Can a terrorist murder me with my car's computer?
March 23, 2013 2:15 PM   Subscribe

Can "hackers" use the power of the internet to control anything that has a computer in it?

I recently read a thriller in which a hacker-terrorist wreaks all kinds of mayhem from his keyboard. He crashes a plane, wrecks a Disneyland ride, and attempts to murder someone by trapping her in a car and revving up the engine to make it explode, or something. In all these cases he not only takes over control of the computers but disables all manual controls - like in the car, the victim tries to manually open the door lock but it doesn't work because "the computer" has control. When reading this it struck me as incredibly implausible - but how implausible is it? Are we talking, yes it could happen but it's really unlikely because of x, y, and z - or is the idea more along the lines of faster-than-light travel and magic powers?
posted by Daily Alice to Media & Arts (17 answers total) 3 users marked this as a favorite
 
If (a) the car's computer is accessible via a network (b) the computer is used to manipulate key systems than yes, it's possible that someone can use exploits to control the car. This has already been demonstrated. CAESS is a research organization dedicated to car IT security.
posted by Foci for Analysis at 2:24 PM on March 23, 2013


Note that in the above example they presumably had physical access to the car. The computers in normal cars do not, as far as I know, have the capability to connect to the internet.

The fact that there have been millions of hacker attacks in the last few decades and none have done anything even a little like this leads me to believe that it's extremely outlandish. The only things hackers have accomplished is defacing/taking down websites and stealing confidential information.

"Cyber-attack" is the latest "threat" the government is hyping up, sure. So I wouldn't be surprised to hear them claim that "terrorists" can do these things unless we surrender all our privacy and let them spy on our internet activity 24/7. But I don't know of any plausible scenarios where they could cause mass physical destruction.
posted by drjimmy11 at 2:43 PM on March 23, 2013


Even with a networked computer, the possibility depends upon the specific thing.

Most cars, for example, do not have "brake by wire" systems, so the brakes are not computer controlled. Thus, while they could control some car functions, disabling the brakes would only be possible in the very small number of cars where the computer controls brakes. And so on. While you could probably damage a cars engine by messing with the computer settings (racers often tweak all sorts of engine settings and htis can lead to damage to the car), I don't think you could make the engine actually explode.

The best car example would probably be something like the Prius acceleration issue, where the car could basically simulate the gas pedal being pushed all the way. You could still throw on the emergency brake and do other things, though, and it certainly wouldnt explode.
posted by wildcrdj at 2:43 PM on March 23, 2013


While it's not directly along the lines of the examples you mentioned, a slightly more plausible scenario would be of a hacker taking advantage of SCADA systems to wreak havoc. These are systems that are in place to monitor everything from power plants to traffic signals to communication infrastructure, etc. and are not always designed with security in mind. One of the reasons for this is that these systems were designed before the internet made networking a trivial and desirable thing. Imagine all the lights at an intersection turning green, or a monitoring system giving false readings at an industrial plant. (I don't like to be fear-mongering about this, and should mention that these are just theoretical attacks at this point in time)
posted by antonymous at 2:58 PM on March 23, 2013


Note that in the above example they presumably had physical access to the car.

I imagine you could cause substantial mayhem if you could hack into OnStar, and then use OnStar to, say, turn off a bunch of cars on LA freeways.
posted by ROU_Xenophobe at 3:06 PM on March 23, 2013 [1 favorite]


like in the car, the victim tries to manually open the door lock but it doesn't work because "the computer" has control

I think this particular one is more plausible than the rest. The wiring harnesses in a car are a surprisingly expensive part of the car, all told. So a major trend in recent decades has been replacing dedicated wiring with various kinds of networks (CAN and LIN buses, eg) that are shared among various mostly-unrelated functions.

Most cars, for example, do not have "brake by wire" systems

Many cars have antilock brakes and possibly traction control (ABS's inverse), both of which are typically implemented in software AIUI.

The computers in normal cars do not, as far as I know, have the capability to connect to the internet

Connections to radio networks aren't too uncommon, though (onstar, fancy navigation aids that deal with traffic conditions, etc.; not to mention increasing cellphone/car integration). The thing that I get from the autosec research is that the various components of a car network aren't generally firewalled off from each other; if you can own one component, you can often mess with all the others to some degree, or possibly own them as well.

Personally, I think the plausibility of hackers doing something deadly to a car without physical access is above the "wouldn't make my eyes roll if encountered in a Tom Clancy novel" threshold, but below the "I would expect heads of states' security people to be especially concerned about it" threshold.
posted by hattifattener at 3:13 PM on March 23, 2013


Many cars have antilock brakes and possibly traction control (ABS's inverse), both of which are typically implemented in software AIUI.

at worst the hacker could disable these systems (or turn them all the way up). In either case the mechanical links of steering and the hydraulic links of brakes would not be affected and still work jsut fine. Computers in cars mostly just control the fuel/air ratio and provide an electronic 'nanny' to keep the system from performing TOO well. For example anti lock brakes pulse/modulate hydraulic pressure minutely to keep the brake system working at the limit of friction for the tires preventing them from skidding. They do not release the hydraulic pressure. and, anyway, ABS systems are really, really dumb and not usually networked in with the engine management computer and I would think pretty immune to hacking. So far all that has been demonstrated is either totally disabling the engine and locking/unlocking the car and remote starting (if your car is so equipped-they can't turn on system that aren't there). All of this was done in controlled enviroments where the hackers had a LONG time to work with that specific car.

NOTHING overrides the mechanical linkage that exists between say the little button/lever on most cars and the door locking mechanism. Most 'embedded' system really just provide feedback to some monitoring system somewhere and don't really 'control' anything directly. For a non car example most traffic signals are controlled by a big, clunky computer in a (very tough and secure) box next to the signal that is physically wired in such a way that two opposing greens CAN'T show at the same time. The little buttons for pedestrians or the loops in the pavement that detect cars just tell the box that hey someone wants to go but don't override the programming. IF the big clunky computer does not have it in the programming to do something than that signal means nothing and is ignored. So if someone with malicious intent gains access (and there is almost always physical lock preventing reprogramming as well as one preventing physical access) the most likely outcome is an all red flashing signal as these things are not easy to program in a way that the computer will accept (really-engineers go to college to learn this stuff)

If you want nightmare fuel most biological testing on water systems are running 12-24 hours behind the water being introduced in the municipal system because it takes time to culture harmful bacteria in the water. This is barely the time most of the water is resident in the system and sometimes longer than that. And most water treatment plants are not as secure as say...schools are (for smaller towns most are fully automated and only have a single person by a day to run those cultures).
posted by bartonlong at 3:34 PM on March 23, 2013


I don't like to be fear-mongering about this, and should mention that these are just theoretical attacks at this point in time

Except that it's not theoretical, and has been reported in various forms in the wild - and these are only the ones that have come to light. Summary: someone (most likely the US) created a worm which affected centrifuges used for refining nuclear material, successfully infected an Iranian nuclear facility with this worm, and wrecked the centrifuges, thereby slowing Iran's progress.

On the other hand, for the moment this type of attack seems to be limited to state-sponsored industrial sabotage, and scenes such as those in Die Hard 4 aren't likely to crop up any time soon.
posted by fearnothing at 4:13 PM on March 23, 2013


One of the obnoxious things I noticed when I was visiting someone in the hospital a few years ago was that the automated pumps controlling the rate at which medication is delivered to patients are controlled remotely by computer. (In that particular facility, haven't seen it elsewhere so far.) Later on I came across a 1993 Law & Order episode that hypothesized killing someone this way. I wonder how long it will be before evidence is discovered of a real case.

If you were going to try to hack into someone's car computer I'd think the thing to try would be getting in to their phone first, then attempt to gain access through the Bluetooth connection.

If a type of product is computer-controlled you can bet some manufacturer somewhere has made a version with stupidly gaping obvious security holes, and that many more subtle exploits exist besides that. I was just reading this analysis from a few years ago analyzing the ways anything with a web browser might be taken over by delivering a virus in the fonts of a web page.
posted by XMLicious at 6:45 PM on March 23, 2013


Response by poster: The part that seems least plausible to me is that "the computer" would prevent the victim from manually unlocking the door (or the pilot from manually landing the plane, etc.) Is there some command that could be triggered that would disable manual controls? Why would they even have that?
posted by Daily Alice at 9:14 PM on March 23, 2013


Well, key fob remotes can lock and unlock the doors of the car... if you want a more feasible scenario to explain the movie, maybe you could say that the hacker has taken control of some nearby device that can simulate the RF transmissions from the key fob and has a way to time it so that he triggers the lock immediately after the victim tries to unlock it, before she can open the door?
posted by XMLicious at 9:52 PM on March 23, 2013


Best answer: While black-hats can exercise a surprising amount of power over real-world things... remember, their power extends to the boundaries of the network, and no further.

In computer security, we call this the "air gap."

Point the first, yes, your car's onboard computer controls the doorlocks. It's why they all lock once you go faster than 10mph or so - safety. They don't want your kids tumbling onto the highway at speed.

Point the second, the onboard computer controls various aspects of engine management, and can be made to do bad, bad things to your motor - hot-rodders maul their engines this way deliberately with a special computer interface that plugs into the car under the dashboard.

This brings us back to the air gap. Your car is off the network. It cannot be reached by bad hackers, because it's designed to operate for a few decades on its own, separate from any connection to a computer network.

But, this brings us to OnStar, GM's system for helping drivers - it will provide directions and roadside assistance if you hit that blue button on your rear-view mirror. Officially, the information from the onboard computer - if the air bags deployed, if the engine registers no oil pressure or an overheat state - can only flow outbound. Unofficially... yeah. It is a concern, as no network of that nature can be truly one-way.

In reality, no hacker has been this smart. In fantasy... yeah, it's a fun thing to woolgather over.
posted by Slap*Happy at 10:07 PM on March 23, 2013


Best answer: I think this particular one is more plausible than the rest. The wiring harnesses in a car are a surprisingly expensive part of the car, all told. So a major trend in recent decades has been replacing dedicated wiring with various kinds of networks (CAN and LIN buses, eg) that are shared among various mostly-unrelated functions.

Besides being just a good idea, I believe it is federal law that the driver's door can be opened simply by pulling the inside handle, locked or unlocked.

OnStar would be the most plausible way to do this right now. I believe I've seen commercials where a certain manufacturer's cars can be locked, unlocked and started via an app on a smartphone, so those would be vulnerable too.
posted by gjc at 2:28 AM on March 24, 2013


Response by poster: Oh, yikes. I hadn't thought about OnStar as a way to access cars' computers remotely.
posted by Daily Alice at 8:56 AM on March 24, 2013


in your car you could do a small test, just hold down the electric lock button and see if someone else can open the door. when I was a kid I would do that to my sister, I'm not sure about the driver, but the passengers couldn't unlock their door manually, the electic locker was too strong. Probably not good for the lock system, but it outlasted the rest of the car...
posted by Iax at 12:49 PM on March 24, 2013


Oh, yikes. I hadn't thought about OnStar as a way to access cars' computers remotely.

It's totally possible that OnStar has their security set up double super good. I don't mean to impugn them in any way. Just that there is a connection that possibly could be exploited.
posted by gjc at 12:54 PM on March 24, 2013


What about the old BlueTooth Sniper Rifle from like 2005? Sure, here's a link: http://boingboing.net/2005/03/13/howto-build-a-blueto.html

Whatever you can get to via BT -- primarily audio gear (phone, stereo, etc.), but who knows these days -- could be messed with. Have they fixed this old problem since then?
posted by wenestvedt at 2:15 PM on March 24, 2013


« Older I have a question about animal containment, or...   |   Pump Up The Soap. Pump Up The Soap. Pump Up The... Newer »
This thread is closed to new comments.