[00:07.030 --> 00:07.470]  Okay.
[00:07.470 --> 00:11.570]  Depending on what networks you use, wherever you are, you may need additional security.
[00:12.450 --> 00:16.610]  Inside the U.S., assuming you're U.S.-based, you can't always trust where you're going.
[00:16.970 --> 00:24.530]  Outside the U.S., whether it's something as simple as just checking a news report or something a lot more sensitive, you always want to make sure that you have optimum security.
[00:25.290 --> 00:30.070]  Sometimes just to protect from something embarrassing, other times it can be a much more serious situation.
[00:30.070 --> 00:37.550]  Our next two speakers are here to discuss using travel routers to increase your security on the road.
[00:38.170 --> 00:40.050]  And here they are, Ryan Lackey, Marc Rogers.
[00:47.030 --> 00:47.470]  Hi.
[00:47.650 --> 00:48.150]  Thank you.
[00:48.250 --> 00:49.130]  I'm Ryan Lackey.
[00:49.530 --> 00:51.230]  I've worked in crypto for a while.
[00:51.570 --> 00:59.190]  I recently sold my company to Cloudflare, and I've been working on some interesting security stuff since back when I was at Sealand and Haven Co. doing this offshore data hidden thing a while ago, about 15 years ago now.
[01:02.830 --> 01:07.790]  And I'm Marc Rogers, and I break stuff.
[01:08.870 --> 01:09.310]  Cool.
[01:09.970 --> 01:13.270]  So we're going to talk about using travel routers to hide your traffic and safety.
[01:14.070 --> 01:19.450]  We're going to go over basically why you want to hide your information, some principles of OPSEC.
[01:19.570 --> 01:24.870]  We work a lot with the Gruk, who's really big into applying OPSEC to a hacker tradecraft.
[01:25.330 --> 01:36.390]  We're going to show some pretty serious OPSEC fails and people who got burned by them, some of what the technical threats are that you can face, things that can be countered by OPSEC, things that require more technical countermeasures.
[01:36.630 --> 01:39.150]  Some of the existing tools and countermeasures to those tools.
[01:39.810 --> 01:43.230]  And then we're going to introduce travel routers and how to use those travel routers to hide.
[01:45.510 --> 01:49.530]  So probably don't need to really put too much time into it, but why do you really want to hide?
[01:50.310 --> 01:51.290]  There's a lot of reasons.
[01:52.230 --> 01:59.690]  We've seen the last couple, maybe 13 months, it's been very, very clear that there are global monitoring systems in place.
[02:01.170 --> 02:04.210]  NSA, GCHQ, but don't forget about the old threats.
[02:04.650 --> 02:08.050]  China, Russia, all sorts of other people monitoring a lot of global traffic.
[02:08.370 --> 02:09.450]  You don't want to get flagged.
[02:09.490 --> 02:10.210]  You don't want to be special.
[02:10.410 --> 02:11.650]  Like you really don't want special treatment.
[02:11.790 --> 02:15.550]  You want to be completely a normal person that nobody really takes interest in.
[02:15.990 --> 02:18.110]  Because the interest they take in you is not usually very good.
[02:18.450 --> 02:20.670]  And there's good reasons just to care about privacy.
[02:20.850 --> 02:23.670]  Like you don't really want your personal innate desire for privacy.
[02:24.610 --> 02:26.990]  You're not necessarily somebody who's doing anything really illegal.
[02:26.990 --> 02:28.770]  You're just sort of this normal person.
[02:29.130 --> 02:32.890]  And you've got to prevent some basic threats.
[02:33.030 --> 02:46.050]  You want to prevent... there's a principle that if you have a lot of individual things that are themselves not terribly sensitive, but you put them all together in one place in a big database, that collection of stuff is much more sensitive than any of the individual things.
[02:46.970 --> 02:58.390]  And to be honest, there are times when you're doing stuff where you really don't want to be monitored in specific, where you're doing something that might be illegal under some people's laws, might be high risk, might get you in trouble with people that are very powerful.
[02:59.050 --> 03:02.290]  You want to bypass some things that are totally arbitrary that you don't believe in.
[03:03.230 --> 03:04.250]  You don't want to get detected.
[03:04.570 --> 03:10.130]  And you want to make sure that if you do something and you don't get caught while you're doing it, it doesn't come back to burn you like 20 years later.
[03:10.750 --> 03:12.850]  So there's a whole lot of reasons why you might want to hide.
[03:13.830 --> 03:20.630]  So as I said, we work a lot with the Gruk, this guy who lives in Thailand right now and is pretty awesome actually.
[03:20.910 --> 03:32.730]  He's studied OPSEC, he studies KGB, he studies all sorts of other people's sort of like operational tradecraft principles starting from pretty much like the whole Cold War up to now.
[03:33.230 --> 03:36.990]  And he also is a fan of Biggie Small's Ten Crack Commandments.
[03:37.090 --> 03:39.330]  And he converted it to the Ten Commandments of OPSEC.
[03:39.430 --> 03:40.470]  And they're here for you.
[03:40.550 --> 03:41.070]  They're pretty clear.
[03:42.090 --> 03:43.150]  They make a lot of sense.
[03:43.650 --> 03:44.350]  So you should follow these.
[03:46.210 --> 03:48.550]  And then the big question is what happens when this goes wrong?
[03:48.730 --> 03:51.070]  This is a sort of benign case of when it goes wrong.
[03:51.410 --> 03:57.610]  I don't know if anyone remembers back in 1992, the U.S. invaded a country in the CENTCOM area.
[03:58.030 --> 03:59.210]  It turned out it was not Iraq.
[03:59.390 --> 04:00.770]  It was Somalia in this case.
[04:01.290 --> 04:02.710]  And they sent the Marines in to invade.
[04:02.930 --> 04:07.810]  But apparently the journalists knew that this invasion was going to happen on this beach at a certain time.
[04:07.810 --> 04:12.810]  So there were like 200 journalists on the beach with cameras and rigs and everything and tents all set up.
[04:13.170 --> 04:16.310]  And then you've got these Marines doing effectively an opposed landing into a bunch of press.
[04:16.830 --> 04:18.690]  So it's pretty hilarious to watch this whole thing.
[04:20.030 --> 04:20.430]  Yeah.
[04:20.670 --> 04:24.310]  So that's like the least bad thing that happens when this kind of stuff goes wrong.
[04:25.750 --> 04:27.490]  And Marc will go into some of these.
[04:30.070 --> 04:37.830]  Most of the big busts that we've seen over the last 12, 24 months have been because of basic OPSEC failures.
[04:39.190 --> 04:47.630]  Silk Road, arguably the biggest drugs bazaar, probably the first major online drugs bazaar.
[04:48.310 --> 04:52.110]  $1.2 billion in transactions over two years.
[04:53.230 --> 04:58.210]  When you look at why Ross Ulrich was busted, it's pretty obvious.
[04:58.590 --> 05:04.190]  He used his own Gmail when he set up his first account, an account under the name Altoid.
[05:04.470 --> 05:10.130]  He posted jobs for Silk Road to employ people to build stuff under that account.
[05:10.990 --> 05:15.030]  He advertised an early version of Silk Road under that account.
[05:16.610 --> 05:20.910]  He obviously realized at some point that he needed to change things up and that there was a risk.
[05:21.070 --> 05:26.110]  So he then went and retrospectively changed that account to make it anonymous.
[05:26.550 --> 05:29.190]  He changed it to frosty at frosty.com.
[05:29.430 --> 05:35.210]  Frosty at frosty.com then went on to become an admin at Silk Road and was found in the Silk Road keys.
[05:35.210 --> 05:42.890]  He was also seen to post snippets of code to Stack Overflow, asking questions.
[05:43.110 --> 05:45.710]  Code that was later found in Silk Road.
[05:46.890 --> 05:50.010]  All of this stuff is really basic.
[05:51.110 --> 05:53.510]  Don't post things using your own name.
[05:54.510 --> 05:56.270]  Don't use your own email address.
[05:57.030 --> 05:58.250]  Don't use your home address.
[05:58.530 --> 06:03.890]  And once you've posted something on the Internet, don't think that you can go and change it and delete it.
[06:03.890 --> 06:07.270]  Because once it's on the Internet, it's out there.
[06:07.830 --> 06:09.190]  That's not how the Internet works.
[06:12.720 --> 06:14.180]  Same thing with Sabu.
[06:14.960 --> 06:18.000]  Sabu was busted just to two OPSEC failures.
[06:18.880 --> 06:19.860]  Sabu was careful.
[06:20.100 --> 06:22.520]  Sabu spent a lot of his time using Tor.
[06:23.100 --> 06:25.280]  He also spent a lot of his time using IRC.
[06:26.400 --> 06:28.180]  One time he forgot to use Tor.
[06:28.480 --> 06:32.840]  He logged on to IRC using his own details, his home address.
[06:33.500 --> 06:34.560]  That was enough.
[06:35.400 --> 06:40.740]  In today's world, there are people wholesale collecting data just in case.
[06:41.040 --> 06:43.820]  The governments are storing vast amounts of information.
[06:45.080 --> 06:47.420]  And they can retrospectively mine it.
[06:47.540 --> 06:52.820]  So that one time that you log in using your own address is going to be there.
[06:53.080 --> 06:56.560]  And as they search back through history, they'll link it.
[06:57.280 --> 06:59.320]  The other failure was even more basic.
[06:59.700 --> 07:01.300]  If you're going to buy stolen...
[07:01.300 --> 07:07.360]  If you're going to buy parts for cars using stolen credit cards, don't have them shipped to your home.
[07:10.420 --> 07:11.700]  LulzSec is the same deal.
[07:12.160 --> 07:15.380]  They use personal Facebook accounts to send defacement code.
[07:15.640 --> 07:18.200]  They use their real names as their usernames.
[07:18.960 --> 07:21.320]  They never compartmentalize their stuff.
[07:22.680 --> 07:25.120]  You may be a really cool lead hacker.
[07:25.780 --> 07:32.460]  But if you go and boast that you're a cool lead hacker using your personal account, someone's going to link the two.
[07:34.180 --> 07:35.780]  Separate business and pleasure.
[07:36.080 --> 07:38.740]  Separate your illegal activities from your legal activities.
[07:39.980 --> 07:43.100]  And finally, the jury's kind of out on this last one.
[07:43.220 --> 07:43.940]  Mark Carpellis.
[07:44.400 --> 07:48.760]  I know if anyone here thinks he's innocent, whether he's guilty.
[07:48.760 --> 07:53.040]  But what I do know is when he posted saying that Mt.
[07:53.100 --> 08:07.900]  Gox had been hacked, that it had been taken down using the Bitcoin malleability attack, and that he only had 2,000 Bitcoins left, I and a couple of friends did some very rudimentary tracing using some accounts he'd used before.
[08:07.900 --> 08:10.420]  I know if any of you remember when Mt.
[08:10.500 --> 08:12.600]  Gox had its solvency issues the year before.
[08:13.080 --> 08:17.700]  But Mark Carpellis moved around some cash to show that he was actually solvent.
[08:18.280 --> 08:27.020]  And what we did was we went back and we looked at the accounts that he'd moved that cash around to, and looked to see what happened to that cash.
[08:27.320 --> 08:31.200]  And what we found was it had been split and moved in some other accounts.
[08:32.120 --> 08:35.760]  And those other accounts held 90,000 Bitcoins.
[08:36.960 --> 08:41.120]  That's quite a bit larger than the 2,000 Bitcoins he was claiming that he had left.
[08:41.760 --> 08:49.520]  And lo and behold, when that accusation was put to him, he suddenly found another 200,000 Bitcoins that he had forgotten about in a wallet.
[08:51.140 --> 08:56.520]  If you're gonna steal money from people, at least be smart and try to launder it.
[08:59.780 --> 09:04.260]  So what are the common mistakes and vulnerabilities that we see that are sort of in common from all these things?
[09:04.640 --> 09:07.100]  I mean, of course, the big threat is always insiders.
[09:07.500 --> 09:10.740]  Either there's infiltration of your organization or somebody gets turned.
[09:10.960 --> 09:18.480]  You bust somebody for some minor thing and then you tell them, oh, you're gonna go away for two or three or four times longer than you could ever possibly stand.
[09:19.680 --> 09:21.580]  And they roll over pretty fast.
[09:22.060 --> 09:23.380]  Sometimes people just make mistakes.
[09:23.540 --> 09:27.520]  The problem with OPSEC is you have to get it right pretty much all the time perfectly.
[09:27.820 --> 09:34.440]  And you make a minor mistake when you're up late at night or when you've done a lot of stuff or something complicated and it's all gone.
[09:34.600 --> 09:37.100]  It's an all or nothing thing and you can lose it all in an instant.
[09:37.100 --> 09:38.500]  And then there's data leakage.
[09:38.740 --> 09:47.480]  Just the problem is people are collecting and organizations are collecting such huge volumes of data that it's no longer a case of they have to be looking just at the right time when you mess up.
[09:47.700 --> 09:48.580]  They've got everything.
[09:48.820 --> 09:52.400]  So if you mess up one time, even if they don't catch it then, they'll look and they'll find it.
[09:53.880 --> 09:58.620]  So there's also the... one of the threats is that people are gonna attack your equipment itself.
[09:58.760 --> 10:00.360]  They're going to investigate you.
[10:00.460 --> 10:03.220]  They'll get a hold of your equipment or they'll get remote access to equipment.
[10:03.440 --> 10:05.780]  They will look at data that's already on the machine.
[10:05.780 --> 10:06.920]  They'll see if you're using Tor.
[10:07.080 --> 10:08.980]  They'll see if there's any sort of temporary files there.
[10:09.080 --> 10:09.840]  All sorts of stuff like that.
[10:10.180 --> 10:15.120]  There's the fact that there are these global pervasive passive adversaries that are monitoring every single thing.
[10:15.280 --> 10:17.360]  So they can look at backbones.
[10:17.480 --> 10:18.680]  They can look at your local traffic.
[10:18.900 --> 10:21.020]  They can look at the traffic on the remote side.
[10:21.220 --> 10:27.680]  The fact that you might be connecting to servers that are operated by people that are in complete cahoots with whoever your enemy is.
[10:27.860 --> 10:37.820]  If you are connecting to a server in some foreign country, the government in that country Siri is totally on board with the company, is much more likely to side with the organization and the server than you.
[10:37.980 --> 10:49.300]  So if you're sort of like some Internet freedom fighter that's going after some infrastructure in, say, Syria, the Syrian government is pretty much Syrian government, Syrian telco, all those guys are all pretty much one organization.
[10:51.120 --> 10:55.980]  And you might build a threat model around resisting passive threats that works pretty well for you.
[10:56.160 --> 11:02.920]  The problem is there's organizations like NSA that are willing to do a whole range of active plus passive attacks.
[11:03.140 --> 11:12.160]  There's things, the whole quantum series of things that were in the TAO playbook are a combination of a very small active attack combined with a passive attack and passive monitoring.
[11:12.780 --> 11:13.460]  There's malware.
[11:13.640 --> 11:24.740]  There was the attack where the Tor onion router site, the Tor hidden site was compromised, was serving malware that would attack specific versions of Mozilla in the Tor browser bundle of people.
[11:25.180 --> 11:28.000]  And there's the fact that they go after the security controls first.
[11:28.180 --> 11:31.040]  They remove your protection first and then they can unmask you.
[11:31.240 --> 11:33.300]  And then there's, of course, the physical and financial audit trails.
[11:33.440 --> 11:39.540]  All those Bitcoin transaction records, Bitcoin being much more anonymous than a lot of the conventional financial systems.
[11:39.780 --> 11:44.100]  So if you touch money at all, there's usually an audit trail and it's a very strong audit trail.
[11:49.460 --> 11:51.100]  So what are they looking for?
[11:51.800 --> 11:56.920]  When someone gets hold of your equipment, the first thing they're going to look for is the low-hanging fruit.
[11:57.160 --> 11:59.560]  They're looking for network logs, content, metadata.
[12:01.780 --> 12:05.640]  The reason it's seen as a low-hanging fruit is it's really easy to collect.
[12:07.060 --> 12:10.360]  You're generating metadata every time you use your machine.
[12:10.700 --> 12:13.720]  It's being stored on your phones, it's being stored on your laptops.
[12:14.280 --> 12:15.820]  It's data about data.
[12:16.140 --> 12:19.020]  And it's seen as low-risk.
[12:19.780 --> 12:23.160]  Who cares how many gigabytes of data you send as an email?
[12:23.180 --> 12:25.600]  Who cares about how many emails you sent?
[12:25.780 --> 12:31.940]  Who cares about the times that you use your phone or the times you access certain accounts?
[12:32.380 --> 12:33.640]  The answer is they do.
[12:34.980 --> 12:42.180]  It may be quite surprising to you, but the vast majority of metadata isn't encrypted, even when you think your device is.
[12:42.420 --> 12:49.780]  There's been a lot of talk, for example, about the iPhone and how easy it is for law enforcement to acquire forensic information off the iPhone.
[12:51.120 --> 12:57.960]  Most people have speculated things like, you know, there's a floor in Apple's crypto or Apple has a back door.
[12:58.280 --> 13:02.540]  The reality is Apple has different levels of storage.
[13:02.800 --> 13:04.440]  Different things are stored in different ways.
[13:04.620 --> 13:11.580]  And some stuff is stored plain text, even when the device is encrypted, because they don't see the need to encrypt it.
[13:11.900 --> 13:14.140]  And this metadata falls into that category.
[13:14.600 --> 13:19.320]  You can access it with physical access to the device and some special tricks from Apple.
[13:20.500 --> 13:21.260]  No sweat.
[13:21.440 --> 13:22.760]  You don't need to break any crypto.
[13:22.980 --> 13:23.820]  You can just pull it down.
[13:26.300 --> 13:27.800]  What can you collect from systems?
[13:28.080 --> 13:30.020]  Well, the list here is huge.
[13:32.580 --> 13:33.980]  The list here is huge.
[13:34.040 --> 13:40.160]  And I think one of the things that should stand out is the vast majority of it is metadata that's not encrypted.
[13:40.400 --> 13:41.900]  It's not protected in any way.
[13:42.440 --> 13:43.340]  It's just there.
[13:44.460 --> 13:47.020]  That doesn't mean to say that you can't control it.
[13:47.180 --> 13:55.640]  And that's kind of the part of some of this talk, is that you can control where your metadata is, if not the creation of your metadata.
[13:55.900 --> 13:59.420]  And if you can control where it is, you can limit people's access to it.
[14:01.780 --> 14:04.320]  Cell phones are even worse in many regards.
[14:05.440 --> 14:10.420]  The amount of data that you can get just from the SIM card in your phone is pretty staggering.
[14:11.320 --> 14:14.940]  With access to a SIM card, I can tell you your phone number.
[14:14.940 --> 14:16.720]  I can tell you which network you're using.
[14:16.880 --> 14:18.180]  I can tell you who you called.
[14:18.380 --> 14:20.080]  I can get location information.
[14:20.360 --> 14:21.840]  I can get ciphering information.
[14:22.820 --> 14:32.160]  And depending on the SIM card, if it's one that's vulnerable or if you're from an operator that doesn't have strong crypto, I can break it and I can clone it.
[14:36.840 --> 14:39.900]  So we're basically in the golden age of signals intelligence.
[14:40.220 --> 14:44.840]  It used to be this really hard thing that was only directed against foreign governments and foreign militaries.
[14:44.980 --> 14:50.080]  Now it's kind of something you can you can do yourself and or that's so easy for governments to do.
[14:50.180 --> 14:54.620]  The marginal cost of each additional target is so low that they can do it pretty much to everybody.
[14:55.840 --> 14:57.320]  SSL still isn't pervasive.
[14:57.600 --> 15:02.760]  I'm actually working on some stuff personally with some people to get SSL much more widely deployed.
[15:04.420 --> 15:07.060]  There's still like single digit percentages.
[15:07.400 --> 15:08.840]  Let this sink in.
[15:09.940 --> 15:17.680]  Over the last 12 months since Snowden's revelations, encrypted traffic has tripled, if not quadrupled in volume.
[15:18.320 --> 15:21.960]  It's gone up from one to two percent to six percent.
[15:23.940 --> 15:27.180]  More than 90 percent of the world's traffic is unencrypted.
[15:29.540 --> 15:29.940]  Why?
[15:30.980 --> 15:36.780]  And then big data makes once you you have all this data in one place, once you've gathered it, it used to be infeasible to search.
[15:36.960 --> 15:39.880]  Now it's kind of something people do just to target ads better.
[15:40.120 --> 15:43.300]  So it's easy for the government to do with even more money and less profit and so on.
[15:47.320 --> 15:53.400]  So there's some specific information that's interesting about monitoring IP networks, which is something I care a lot about.
[15:53.940 --> 15:56.560]  IP networks are great.
[15:56.680 --> 16:01.680]  You use them all the time, but even if the data is encrypted that you're sending, you still have to leave a lot of metadata.
[16:01.820 --> 16:04.120]  There's some essential routing information that you just can't hide.
[16:04.340 --> 16:10.780]  There's the IP addresses, source destination, ports, the sizes and patterns of your flows, traffic type.
[16:10.980 --> 16:14.820]  All this stuff is essentially like part of the addressing information.
[16:14.960 --> 16:16.320]  You can't get rid of it or it'll stop working.
[16:16.500 --> 16:24.740]  And then the content, even if it's unencrypted, there's a lot of information you can gather just from the patterns of encrypted data versus the contents of the data themselves.
[16:30.180 --> 16:34.660]  Obviously, if you have access to cell phone equipment, you're going to get even more.
[16:35.360 --> 16:41.020]  As somebody who has access to the operator network, you can get pretty much the keys to the kingdom.
[16:41.240 --> 16:44.400]  You have call time, duration, call and call party.
[16:44.700 --> 16:53.320]  You have a rough location, not necessarily precise unless it's transmitted, although that is accessible with the operator's cooperation.
[16:53.780 --> 16:56.240]  You have data traffic, SMS traffic and voice calls.
[16:56.780 --> 16:58.620]  Over the air, it's a little bit trickier.
[17:00.420 --> 17:01.580]  You can get the IMSI.
[17:01.980 --> 17:03.380]  Everyone's heard about IMSI catchers.
[17:03.460 --> 17:04.960]  They're being deployed everywhere at the moment.
[17:05.880 --> 17:09.440]  What most people don't realize is just how sensitive the IMSI is.
[17:10.120 --> 17:13.360]  The government treats the IMSI as unsensitive metadata.
[17:13.600 --> 17:15.400]  You don't need a warrant to seize it.
[17:15.780 --> 17:18.240]  Stingray devices are being deployed all over the place.
[17:18.880 --> 17:23.940]  The reality is the IMSI is a unique identifier that's attached to you as a cellular subscriber.
[17:24.480 --> 17:26.620]  With the IMSI, I can track your phone.
[17:27.460 --> 17:29.520]  And it doesn't take a lot of equipment to do that.
[17:29.960 --> 17:37.100]  A colleague of mine, Kristen Padgett, demonstrated at Defcon that just for a few hundred dollars, you can build an IMSI catcher to track people.
[17:37.440 --> 17:43.120]  The reality is it's actually got even cheaper because now we have these things called femtocells.
[17:43.280 --> 17:47.220]  And what a femtocell is, is a cell-based station in a small box.
[17:47.380 --> 17:50.840]  It's the Wi-Fi router equivalent of the telco network.
[17:51.560 --> 17:54.200]  And these things aren't secure.
[17:54.400 --> 17:56.580]  Well, some of them are secure, but they're not that secure.
[17:56.920 --> 18:02.160]  And if you route a femtocell or if you modify it in any way, you have a cheap IMSI catcher.
[18:02.420 --> 18:05.660]  In fact, more than that, you have an active base station attack.
[18:06.020 --> 18:09.120]  Something that previously you could only buy for $200,000.
[18:09.660 --> 18:12.540]  Now you can buy at Walmart for about $100.
[18:14.760 --> 18:17.040]  What can you get with a man in the middle?
[18:17.160 --> 18:19.600]  Well, you get full precise location.
[18:19.640 --> 18:25.300]  You get data, you get voice, you get SMS traffic, and you get to substitute traffic with your own.
[18:25.400 --> 18:29.700]  That means you can pass malicious payloads to the browsing device.
[18:29.900 --> 18:31.600]  It means you can see what the user's doing.
[18:31.780 --> 18:33.500]  You can listen to what the user's saying.
[18:34.660 --> 18:36.300]  Pretty much as much as the operator.
[18:36.420 --> 18:43.500]  In fact, I would argue more than the operator because you have the opportunity to do stuff that the operator wouldn't do because of legal constraints.
[18:48.860 --> 18:56.040]  A couple of years ago, I went out to China to take a look at the deal of what's going on there and exactly what the Great Firewall of China is like.
[18:57.440 --> 19:01.100]  The Great Firewall of China is kind of a poor term because it's not just one system.
[19:01.260 --> 19:07.580]  It's a very distributed network of machines that interact with everybody and anyone who's in China.
[19:07.580 --> 19:13.880]  It does everything from blocking IP addresses, hijacking DNS requests, not even ones in China.
[19:14.180 --> 19:19.180]  If you come too close to Chinese IP space, they will substitute some of your requests.
[19:19.340 --> 19:20.760]  It filters by keyword.
[19:21.040 --> 19:25.380]  It does deep packet inspection for plain text looking for forbidden words or forbidden strings.
[19:25.800 --> 19:31.980]  And most recently, in fact, in 2011, it developed the ability to look for Tor.
[19:32.580 --> 19:40.460]  And what they did was they built a a signature detection system that looked for the handshake, the ciphering handshake that Tor did.
[19:40.560 --> 19:55.860]  And if they detected it, it then actively scanned the IP address they suspected and spoke Tor to it and set up a one-hop onion circuit to validate that this was truly a Tor bridge or a Tor node and then blocked it.
[19:57.200 --> 20:03.300]  That's pretty advanced stuff and a major setback for people who love privacy like me.
[20:06.580 --> 20:12.900]  So the good thing is you don't have to rely on having perfect OPSEC 24-7, 365 to defend yourself.
[20:13.040 --> 20:15.560]  There are actually some tools out there that make things a lot easier.
[20:15.680 --> 20:16.520]  Oh, we're going to have the slides on.
[20:16.660 --> 20:18.160]  You don't need to... we'll have the slides up.
[20:18.780 --> 20:22.460]  Um, so we're going to have... there's a bunch of tools that will help the user protect themselves.
[20:22.900 --> 20:29.180]  Some of these things are just regular tools that you happen to use versus other tools that are just safer to use.
[20:29.380 --> 20:36.520]  There's other tools that are not really so safe in their default configuration, but we can tweak them with some new settings to make them a lot safer.
[20:36.940 --> 20:47.200]  And then there's a bunch of tools that people have been working on really for the past maybe 15 years, but especially in the past 5 or 10 years, to sort of intentionally hide information and provide privacy.
[20:47.400 --> 20:50.600]  Not merely hide content, but also provide anonymity.
[20:51.200 --> 20:52.460]  Um, there's some great things.
[20:52.560 --> 20:56.280]  Tor is probably the most successful to date, and there's a lot of other good work.
[20:56.420 --> 20:58.740]  And then we're introducing some new tools that are software and hardware.
[20:59.840 --> 21:03.740]  Uh, to start with, there's the tools that are a little bit more privacy-protecting by default.
[21:04.020 --> 21:12.380]  There's some basic characteristics that will make a tool, I would say, generally more privacy-protecting, or at least amenable to privacy, than other tools.
[21:12.740 --> 21:20.860]  Message-based systems like email versus chat... or email versus live stream, things like that, are just generally much easier to secure.
[21:20.980 --> 21:21.760]  You can add latency.
[21:22.060 --> 21:31.980]  One of the big attacks on all these systems is you can correlate when someone's online, when they're using the traffic, and when their traffic is spiking on their local network link, with a remote node.
[21:32.200 --> 21:36.680]  If you send email, there's less of a direct correlation, especially if it's got intermediate systems.
[21:36.780 --> 21:38.660]  If you've got a peer-to-peer system, it's decentralized.
[21:39.460 --> 21:45.540]  You don't have a single organization that has everybody's records to go after legally, or extra-legally, or illegally.
[21:46.640 --> 21:50.960]  And it's also why bitcoin is so awesome compared to a lot of the existing financial systems.
[21:51.340 --> 21:55.300]  Systems that just, even if they're not totally secure, that just sort of passively encrypt everything.
[21:55.560 --> 22:00.940]  Like, one of my favorite things is how a lot of Internet email finally, in 2014, is encrypted.
[22:01.340 --> 22:05.900]  Not because of PGP, which is an awesome message-based crypto system, but because of StartTLS.
[22:06.120 --> 22:10.440]  People just set up their mail servers to passively encrypt traffic that goes to other mail servers.
[22:10.660 --> 22:14.000]  Gmail does it for virtually everybody they talk to, except, I guess, for Yahoo still.
[22:14.920 --> 22:16.800]  Everybody who runs their own mail server should do this.
[22:17.360 --> 22:23.660]  It's not at all secure against an active attacker, because there's really no PKI for it or anything, but against a passive adversary, it's great.
[22:23.900 --> 22:28.500]  So that's an example of where just sort of pervasive, passive, kind of weak encryption helps you.
[22:28.680 --> 22:30.640]  And then you can layer better systems on top of it.
[22:30.820 --> 22:36.140]  And then stuff where data and executable code are distinct is really, really important for a lot of security.
[22:36.380 --> 22:43.060]  And a lot of the violations of privacy of user data happen when security is subverted by an active, passive hybrid system.
[22:43.240 --> 22:45.600]  So those are all really important principles in general.
[22:47.620 --> 22:56.060]  One of the things that's actually really awesome that's sort of one of these sort of like accidental benefits that come from something else is that cheap hardware itself helps you a lot.
[22:56.260 --> 23:02.980]  If you have to worry about securing your hardware, so there's a standard for the DoD for like how you can declassify a hard drive.
[23:03.540 --> 23:05.500]  Essentially, it is you physically destroy the hard drive.
[23:05.640 --> 23:13.300]  We've basically completely given up on the ability to take a hard drive, delete data from it securely, and reuse that drive.
[23:13.300 --> 23:14.740]  You effectively run it through a shredder.
[23:15.040 --> 23:22.200]  So the fact that the military has this crazy huge budget of like $800 billion a year lets them do that with top of the line equipment.
[23:22.440 --> 23:33.260]  I could never do that with my personal equipment until now where I can buy a laptop for like $200 and I can dedicate it, use it for a couple of weeks, use it for a couple of months, and then once it's served its purpose, I can get rid of it.
[23:34.920 --> 23:42.020]  Disposable hardware lets you dedicate things to tasks, so you don't have to worry about how you securely handle multiple projects on the same machine at one time.
[23:42.200 --> 23:44.160]  You just buy a couple separate machines and use those.
[23:44.700 --> 23:46.220]  And phones are even cheaper than laptops.
[23:46.520 --> 23:55.440]  You can buy phones, like pretty soon, phones are going to be included for free with like serial that are like unlocked phones, but you can get a phone, a smartphone now for like $50, $100 for a decent phone.
[23:55.720 --> 23:57.740]  So that's getting pretty crazy.
[23:58.020 --> 24:06.380]  And then there's the model of running live systems out of memory where you're not actually writing anything to disk makes it a lot easier to secure stuff.
[24:06.480 --> 24:09.700]  If it's never on the disk, it's never available to anyone else to attack.
[24:12.020 --> 24:14.760]  One of the problems here is, of course, getting access to this hardware.
[24:15.000 --> 24:19.120]  It doesn't do you a whole lot of good if you're leaving like a network monitoring device in somebody's location.
[24:19.280 --> 24:25.380]  If it's got a serial number and you can trace where that thing was purchased with your credit card, shipped to your home address.
[24:25.720 --> 24:28.740]  So you want to be able to buy your hardware in a non-attributable way.
[24:29.420 --> 24:30.900]  Bitcoin is really awesome for this.
[24:31.500 --> 24:35.920]  As we saw with Sabu, there was the problem where he had some stuff shipped to his house.
[24:36.020 --> 24:37.080]  They shouldn't have had shipped to his house.
[24:37.700 --> 24:40.840]  With Bitcoin, you, of course, want to buy it through like local Bitcoins or something.
[24:41.080 --> 24:42.160]  But there's that.
[24:42.340 --> 24:44.160]  There's a myth that Bitcoin is completely anonymous.
[24:44.540 --> 24:45.100]  It isn't.
[24:45.660 --> 24:56.000]  Because the transaction ledger is public, unless you take steps to hide your transactions in the cloud of other things going on, it's actually not too hard to trace what's going on.
[24:56.460 --> 25:09.000]  But the reality is the steps that somebody like Marc Capelli should have taken is things like Bitcoin tumblers, proxies that will hide your transaction in a cloud of other transactions are accessible and available.
[25:09.360 --> 25:11.020]  And they're relatively easy.
[25:11.160 --> 25:14.320]  But you just have to think about what you're doing and understand the implications.
[25:14.800 --> 25:26.060]  Because if you run out after drawing $500 from your cash point, from your ATM rather, and go and buy yourself a phone from AT&T, don't have any illusions that it's completely anonymous.
[25:26.440 --> 25:29.540]  It's going to take them no more than a couple of days to trace that phone to you.
[25:31.380 --> 25:34.480]  And then the other... the third part of this is the disposable accounts.
[25:34.680 --> 25:37.200]  So you dedicate an account to a single task.
[25:37.440 --> 25:43.500]  You don't run your personal email and use that to contact people for your hacking thing.
[25:43.840 --> 25:44.820]  You've changed systems.
[25:44.980 --> 25:45.880]  You do all sorts of stuff like that.
[25:46.000 --> 25:48.340]  This is basically good OPSEC principle.
[25:48.480 --> 25:50.960]  The problem is it's a bit of a pain to manage all this kind of stuff.
[25:54.320 --> 25:54.760]  Yeah.
[25:54.760 --> 25:56.280]  There's also VPN technology.
[25:56.440 --> 26:02.500]  So VPNs are kind of near and dear to my heart because I ran a VPN provider for a lot of last year.
[26:02.600 --> 26:10.820]  And we shut it down after the lava bit fiasco because of some security issues that were some legal policy issues that became very apparent.
[26:12.260 --> 26:19.440]  But you've got SSH, you've got a bunch of free VPN services you can use, commercial services that are falling to, like, almost zero dollars in cost.
[26:19.700 --> 26:21.120]  And then you can, of course, run your own VPN.
[26:22.140 --> 26:25.540]  The problem is they're not really terribly good for anonymity.
[26:25.680 --> 26:27.500]  They're really not anonymity at all.
[26:28.360 --> 26:32.380]  You can use them to support an anonymity system, but on their own, they're not at all anonymous.
[26:32.800 --> 26:33.880]  They're not end-to-end.
[26:34.140 --> 26:38.500]  Some of them are run by really sketchy people that like to log all your traffic and do all sorts of bad stuff to you.
[26:39.840 --> 26:40.920]  And a lot of them are blocked.
[26:41.060 --> 26:46.520]  Like, I've been trying to get Chinese users to connect to something, and it's sort of a cat-and-mouse game.
[26:47.160 --> 26:49.600]  And then you can leak data.
[26:49.760 --> 26:55.340]  Like, if you've got a VPN for some of your traffic but not all of your traffic, it's often that traffic that goes around the VPN that's really interesting.
[26:56.880 --> 26:58.620]  Also, same as with the burner phone.
[26:58.840 --> 27:05.280]  Don't fool yourself into thinking that you can buy a VPN service with a credit card and then go and use that VPN service to do illegal things.
[27:05.700 --> 27:08.220]  Because, I mean, come on, really?
[27:09.040 --> 27:14.840]  The same thing is, the more public a service like that is, the less likely that it's going to fly under the radar.
[27:15.140 --> 27:30.800]  So if your purpose is to find a service that's going to allow you to punch through some kind of national firewall, such as the Great Firewall, then public VPN services probably aren't going to help you because their nodes are being shut down almost as fast as they're setting them up.
[27:32.400 --> 27:35.520]  So one of the best tools that's out there today is Tor.
[27:36.920 --> 27:42.220]  Correctly used, it will keep you anonymous, I believe, at least until it hits the exit node.
[27:42.600 --> 27:45.220]  And it has some... it generally works.
[27:45.340 --> 27:45.840]  I use it.
[27:45.940 --> 27:46.700]  It's a great tool.
[27:46.840 --> 27:47.960]  But it's got some disadvantages.
[27:48.340 --> 27:50.160]  It's easy to block if you know the signature.
[27:51.280 --> 27:56.500]  A lot of people are looking at it because it's a really interesting academic project, as well as a useful tool.
[27:56.800 --> 28:01.280]  It's not terribly performant, and it's pretty complex.
[28:01.540 --> 28:08.500]  So there's a lot of potential for error, especially if you run it on your local computer with all of your sensitive data on it, dedicated... not dedicated to the task.
[28:11.220 --> 28:14.760]  DPI and active network analysis are the greatest enemy of it.
[28:15.180 --> 28:21.640]  And luckily, the Tor community has a lot of great developers that are working on ways to mitigate these disadvantages.
[28:22.040 --> 28:31.680]  The best thing is actually to run format transforming encryption to... or pluggable transports to change the way that Tor traffic looks to the outside world as it transits.
[28:31.960 --> 28:35.780]  They've been working on this for, I think, three or four years now, at least.
[28:36.200 --> 28:39.180]  And it's pretty effective, and we'll go into that.
[28:39.340 --> 28:47.400]  So what the Tor project realized is that we're in a world where we have access to tools that can analyze big data.
[28:47.720 --> 28:52.380]  We can now take huge swaths of data and look for signatures, patterns.
[28:52.800 --> 28:57.040]  We can look for ways to attribute the types of services sending these things.
[28:57.240 --> 29:03.360]  And that means that for our traffic to stay hidden in this giant cloud, it has to become a moving target.
[29:03.500 --> 29:15.420]  We have to come up with ways either to hide the tree in the forest, or we have to come up with ways to make sure the tree is constantly changing shape, so that when they look at it, it's a bush, now it's a blade of grass, now it's another tree, now it's a bush,
[29:15.580 --> 29:18.000]  and that way no one can pin it down and track it.
[29:18.180 --> 29:23.120]  And that's where the pluggable transport concept is fantastic, because it's dynamic.
[29:23.280 --> 29:24.660]  It means you can load different modules.
[29:24.960 --> 29:28.400]  Right now, there's seven live pluggable transports that you can use with Tor.
[29:28.720 --> 29:41.560]  There's OBFS Proxy, there's Flash Proxy, Format Transforming Encryption, Scramble Suite, Meek, OBS4, OBFS Client, and there are things like three others which are coming live.
[29:41.840 --> 29:48.400]  OBS4 but is by far the most popular, although Format Transforming Encryption is probably my favorite.
[29:49.500 --> 29:52.260]  OBS4 is basically just an encryption layer on top of Tor TLS.
[29:52.560 --> 29:57.140]  It negotiates the session keys and encrypts everything between the client and server with no plain text headers.
[29:57.860 --> 30:02.680]  By stripping out the plain text headers, you remove the opportunity for signature detection.
[30:02.800 --> 30:08.580]  It's just random data that they see makes it very difficult for them to say, this is anything.
[30:08.740 --> 30:09.680]  It just looks like noise.
[30:12.040 --> 30:19.840]  There's a visual representation here which shows the difference between ordinary Tor traffic and OBS3 traffic.
[30:20.060 --> 30:22.700]  I couldn't find a representation for OBFS4.
[30:25.940 --> 30:33.280]  Essentially, it just adds additional layers onto it, but packages it inside an encrypted tunnel that makes it impossible to see.
[30:34.080 --> 30:38.740]  Scramble Suite is another transform which aims to make it look like uniform random bytes.
[30:39.120 --> 30:43.940]  Similar to OBFS, it offers a few... but it offers a couple of extra features.
[30:44.140 --> 30:46.400]  It randomizes the size and the timing of packets.
[30:47.020 --> 30:52.780]  And the server also resists active probing by requiring a secret key from the client before it will respond.
[30:53.040 --> 30:59.780]  This is an attempt to resist the active probing done by the great firewall and other sensor systems that are coming online.
[31:00.840 --> 31:06.100]  Flash proxy allows the creation and utilization of very sudden short-lived JavaScript-based proxies.
[31:06.180 --> 31:17.240]  They don't do much to try and hide it, but they set it up in a way that you can just create a cloud of browsers and have them just create a flash network very quickly, put it up rapidly.
[31:17.920 --> 31:22.300]  It doesn't matter if it gets found because hopefully you've used it by then and it can pop up another one.
[31:22.520 --> 31:25.680]  And new ones can come up as fast as old ones are taken down.
[31:28.140 --> 31:30.060]  Format transforming encryption is my favorite.
[31:30.740 --> 31:34.580]  This encodes data in a way that it matches an arbitrary regular expression.
[31:34.780 --> 31:42.520]  The idea is they've looked at how most of these censoring systems automatically identify traffic and attribute it to a type of traffic.
[31:42.840 --> 31:48.860]  Most of the time it's just a regular expression that looks for particular byte sequences or looks for particular keywords.
[31:48.860 --> 31:51.820]  And they map it to various things.
[31:52.100 --> 31:59.220]  So the example you can see below is they've encoded the Tor traffic so it looks like SSH traffic.
[32:00.780 --> 32:01.880]  It's not perfect.
[32:02.140 --> 32:06.780]  It doesn't stand up to manual scrutiny, but it does full automatic systems.
[32:07.340 --> 32:13.440]  And at the end of the day, we know that manual inspection of this traffic is too expensive.
[32:13.640 --> 32:15.780]  They can't afford to manually inspect everything.
[32:15.820 --> 32:21.020]  So they rely on their automated systems to flag things of interest and drill into it.
[32:21.220 --> 32:29.460]  So so long as you can do something that keeps you under the radar and avoids the manual manual inspection, it doesn't matter if it doesn't stand up to manual scrutiny.
[32:30.440 --> 32:35.460]  And one of the disadvantages for the attacker is that those automated systems have to run incredibly fast.
[32:35.580 --> 32:42.140]  They're running on 10 gigabit Ethernet links, multiple links, and they can't add too much latency to the system.
[32:42.240 --> 32:43.600]  They have to be reliable and everything else.
[32:43.600 --> 32:47.100]  I've worked on a high-speed proxy that needs to do the same kind of thing.
[32:47.280 --> 32:48.680]  And we worry about microseconds.
[32:48.960 --> 32:53.980]  So you really don't have a lot of processing time to do this kind of pattern matching in IDS.
[32:55.160 --> 32:57.740]  We can... we'll put the rest up in the slides and you can skip through.
[32:58.000 --> 33:00.360]  One last one that I particularly like, Bananaphone.
[33:01.200 --> 33:11.480]  What Bananaphone does is it takes a literary construct, a library of keywords, and it uses that to disguise it as something that resembles natural language.
[33:11.820 --> 33:15.960]  Again, if you look at it, it doesn't stand up to manual scrutiny.
[33:16.200 --> 33:22.880]  But when it comes to automatic detection, it's going to be very hard for the system to flag it.
[33:24.560 --> 33:27.440]  And there's an expansion factor, of course, so there's inefficiency.
[33:27.780 --> 33:32.240]  But again, you've got really fast links now because stuff is so cheap and there's video streamed to your devices.
[33:32.520 --> 33:35.880]  So you can afford a hundred-time expansion factor if you're sending text.
[33:38.860 --> 33:41.720]  So outside of the tour world, there are still a bunch of great tools.
[33:41.940 --> 33:48.820]  There's some that have been around since... I think 1998 was when DNS Tunnel was proposed on BUGTRAQ, which I actually remember.
[33:49.260 --> 33:50.980]  And there's HTTP Tunnel.
[33:51.100 --> 33:54.980]  There's various other systems that effectively tunnel one kind of traffic in another kind of traffic.
[33:55.180 --> 33:57.600]  The idea being that... well, DNS Tunnel has a couple purposes.
[33:57.740 --> 34:00.780]  DNS Tunnel is also so you don't have to pay for Internet at your hotel.
[34:00.900 --> 34:02.720]  You can bypass the authentication thing.
[34:02.820 --> 34:06.840]  But it also means that DNS traffic isn't usually very suspicious, so people just sort of let it go by.
[34:07.180 --> 34:10.600]  And then HTTP traffic is also not usually suspicious, so they just let it go by.
[34:11.200 --> 34:19.180]  The two obvious limitations with both of these is that you have to have servers that speak these tunneling protocols ready and waiting to catch your connection.
[34:19.980 --> 34:23.460]  That means you have to kind of preset everything up and have the infrastructure waiting.
[34:23.640 --> 34:33.120]  And if it's just you setting up some infrastructure for yourself, you can guarantee that it's going to get flagged eventually and it's probably not going to stay around forever.
[34:33.420 --> 34:41.380]  The big advantage of Tor is there is a network of ready proxies growing all the time there to connect to your session.
[34:42.420 --> 34:48.780]  Although it's even worse if you tried to use a commercial provider that ran one IP for DNS Tunnel or something with a bunch of people using it.
[34:48.880 --> 34:51.060]  It would get shut down in a very short time.
[34:52.300 --> 34:55.680]  So the problem we see all here is that there's really no silver bullet.
[34:55.800 --> 34:56.620]  All this stuff is hard.
[34:56.800 --> 34:59.160]  There's a bunch of incremental things that will give you small advantages.
[34:59.160 --> 35:00.180]  They're hard to use.
[35:00.180 --> 35:03.380]  And if you screw up, there's a lot of bad stuff that happens to you.
[35:03.640 --> 35:07.280]  You might have your photo taken on the beach, which is one thing bad to happen to you.
[35:07.660 --> 35:15.640]  But these EMC catchers and IMEI monitoring, um, it's how we target drones to shoot missiles out of the sky to, like, people that are driving around in their cars.
[35:15.980 --> 35:18.240]  So some people really don't want to be located.
[35:18.840 --> 35:21.920]  Um, so it only takes one mistake to screw up.
[35:22.060 --> 35:24.780]  And it's really hard for humans to not make mistakes.
[35:25.080 --> 35:30.140]  It is, however, really easy for computers to only make the same mistake over and over again and not make new mistakes each time.
[35:30.380 --> 35:35.780]  So we think there's an opportunity to use software tools to sort of automate the meta process here for all this stuff.
[35:36.180 --> 35:41.080]  One of the, uh, the ideas behind the concept is we can't solve all of the problems.
[35:42.160 --> 35:49.940]  It's, you're going to end up with a system that's so complex that by the time you've created it, some part of it's going to be obsolete and not worthy.
[35:50.340 --> 36:02.160]  But what we figure is if we aim at the low-hanging fruit, the stuff that there is tools and technology, for example, like Tor, out there to solve, you can solve 50 or 60% of the problem of reducing the metadata.
[36:02.480 --> 36:11.620]  And that allows you to spend the rest of your time and focus all of your energy on the remaining section, which means there's less chance of mistake.
[36:11.860 --> 36:14.360]  You're not going to make a mistake in the bit that's being done automatically.
[36:15.020 --> 36:19.920]  And hopefully, you can put some additional effort into ensuring that the bit you do do is done right.
[36:22.320 --> 36:23.540]  So, we thought about this.
[36:23.600 --> 36:33.280]  I've been working on this since, kind of, uh, October 2010, and thinking about it, talking with Mark, talking with Grok, and a few other people, trying to figure out what the best approach to solve this is.
[36:33.580 --> 36:41.320]  Um, everyone's moving to the cloud for everything, but this is one of those things where you actually can't use the cloud as your only solution, because, uh, what would you connect to?
[36:41.440 --> 36:43.620]  And it's the whole monitoring problem, so it doesn't make any sense.
[36:43.980 --> 36:48.880]  Client software, it's a long, it's like 2014, and it's still a pain to write client software, even for one platform.
[36:49.140 --> 36:59.460]  But if I wrote something that only worked on, like, FreeBSD, people would complain that there's no Linux version, there's no Mac version, no Windows version, and then trying to keep something that you're actively developing in sync on all these platforms is a huge plan.
[36:59.640 --> 37:01.280]  Plus, you want to use it on iPhone and Android.
[37:01.540 --> 37:06.420]  Um, then you have to get approval from this, like, walled garden to be able to deploy software there.
[37:06.980 --> 37:08.540]  PGP is an effective suite.
[37:09.640 --> 37:15.260]  You use it right, it effectively encrypts your encryption, yet its adoption rate is terrible.
[37:16.080 --> 37:18.920]  Most places, the only reason it gets used is because it's mandated.
[37:19.660 --> 37:22.440]  Like, various companies insist that you have to use it as part of the policy.
[37:23.100 --> 37:29.280]  How many of you know people who aren't in our community, who aren't geeks, that use PGP at home?
[37:30.320 --> 37:30.980]  Almost none.
[37:31.380 --> 37:31.760]  Why?
[37:32.200 --> 37:35.980]  Because there's an overhead in getting this thing set up and managing it and using it.
[37:36.540 --> 37:42.680]  I actually, in my day job, work in the cell phone business, and we did a survey not too long ago on a really simple security measure.
[37:42.980 --> 37:45.160]  The pin code on your phone.
[37:45.260 --> 37:48.260]  How many people do you think implement a pin code on their phone?
[37:48.740 --> 37:50.480]  The answer is roughly 50%.
[37:52.020 --> 37:58.520]  And the reason given, the number one reason, is a four-digit pin is too difficult to manage.
[37:59.800 --> 38:04.100]  So there's cool stuff like the Tor Browser Bundle, which sort of has the same problems with client software.
[38:04.440 --> 38:10.140]  And some of the most secure systems we have right now are VM or bootable image systems, like Tails.
[38:10.700 --> 38:11.780]  They work pretty well.
[38:11.980 --> 38:14.960]  It's still a pain to have to run VMs on your system and everything else.
[38:15.120 --> 38:18.000]  You have to use different tools inside the VM, and it's kind of clunky.
[38:18.440 --> 38:20.200]  And it doesn't work on mobile phones or tablets.
[38:20.380 --> 38:24.140]  There still, surprisingly, is no great virtualization system for mobile phones.
[38:24.320 --> 38:27.180]  I'm really surprised by that, but it doesn't exist yet.
[38:27.800 --> 38:30.800]  And the other option is to use dedicated or disposable hardware.
[38:30.960 --> 38:40.140]  So you've got your work laptop, your personal laptop, your secret project laptop, your other malware analysis laptop, and pretty soon you're not going to be, like, carrying the stuff around with you.
[38:40.260 --> 38:44.200]  And what happens if you're out and you only have one of your laptops and you need to use your other laptop to do something?
[38:44.380 --> 38:47.960]  It's just... then they can just find the guy who's carrying five laptops and pick him up.
[38:48.160 --> 38:49.340]  So it's not really so great.
[38:49.760 --> 39:02.320]  So we thought about it, and we thought about what is the small device that you want to have for some other benefits that you carry around with you, or it could carry around with you everywhere, that will actually be able to run all the software, all these tools,
[39:02.600 --> 39:03.540]  on the device.
[39:03.740 --> 39:08.240]  And we thought that a custom-secure travel router is actually the sweet spot for all this stuff.
[39:08.380 --> 39:13.800]  It mitigates a lot of the disadvantages of this stuff and is an easy way to deploy these tools.
[39:16.810 --> 39:18.630]  The reasons travel routers are great?
[39:18.970 --> 39:19.870]  They're super cheap.
[39:20.210 --> 39:27.090]  I remember when travel routers were something that was about the size of a hardback book and it cost a couple hundred dollars.
[39:27.450 --> 39:40.450]  Now you can go and buy little routers from TP-Link for about 20 to 25 dollars, which run Linux and which are flashable and which have all the hardware features that you might want to take advantage of to do something cool.
[39:40.590 --> 39:53.590]  Yeah, people use these when they go to hotels and they want to share their network with their cell phone, their Kindle, their tablet, their wife's computer, their computer, all these five or six devices and they get charged per MAC address by the hotel or per connection.
[39:53.950 --> 39:58.430]  Or if they want to have a cell phone connection, they want to share it with a bunch of devices.
[39:58.610 --> 40:01.970]  So there's actually pretty good reasons to use these things that are totally unrelated to security.
[40:02.150 --> 40:03.530]  So that's why they're so cheap.
[40:03.630 --> 40:04.730]  They're not a security product.
[40:04.930 --> 40:10.210]  They're a save you money on paying for Wi-Fi at hotels and make it easier to connect product.
[40:10.410 --> 40:11.930]  So they're really price sensitive.
[40:12.090 --> 40:17.410]  They're made by these Chinese companies that make them super cheap and they're everywhere.
[40:18.350 --> 40:29.470]  Of course, the other advantage is if you're going to use something as a hardware security product and you're going to take it into hot zones like, for example, China, you don't want a security product that screams security product.
[40:29.970 --> 40:38.250]  Because if you take out something that looks like a spy tool and you set up in your hotel room, you can pretty much guarantee you're going to get a visit from the PLA within a few minutes.
[40:39.550 --> 40:48.970]  Using something that almost every other business traveler has or at least looks like something that every other business traveler has and stands up to some reasonable scrutiny is a massive advantage.
[40:49.430 --> 40:52.590]  And the other advantage for us is that they're actually really easy to program for.
[40:52.910 --> 40:54.130]  They're system-on-chip.
[40:54.230 --> 40:55.250]  They're Linux-based.
[40:55.410 --> 40:56.310]  There's OpenWRT.
[40:56.430 --> 40:57.510]  There's DDWRT.
[40:57.770 --> 40:59.010]  They're really simple to work.
[40:59.070 --> 41:02.150]  They're not as simple as like general purpose computers, but they're still pretty simple to work with.
[41:02.590 --> 41:11.110]  You can't really self-host the development environment, but if you're familiar with Arduinos and with Raspberry Pis and things like that, then it's pretty easy to do all this stuff.
[41:13.850 --> 41:16.390]  So we weren't the only people who ever came up with this idea.
[41:16.510 --> 41:18.770]  There's actually several projects in this space.
[41:19.150 --> 41:25.930]  Of course, there's the PogoPlug, which is sort of like a plug-in wall wart kind of company that used to be for sharing USB drives.
[41:26.230 --> 41:31.390]  They came up with something called the SafePlug, which is a device that's not really portable.
[41:31.550 --> 41:32.930]  It's designed for home network connection.
[41:33.250 --> 41:36.090]  It runs basic Tor, doesn't do pluggable transport.
[41:36.250 --> 41:36.790]  It's pretty basic.
[41:36.910 --> 41:38.150]  I've never actually seen one in person.
[41:38.430 --> 41:43.630]  There's the OnionPie, which Adafruit, which is this awesome New York-based hardware hacker company has.
[41:44.130 --> 41:46.530]  It's really a Raspberry Pi plus a bunch of other stuff.
[41:46.750 --> 41:53.690]  And as far as I can tell, it's as much a learning tool for people to really experiment with the Raspberry Pi to do a practical thing as it is a finished product.
[41:53.870 --> 41:54.710]  But it's a pretty awesome tool.
[41:54.870 --> 41:55.550]  It's very configurable.
[41:55.650 --> 41:56.430]  You should totally check it out.
[41:56.710 --> 42:03.510]  And then there's Portal, which the Gruk, this awesome hacker, came up with, I think 2011 was when he started working on it, maybe a little bit before.
[42:06.030 --> 42:10.250]  Basically, he started out really focused on Tor and pluggable transports.
[42:10.670 --> 42:16.730]  And when I found out this thing existed on the Internet, I mailed him and I met up with him a couple times and talked with him about it.
[42:17.190 --> 42:18.690]  It does a lot of what I want.
[42:18.770 --> 42:20.190]  And it's a lot of the hardware that I want.
[42:20.330 --> 42:21.550]  But it doesn't really have all the tools.
[42:21.750 --> 42:24.110]  So he was limited in his bandwidth to sort of work on the tools.
[42:24.190 --> 42:26.570]  So we worked on it together, which is great.
[42:26.890 --> 42:27.910]  But it didn't do enough.
[42:28.130 --> 42:29.830]  So we hacked it.
[42:29.930 --> 42:30.570]  And now?
[42:33.470 --> 42:35.170]  So we've been working on it for a while.
[42:35.650 --> 42:37.250]  It's basically dedicated hardware.
[42:37.430 --> 42:43.730]  It's a Qualcomm Atheros AR9331 chipset with 64 megs of RAM and 16 megs of ROM.
[42:44.270 --> 42:46.170]  It's a pretty low-end computer.
[42:46.370 --> 42:48.170]  It's a 400 megahertz like MIPS 24K.
[42:48.450 --> 42:50.890]  But for what we need, it's pretty awesome.
[42:51.310 --> 42:52.550]  We built some images for it.
[42:52.630 --> 42:54.410]  It's got Ethernet ports, USB ports.
[42:54.630 --> 43:05.910]  The idea is that it has multiple network interfaces, some of which are on the high side or the unencrypted side, the raw data side, some of which are on the public side, where only encrypted traffic is going.
[43:06.070 --> 43:08.190]  So you can keep those separated in the device itself.
[43:08.450 --> 43:11.510]  It's got a USB port, so you can plug in a 3G or 4G modem.
[43:11.750 --> 43:19.270]  The cool thing when you go to China is you can actually buy or rent a or buy a cheap USB modem, plus like a week of service for like $20.
[43:19.770 --> 43:25.930]  So instead of worrying about roaming cellular phones, you just buy one of those things, plug it in, use IP services for everything.
[43:26.410 --> 43:31.170]  And rather than just working with Tor itself, we added a lot of other transforms and tools to it.
[43:31.230 --> 43:34.810]  So we built a GUI for it, make it manageable, everything else.
[43:35.650 --> 43:45.990]  The idea behind it is to take all of the flexibility and simplicity of a travel router, the power of something like Tor pluggable transports, and to make it really easy to use.
[43:46.270 --> 43:51.950]  So you can simply go in, set it up, choose which transports you want to use, enable them, and run them.
[43:52.110 --> 44:00.570]  But most importantly, something that's pretty simple and gives you a very visual confirmation that it's functioning, and it's doing what it's supposed to do.
[44:00.730 --> 44:09.270]  Because one of the biggest issues I have with Tor is that sometimes your circuit can go down, and before you're realizing it, you're actually sending stuff out over the browser, and it's not encrypted.
[44:09.790 --> 44:20.550]  And sure, you can check it by going and checking what your IP address is, and seeing if it's one of the Tor ones, but the reality is it's not hard to make a tool that does that, and have it flag you the moment something goes wrong.
[44:21.390 --> 44:26.690]  And the nice thing about it being a physically separate device is it's really easy to conceptually understand.
[44:26.870 --> 44:35.670]  It's really hard for me to understand what a piece of software is doing when it's on a machine, especially if it's running like Mac or Windows or something, and I don't really understand every piece of software on that machine.
[44:35.750 --> 44:36.970]  There's all sorts of stuff that can interact.
[44:37.170 --> 44:41.530]  If it's a piece of hardware that only connects by one cable, and you pull the cable, it's not connected.
[44:41.750 --> 44:43.590]  So that's a very simple principle.
[44:43.930 --> 44:47.930]  And I did some defense contracting, and I saw how the government side of all this stuff works.
[44:48.150 --> 44:50.510]  They rely extensively on hardware crypto.
[44:50.630 --> 44:54.590]  They don't worry about software modules or everything to the extent possible.
[44:54.650 --> 44:57.550]  It's all hardware modules, physical cables, easy tracking.
[44:57.890 --> 44:59.390]  They worry about serial numbers, tracing everything.
[45:00.010 --> 45:13.310]  Much, much simpler way to operate that when you're in an environment that's either high stress or lots of users, or maybe some of your users aren't quite as smart as some of your other users, and something can go wrong, it will go wrong, you make sure the tools are as simple to use as possible.
[45:17.230 --> 45:19.570]  So, at the end of the day, monitoring is pervasive.
[45:19.730 --> 45:20.410]  It's here to stay.
[45:21.350 --> 45:22.770]  We can do a lot to change it.
[45:22.850 --> 45:27.950]  We can ensure that all our applications are using encrypted transports at their back end.
[45:28.150 --> 45:30.810]  We can ensure that we use HTTPS all the time.
[45:31.290 --> 45:35.390]  Reality is, only 6% of traffic is using SSL.
[45:35.590 --> 45:42.710]  And it's only gone up like 5% in the last 12 months.
[45:42.830 --> 45:44.250]  It's not going to change radically.
[45:44.490 --> 45:50.850]  So we need to take radical steps to protect ourselves, because they are monitoring.
[45:51.370 --> 46:05.930]  And whether you want to protect your privacy, because you're just an ordinary user who wants to ensure that you don't get flagged and your traffic isn't spied on, or whether you're someone who's doing high-risk activities and you want to stay safe, extraordinary steps need to be taken.
[46:07.470 --> 46:08.610]  So, we've got some boards.
[46:14.170 --> 46:14.550]  Yeah.
[46:14.550 --> 46:15.790]  We're not quite ready for demo.
[46:15.910 --> 46:20.470]  And to be honest, it's kind of, it's a little bit pointless to show and tell from this distance and this size of a room.
[46:20.770 --> 46:23.350]  But look, circuit boards.
[46:25.650 --> 46:27.890]  These are cheap commodity boards.
[46:28.430 --> 46:30.570]  These systems will not be expensive.
[46:30.870 --> 46:32.170]  You'll be able to use whatever hardware.
[46:32.350 --> 46:33.350]  It's all open source software.
[46:33.570 --> 46:38.390]  So it'll be... because there's no way you could trust a closed source piece of software to keep you safe for this kind of thing.
[46:38.810 --> 46:40.450]  And it'll be easy to use.
[46:40.950 --> 46:45.230]  One of the amusing things I did over the weekend was playing with this.
[46:45.690 --> 46:58.090]  I plugged in a GSM board and activated my own little base station running OpenBTS and found that actually I can run quite a secure telephone network.
[46:58.650 --> 47:01.070]  The only downside is it's obviously very legal.
[47:02.110 --> 47:03.730]  Which is a downside sometimes.
[47:06.010 --> 47:07.390]  Educational purposes only.
[47:11.090 --> 47:12.970]  So we'd be happy to take questions.
[47:18.060 --> 47:20.560]  Uh, uh, no comment.
[47:21.220 --> 47:25.480]  Can we post these slides someplace if you haven't?
[47:25.640 --> 47:26.440]  Uh, yes, we will.
[47:28.720 --> 47:31.180]  Uh, Twitter, at Octal.
[47:31.440 --> 47:31.660]  Follow.
[47:31.900 --> 47:32.600]  I'll tweet the location.
[47:37.690 --> 47:38.570]  Thank you very much.
[47:51.110 --> 47:53.130]  What kind of the GSM base station is running?
[47:54.250 --> 47:54.950]  No comment.
[47:55.690 --> 47:56.390]  Oh, really?
[47:56.790 --> 47:57.270]  It's...