's segmented and isolated from the telematics and the brakes or make sure you do the commensurate care to make sure you've done adversary resilience testing you haven't just paid a pen tester to do so so that's what we showed them and I can take questions or one of the things I was going to do for like the last ten years we've been working really closely with congress critters on both sides of the aisle and both the house and the senate because we know what they're trying to do and until they can talk to more researchers or more auto makers in a more candid way they may make really bad decisions so I thought about telling you what I liked and didn't like about going past something and that something will be really bad sort of bad pretty okay or might be really awesome and make sure that you can do the research without threat of criminalization that you can get safer cars sooner and as such we can talk through some of those things as well does anybody have a strong opinion any value in this we just want to poke out so the question for the camera was the car companies said we can't fix it for one more time oh double digit percent of the yeah so we're in a bad position so if it's 20 percent of the GDP and they're currently vulnerable and we can't fix them until 2020 at best how do you resolve these the answer is we're in a very precarious time in history and while I would no one's done and I'm trying to be very fair here and let me let me put it from the perspective of how Washington saw it number one the research was top notch really good stuff they did a good job on the research number two they told the auto manufacturer discreetly in October so they gave them advance notification before coming here plenty of advance notification before coming here number three they shattered it requires physical we don't have to do anything about it yet and while there are several really good things in fact that's one thing I should not skip right away we were introduced to the side of automotive engineers right away we were introduced to the department transportation right away we got involved with DHS and parts of NISTA so we've found incredibly talented people in industry and they're doing smart things now some of their aim was off they were trying to do something for that segmentation and isolation that has never worked for us but they were doing the right things and I think they deserve mad props and credit for that but the transition period between adding remote connectivity exploitable stuff knowing they have and being both impact and kinetic impact and we're kind of hoping that no one noticed and the adversaries don't take advantage of it and they weren't a foot race between fixing these things robustly and having very large consequences of failure so back to the Charlie analysis right if it's great research they did disclosure without you know full drop in O-day and they shattered a long-held excuse that kept some of those people they did several more years and they don't they never did but they shattered that excuse and by the way just so I'm precise this is not the first remote over the air hack it's just the first one that rose the level of public consciousness right I mean Savage and Yoshi did it what 1300 miles away if I think in their white paper and the U6 guys also did it but they did not test track and they were academics and they didn't it went off the rails and I know it went off the rails for congress critters is by doing the test on a highway without consent without police escort without a closed loop and by screaming you are doomed over the phone I spent most of my day getting screamed at and having the first time I ever heard F-bombs dropped by congress critters because they're trying desperately to help make sure their peers understand that white hat research is vitally necessary for public safety and they're trying to help us fight and there's a new one that's called ICPA which none of you have even heard of which is the International Crime Prevention Act and there's the Wastner stuff and there's also the executive order from April 1st these five things are an existential threat to researchers writ large if they're done wrong and these guys who have been advocating for a year and a half for people like Jen Ellis and myself and Tray Ford and Space Rogue they said you're making impossible for us to help you when you do stuff that companies have a lot of growing up to do but this stunt convinced us that you do too so while we're getting a lot of the results we have to make sure we do it in a way that doesn't deploy antibodies and the truth is we're never going to be able to control the choices of every given researcher but I hope this group doesn't say this is a success story and the new normal because it's going to take that helping hand not just the pointing finger and what kills their coordinated disclosure policies through the cavalry during this week and they got it canceled because their lawyers saw the G-PAC and saw the responses and they like see researchers can't be trusted so if any one of you ever get sued by one of those two companies because they decided not to do their coordinated disclosure policy that's one of the collateral pieces of damage from this sensationalism and third we actually had to cancel that too so those are three concrete setbacks from too much too extreme response that scared the more conservative executives at these companies they're great researchers and great security teams most of these companies they're not all the same level of great but because of that fear factor it at least set us back for six months maybe longer and if one of those bad laws passes if you guys it's the computer information sharing act it's basically the information sharing act in the senate it's likely going to pass and somebody when they saw this video said maybe we should squeeze ICPA in which is the international crime prevention act which if it's badly worded it won't even go to debate it's just going to pass as an amendment to the CISA thing because they were scared by the news about the G-PAC so I don't want fear based response we put a lot of scaffolding in place so we can have you know it's going to take a little bit of everybody's different talent pool and I'm really appreciative for the fact that they've now changed the dialogue but we have to make sure we're not scaring people so badly that they come after the researchers because then we're not going to be safer sooner long answer to your question but I kind of had to get off my chest yeah well one thing that's very for the camera what do you say to people when they say they can't do this and a year goes we can be safer sooner together if you didn't see it the fact that there was a recall here's the good part about the sensationalism because the toothpaste is out of the tubes let's just make the most of it that recall sent a message to all the other car companies that if you have a remotely exploitable vulnerability you're going to have to go through a very expensive recall and they know it's going to lead to a board level discussion and that might accelerate the making better future choices prior to that recall Craig and I spent time with different executives in Detroit and they said well you know it's going to be by 2020 before we can do this thing and we said no one's forcing you to put the 4G LTE application was more theoretical until the recall the second thing I want to point out just so I can give some kudos to Billy Rios and some of the cavalry work what didn't get nearly as much press is the hospice drug infusion pump that Billy found a remotely exploitable flying we had been told for years that the FDA can't act on a cyber recall until there's proof of harm until there's no and Billy didn't do any testing on a highway he did testing on devices in his kitchen there was no one put in danger there was no sensationalism he got a similar effect what gives me hope is if the FDA is willing to be proactive and have a financial and monetary impact this sent a message to every individual so you can't use the current model and you can't buy their place until December guess who's going to lose a lot of business and it's not meant to be punitive but because those two events now have board level attention that you have to take research seriously this might be the next stage of this evolution so we're currently prone maybe this but we're still in that period where it's going to be pretty bad for a little while my hope is that you don't see ideological adversaries or criminals taking advantage of this sooner than we can actually do something about it but if it does come to that those cars will be pulled from the market it'll be very costly but you'll have a much more egregious response it'll have three security three privacy things it did get better the original draft had a line in it said any wireless capability must be subjected to a pen test and I said okay why just wireless you know there's ODB2 there's USB there's the app stores there's a whole bunch of remote attack services there's progressive dongles and they're not very comprehensive different people find different things I said according to which standard there isn't a pen test standard by whom there isn't a certification for pen testers and what do you do with the residual findings who's going to absorb that liability when you have known findings that cannot be fixed in that make a model so they did in fact make that better with some of that feedback and it could end up becoming a PCI for cars which would be a terrible terrible mistake whereas the House Energy and Commerce Committee has put out a letter and they took a lot more briefings from a lot more security researchers including the five-star and that letter is also public and I would love to maybe aggregate through a project or a slack here's what we like about these survey questions here's where we would like clarification and we can formally submit an expert commentary on the parts that are more likely to cause things to be safer sooner and the things that may be a distraction decoy or create something that will take five or six years to undo so if anyone's interested in that we have lots of working groups and we do aggressive work on the hill and pull more people in the fold because if they can hear a voice of reason and technical literacy from this crowd it becomes an antidote and a counterpoint to the fud or the distractions of some of the lobbyists who want to keep things the same way they are so I got a late start but I'm going to make sure we stay on track for the time but I think the talent in this room is going to be making difference because they have and this is not a platitude just a small handful of us have tangible results in specific car companies who are now adding coordinated disclosure policies or hiring security researchers or who have announced like forward announced they are going to have a segment of recalls so it was actually their finance guys that finally understood why to do it not so much their cyber guys but this kind of approach is having concrete tangible results and just go talk to the Tesla guys at how awesome they were they spent they had their entire engineering team in our besides track all day on Tuesday and are highly engaged not only do they have a coordinated age three of their coordinated disclosure policy and we can make fun of how little it was at first or we can encourage it and then you can crawl a walk and run but there's quite a few bright spots and I think the more help we get from more of you the more likely it will be safer sooner so thank you