Hackers Reveal Nasty New Car Attacks

schwit1 writes "Stomping on the brakes of a 3,500-pound Ford Escape that refuses to stop–or even slow down–produces a unique feeling of anxiety. In this case it also produces a deep groaning sound, like an angry water buffalo bellowing somewhere under the SUV's chassis. The more I pound the pedal, the louder the groan gets–along with the delighted cackling of the two hackers sitting behind me in the backseat. Luckily, all of this is happening at less than 5mph. So the Escape merely plows into a stand of 6-foot-high weeds growing in the abandoned parking lot of a South Bend, Ind. strip mall that Charlie Miller and Chris Valasek have chosen as the testing grounds for the day's experiments, a few of which are shown in the video below. (When Miller discovered the brake-disabling trick, he wasn't so lucky: The soccer-mom mobile barreled through his garage, crushing his lawn mower and inflicting $150 worth of damage to the rear wall.) The duo plans to release their findings and the attack software they developed at the hacker conference Defcon in Las Vegas next month–the better, they say, to help other researchers find and fix the auto industry's security problems before malicious hackers get under the hoods of unsuspecting drivers."

Comments

High risk

[suso]

"The duo plans to release their findings and the attack software they developed at the hacker conference Defcon in Las Vegas next month–the better, they say, to help other researchers find and fix the auto industry's security problems"

As a security researcher who believes in the spirit of the open release of vulnerabilities, I feel that this is irresponsible behavior on the part of these security researchers. We're not talking about releasing a vulnerability that will compromise someone's e-mail. We're talking about a high risk vulnerability that could cost some random person their life. These two gentleman should take a deep breath before releasing this information to the computer industry first rather than the auto industry. The auto industry may not have a tradition of attending these types of conferences and so by releasing the information at Def-con you're giving the wrong people a head start. Sure, the auto industry already knows about these problems, but you have to try to give them the benefit of the doubt when you confront them about the problems that they will try to fix it.

Re:High risk

[Xaedalus]

The mere fact that this has been announced has already started the wrong people working on it. At this point, releasing at Def-Con is the right thing to do, because not only will that patch get fixed, but others will come to similar conclusions and keep an eye out for peers who are going to exploit this. Black hats have family too.

Re:High risk

[Anonymous Coward]

Right now they have to hook directly into the odb plug to do this, the same person with that kind of physical access can do any number of nasty things to your car.

They are more warning about the lack of security when this stuff becomes accessible remotely (cellular or otherwise wireless) that there are going to be serious security issues as anyone breaking into that remote access path can do serious things.

Re:High risk

[dyingtolive]

Or the attacker just cut your brake lines.

That's not a hack though, more of a snip.

Re:High risk

[viperidaenz]

Once someone has physical access to a vehicle, there are worse things they can do than mess with the traction control and abs systems.

Re:High risk

[Anonymous Coward]

You mean like if there was some embedded computer plugged into the same CANbus as the OBD port, that had a cellular radio on it that was already shown to be vulnerable to attack? One sold on every new car from a certain major manufacturer?

Yeah, in the future, when OnStar exists, there will be serious issues. Wait, was "future" the right word?

The underlying problem is that CANbus was designed by automotive engineers and not network security people.

Re:High risk

[radiumsoup]

You speak as if all companies are equally bad. Somehow, I think you're either young or more sheltered than you believe you are.

Re:High risk

[Joining Yet Again]

Argh, sophomores everywhere.

Security through obscurity isn't "no security at all". It's just inadequate. There's still the hurdle of overcoming obscurity.

Just like strong cryptography is great but not perfect because 1) implementation is often flawed; 2) rubber hose.

Re:High risk

[HornWumpus]

Honda and Accura nav systems are also apparently hooking into the OBD port. They report codes on the nav screen, can't (or won't) clear them.

Re:High risk

[HornWumpus]

OBDII is not wireless.

Re:High risk

[Anonymous Coward]

But the engine controller is going to have some form of authentication required and the hackers are going to be stopped right there.

Yes, I too had noticed that authentication systems were 100% proof against hackers, especially those implemented by companies that obviously have no prior interest in security.

Re:High risk

[suso]

Because we all know that if the researchers quietly tell the auto manufacturers they will fix the issues and make sure everything gets updated. Our upstanding auto manufacturers would certainly not try to bury issue and sue the reporters out of existence!

As a security researcher you should be used to companies trying to deny, bury and ignore reports instead of correct them.

Seriously, the only way to get a company to fix a flaw is when the pr nightmare becomes so great that it is cheaper to fix the problem than deny it.

Yes and I also know about technically minded people denying that problems are real issues too (See libvte vulnerability). DARPA has known about these issues for a while now and apparently the issues are a lot more real and scary than most people realize. We're talking about the ability for a hacker to do something to your car simply by playing a song over your iPod or on a CD. Or a program being injected the next time you get an oil change because the service center's computer had been hacked remotely.

And we're not talking about ego maniac hackers sitting in their basements causing a few cars to honk their horn because they think it will be funny, we're talking about terrorists and countries writing a song that one day everyone plays one day and we have 1 million 60mph 2 ton missles with families in them flying up the road all the same time. That will be a very bad day. But that's ok, because we tried to tell the auto manufacturers and they just didn't listen, so its their fault right?

What these researchers are doing here is treating this vulnerability as if its any other vulnerability, which its not. Human life is at stake, not your email or bank account password. Yes, they do recognize the dangers, but they don't seem to realize that they should be changing their approach accordingly. For instance, they do their tests out in the open on public roads and put someone behind the wheel who doesn't know what is going to happen. You don't really need to do that to demonstrate that there is a problem.

Re:High risk

[chiefmojorising]

Seriously. I've got a hack that'll disable the brakes on any car ever made. It's called a hacksaw (heh) and requires even *less* access than these guys had.

Re:High risk

[suso]

And what cars are those?

Me, I stay safe and only drive cars with carburetors.

Until one of the hacked cars hits you head-on at 60 mph.

Re:High risk

[dgatwood]

While your argument has merit, I'm going to simply stick to the strategy of buying cars that do not attach a wireless communication device to the same bus that the engine control unit sits on.

As for me, I'm going to stick to buying cars in which the brake master cylinder is physically depressed by the pedal, and in which the emergency brake lever is physically connected with a mechanical cable....

I drove a rental car the other day with an electronic emergency brake. I've never been more uncomfortable driving a vehicle. Besides having "safety" features that made it really clumsy to drive (you can't release the emergency brake unless your foot is on the brake pedal, for example, which doesn't make any real sense if the vehicle is in a flat parking space, with the transmission in Park), I just can't see myself ever trusting a car in which a computer failure could kill the emergency brake entirely, and in which there's no way to apply more force on the emergency brake in the event of an actual emergency. That design pretty much defeats the whole purpose of having an emergency brake.

Ugh.

Re:High risk

[Roskolnikov]

unless you add a wireless dongle (they come in Bluetooth and wifi but they still require physical access and close proximity).
every person that has done a 'reflash' on their car to get more performance has done similar things, I can with the right parameters make my cars motor throw a connecting rod through the block, I don't consider this hacking, I consider it sky is falling stupidity... if they had done this through on-star, now that, I would consider hacking and truly the danger that should be exposed by this article.

Re:High risk

[iamgnat]

Then you've never heard of the CAN bus, which is in use on every car produced since 1996. You'd have to avoid anything with obvious wireless access, which means no lock/unlock/panic/remote start systems, and likely not even a car radio since many are on the bus as well.

No, ODB-II was mandated on every new car sold in the US starting in 1996. CAN didn't gain mass adoption for quite a while yet (I have a 2001 with out it and just replaced a 2004 not too long ago that didn't have it).

All of the things you listed as not being possible without CAN were also around long before CAN (and well before ODB-II (though entirely unrelated) was mandated).

Even for the cars that are built today, there are still a fair number that do not have any wireless access to the bus (e.g. cars without OnSTAR or the like). I just bought one in fact. The wireless access was his concern and he still has plenty of options to avoid that while still having all the other benefits of a CAN based car.

Re:High risk

[WaffleMonster]

Right now they have to hook directly into the odb plug to do this, the same person with that kind of physical access can do any number of nasty things to your car.

TFA asserts otherwise. Apparently onstar and integrated infotainment systems can obtain same access to CAN bus access as the OBD port.

They are more warning about the lack of security when this stuff becomes accessible remotely (cellular or otherwise wireless) that there are going to be serious security issues as anyone breaking into that remote access path can do serious things

"When this stuff" ??? This is 2013.

Re:High risk

[Charliemopps]

Uh... it's already being exploited...
http://news.yahoo.com/blogs/lookout/hastings-crash-witness-113514329.html

Re:High risk

[LinuxIsGarbage]

The problem is most drivers don't know that it's a redundant system, and never think of trying the parking brake if the brake pedal fails. This is one area where linguistic drift has hurt us. They were originally called the emergency brake, whose name clearly implies they're to be used in an emergency if the regular brakes fail. But since they were also used to keep manual transmission cars from rolling when parked, they've colloquially been called parking brakes. To the point where most people refer to them as parking brakes now and don't know about their emergency braking function.

It's also referred to as a hand-brake (especially outside of North America where front bench seats with foot operated e-brakes where not near as popular). I've heard of people trying to use it in an emergency, a panic stop situation. In which case it's far worse than the service brakes, unless the service brakes have failed.

The truth is a frightening number of people don't understand how the cars they're driving work, and it's not just limited to e-brake / p-brake / h-brake, and it's not due to the name. Many don't understand basic concepts of gears, how and when to use manual modes of an automatic, how to shift into neutral or kill the ignition in the case of a stuck throttle. A shocking number of people don't understand that an oil light means a loss of oil pressure and the car should be pulled over and shut off immediately. A shocking number of people don't know how to jump-start a car, or change a flat, or check / adjust their tire pressure, or oil / tranny / brake / power steering fluid. A shocking number don't know that a quick blinking turn signal means you have a turn signal bulb out.

Re:High risk

[lennier]

The underlying problem is that CANbus was designed by automotive engineers and not network security people.

A good point. Another way of phrasing the problem I think is:

Systems are too often specified, designed and tested entirely in terms of their positive capabilities, rather than their negative capabilities. In the networked remote security environment, we need a design process that guarantees both.

In other words, most of our design process up to now has been all about "what a system CAN DO". But securing a system from to intelligent attackers is about what that system CAN'T do, even in the worst case. And since the number of things a Turing-complete computer with an always-on connection to the Internet CAN buut SHOULDN'T do is potentially infinite, that can be really difficult.

Tests generally only cover the positive features. It's hard to achieve complete test coverage by trying every possible combination of bad input (though fuzzers seem to be doing quite well at finding vulnerabilities, and it's embarrassing that amateurs keep finding bugs that the professional developers didn't.) Typing seems to be more useful in limiting capability, but our current type systems are very limited - for example, in most OO languages, the type system only guarantees that the call signature of a method is correct; it doesn't give any way of describing any other invariants that should be preserved during the computation; and the entire architecture of OOP is based on methods with side-effects which scales really badly to concurrent processing.

I think we've reached the limit of what can be safely achieved with loosely-typed imperative side-effectful OO languages like C++. These languages give us enormous power to create positive capability, but very little in the way of assuring negative capability. I'd like to think that Haskell or Erlang might be a way forward, but I've yet to wrap my head around either of them. I'm hoping we can eventually get something simpler, that allows creativity where it's needed but also lets us place hard limits on what unexpected interactions can arise.

Re:High risk

[AJH16]

Apparently in their test case, the telematics unit did have access to all 3 speeds of network. That's really goofy since it shouldn't need access to all the networks. Basically CAN buses have 3 speeds of network, a low, medium and high speed network with different types of data on each. TPMS for example is generally low, ABS is normally high speed and your typical error codes and car locks and a lot of the status reporting is on the medium speed. Many ODBII connectors won't connect to multiple of the networks unless you get more expensive units and internally not all components in the vehicle are capable of talking on all of the networks.

Re:High risk

[adolf]

It's already a risk. And it's cheap.

Plug a GSM modem into an RS-232 ODB2 interface.

Programming it is really hard, and stuff: ATS0=1&W

Power is even harder: Ignition-switched power is supplied by the ODB port.

Using it is perhaps the most difficult part: Call the modem, it answers and connects, and...done.

Yeah, sure: I can do all kinds of obvious and nefarious things to a car if I have physical access to it.

But this way, I can have the car work perfectly normally for a week or a month or whatever.

And then, long after they forget about having their window smashed and the change holder looted in the middle of the night, I can have it misbehave at exactly the moment that it is perhaps most dangerous.

Just sayin'.

Re:High risk

[JWSmythe]

    Apply Occam's Razor. User or mechanical failure are much more likely than his car being hacked.

    The story talks about a *wired* port by the parking brake. That would mean the attacker was in the car, or a remote device was attached, which investigators would (or at least could) find. It also only addresses a specific Ford vehicle, which has no relationship to a Mercedes.

    Significant user failure would seem to be present. Options are available when the brakes don't work. Downshift. Turn off the key, let the engine stop, turn the key on to unlock the steering wheel. Spin the car. Even hard maneuvering will bleed speed off. Ask any racer. Turn the key off, let the steering wheel lock, and have a slower speed impact into a fixed object.

    The option of driving as fast as possible, and dying in a fireball is the poorest choice. A conspiracy is one the must unlikely scenarios, only slightly better than alien abduction/intervention, and poltergeists taking over the car.

    I'm kind of fond of the alien theories.

    If it were the feds, wouldn't it be easier to pay a thug to do a random carjacking? A home invasion gone wrong? Shot by SWAT in a drug raid at the wrong address? There are a million other ways to remove someone without needing a high tech solution that doesn't exist yet.

Re:High risk

[Ol Olsoc]

TFA asserts otherwise. Apparently onstar and integrated infotainment systems can obtain same access to CAN bus access as the OBD port.

Onstar can do many things to your car outside your control. Remember when they were bragging about how they could disable your car if someone stole it? It worked by disabling the throttle, forcing the vehicle to idle, so the perp would pull over to the side of the road. My guess is that if they can do that, they can controll a whole lot more. They can remote diagnose car issues, so that means they access things like timing, engine temps, vacuum lines, no doubt much more. And if they can read them? Who knows?

And you don't even need to subscribe! http://www.consumeraffairs.com/news/gm-includes-free-remote-start-on-2014-models-060713.html

How about that? They can start your car remotely. Umm, that means they can stop it remotely.

How about this? They track you for free, and sell the data. Of course you are anonymized. Until you aren't.

http://www.wired.com/threatlevel/2011/09/onstar-tracks-you/

Re:High risk

[Jah-Wren Ryel]

If it were the feds, wouldn't it be easier to pay a thug to do a random carjacking? A home invasion gone wrong? Shot by SWAT in a drug raid at the wrong address?

I'm not going to speak to the larger question of how true the theory is, but to this question there is a good reason why not. If they can make it look like the crash was completely the driver's fault then that would eliminate any question of it being a government hit. All those other options involve third parties that, exactly as you postulated, could be hired to do the hit.

Re:High risk

[AmiMoJo]

CANbus is quite old, originally specified back in 1986. It is designed primarily for robustness in a noisy automotive environment. Back then there was no OBD, no internet, no mobile phone network. It's hard to see how the designers could have predicted all that and designed in security based on algorithms that had not been invented at the time.

Security could be added now but it would push up costs a lot. Most CANbus devices are very simple embedded systems, and there are hundreds of them in a modern car.

The problem is that the CANbus and everything attached to it should never have been made externally accessible. Forget physical access, once you have that there is nothing you can really do, it's the systems like OnStar that allow remote access which are the problem.

Re:High risk

[pla]

Downshift.
Does nothing on an automatic until your speed drops below an appropriate threshold. Even reverse won't engage until you come to a stop. Park theoretically jams the output shaft with a pawl, but even that can't "catch" above a certain (very low) speed.

Turn off the key
Many new cars (Priuses, for example) don't have mechanical keys, just a button that even under the best of conditions doesn't always do quite what you want it to - Hold it just a hair too long or too short, or have the car in the wrong gear for what you want to do, and it just laughs at you.

Spin the car.
At 80MPH, "spinning" the car means flipping the car, and will likely get you just as killed as the "brick wall" method of decelerating.

Even hard maneuvering will bleed speed off.
This one really will always work, but as with spinning, careful just how hard you maneuver at high speeds.


Overall, Sorry for the negative tone I have here, because I completely agree with you in spirit. If the driver doen't panic, he can do a lot to slow down a car with no brakes and/or a stuck accelerator. Most people don't expect that to happen, though, and simply go into a mental freeze, stomping uselessly on the brakes harder and harder rather than taking other corrective measures. As you say, "Significant user failure would seem to be present".

Locking down the cars for security

[IndustrialComplex]

I can appreciate applying Anti Tamper and other IA techniques to 'harden' cars, but I hope this doesn't return us to where only ''licensed' repair facilities can work on cars.

Meh... Give me access, I own your computer

[Mr Krinkle]

So

if I'm sitting in your car, plugged in to the canbus, I can control things on the canbus....

Yeppers....

Just like if I have access to your laptop for long enough, I can get whatever is on it. (encryption will slow it down, but like I said, given time and access?)

But you'll probably notice me sitting in your car, plugging a cord into the port before I take the time to crash your car, with me riding in it.....
While this is amusing, I'm not that nervous about "security through not having some donkey plug his laptop in your car with a death wish while you are hurtling down the highway"

Having them use the "open" canbus specs, you can add aftermarket devices, and not have to take your car to the dealer for any service.

If they fully lock it down, the dealer will be the ONLY place that could work on it. And the ONLY parts you could add to your car.

Accura/Honda Door-lock Exploit

[bradgoodman]

While they're at it - I don't think anyone has really discovered what the deal was with the Accura/Honda remote-control doorlock gadget that thieves were reportedly using to effortlessly break into cars. All the article said was "police are stumped" (duh).

Not News: They put it into brake service mode.

[Anonymous Coward]

To enter the Pad Service Mode, perform the following with the vehicle stationary:

1. Place the vehicle in Park and turn the ignition to the ON position.
2. Apply the brake pedal.
3. Turn the ignition OFF, then ON three times and then release the brake pedal. The total time elapsed for the three ignition cycles and brake release must be less than 3 seconds.

That's how you replace the brake pads. If they figured out how to do it through the OBD connector, whooptie do.

I have one of these vehicles. Fly-by-wire regenerative brakes are a little creepy, but supposedly if something goes wrong and you mash the pedal all the way to the floor, there's a hydraulic backup down there somewhere. I haven't had to try it.

Oh, and all this is no different than your holier-than-thou Toyota Prius, so don't blame Ford.

Re:So?

[mrex]

So they had hard-wired physical access to the car's data network and they were able to cause trouble? News at 11! (aka so what?)

So what? So I could bump key my way into your car, trojan one of the devices sitting on your car area network, and cause you to crash and burn on the highway with no meaningful evidence that anything was amiss.

(RIP Michael Hastings)

I don't get it ...

[recrudescence]

... can some one explain it to me with a car analogy?

Re:Nothing to see here...move along...

[HornWumpus]

Can you imagine where the motherfuckers would have hidden the plug had they not been told more or less where it had to go?

Re:Rev Up Those Conspiracy Theories -

[Anonymous Coward]

Or a reporter (Michael Hastings) whose award winning work caused Stanley McChrystal's resignation mysteriously dying in a single car accident with a tree; without skid marks and the engine winding up 200 feet away...

Re:And the NTSB wants cars to talk to each other??

[Obfuscant]

Or, even worse "OMG! YOU'RE GONNA HIT SOMETHING! EMERGENCY STOP!" to all the cars you pass.

I had something kinda like that 20 years ago. A microwave transmitter from an automatic door opener sensor. $15. A battery. $1. A switch. $1.

Watching the tail lights light up on all the cars that have just zipped past you on the freeway as the radar detectors in those cars start squawking. Priceless. Passing them as they slow to well below the speed limit. Priceless. Watching them zip past again, slam on brakes again, get passed again. Priceless.

Indeed there must be many ehtical companies ...

[golodh]

apart from the banks, the tobacco industry, the arms industry, big pharma, big oil, marketing firms and so forth.

If only because their helmsmen are required, by law, to maximise shareholder value. Nothing else. In fact: senior management can be sued if they don't set policy to that effect.

The upshot is that no publicly traded company can really afford a moral or ethical compass. What passes for ethics in companies is usually nothing but well-understood self-interest (as in: avoidance of PR damage and a resulting slump in sales through bad publicity).

Whilst I'm against releasing any kind of software vulnerabilities before the responsible parties have had a decent chance to fix it, I'm just as skeptical as most regarding the inclination of e.g. car manufacturers to improve security unless there is a massive PR debacle. For massive PR debacle read: a nasty and widely covered crash involving a photogenic celebrity (ugly celebrities won't cut it) and his/her children, that can be traced unequivocally to the lax security of a car's on-board datacommunication infrastructure.

That's the main thing I can see as getting their attention and lending the issue any kind or urgency. If only because of CYA considerations on part of top management. The only alternative would (in my view) be compulsory network safety standards for cars.

Re:Indeed there must be many ehtical companies ...

[SleazyRidr]

The board of directors do not have a requirement to "maximise shareholder value." Most companies could acheive this by liquidating their assets and investing a another company which is doing better.

The board of directors do have a requirement to uphold the company charter. Many charters include wording like "maximise shareholder value" but it is also clear that they intend to do this by providing goods and/or services to people, not by kidnapping college students and selling their kidneys.

Re:Indeed there must be many ehtical companies ...

[lennier]

The board of directors do not have a requirement to "maximise shareholder value." Most companies could acheive this by liquidating their assets and investing a another company which is doing better.

And isn't that exactly what the tidal wave of mergers, acquisitions and restructurings from the 1980s on have all been about? Buying and selling shells of companies, liquidating their assets, closing the factories, selling the brand to someone else, and then outsourcing the production to China and Mexico while centralising the banking in London, the paperwork in the Cayman Islands and the corporate headquarters in New York.

Doing this kind of shell game creates a reputation for a CEO as a "miracle worker" and "turnaround artist" and billions of dollars in share value. But if you look behind the scenes you see an increasingly hollow stack of cards that's propped up by debt and gambling rather than production.

Re:Nothing to see here...move along...

[rahvin112]

Yes I can, it would be on the back of the engine and would require a special tool only sold by the dealer to open the door and would likely require the removal of the starter motor and timing belt/chain to access and for bonus points someone like Porsche would require removal of the head gasket to reach the port.

Putting it within 2 feet of the driver was smart, it should have had the additional requirement to be within 6 inches of both the radio and climate controls because if they had everyone would notice some strange object plugged into the port.

This has already been done. (On another(?) car.)

[tamyrlin]

There is at least one car model where researchers has been able to get access to the CAN bus and do all sorts of shenanigans through the following means:

• * Specially crafted file on a CD inserted into the CD player
• * Exploit weakness in the car bluetooth interface
• * Exploit weakness in built in GSM modem

For the details, see http://www.autosec.org/pubs/cars-usenixsec2011.pdf. (Pretty scary reading. In this case they are also able to disable the brakes and they are also able to engage the brakes on only one of the front wheels for all sorts of "fun"...)

Re:Not News: They put it into brake service mode.

[ChumpusRex2003]

Correct. On this type of hybrid vehicle, there is a regenerative braking system.

Under normal driving conditions, while the vehicle is in motion, the motor/generator will be used to retard the vehicle. The brake pedal is connected to a electronic pressure sensor, and also mechanically to a hydraulic master cylinder.

Unlike on conventional vehicles, there is no vacuum powered booster, instead the master cylinder hydraulics are used to operate an electro-hydraulic servo, with electronic override. This way, under emergency braking, you get full hydraulic force applied to the wheel cylinders with minimal pedal effort. The electronic hydraulic control will also apply hydraulic pressure when the vehicle is stationary and the brake pedal depressed, and also periodically applies hydraulic pressure when the vehicle is stopped and the transmission in P (for self-test purposes) and when the vehicle is powered on.

The hydraulic servo mechanism can be disabled in order to permit brake maintenance (this releases hydraulic pressure in the booster and prevents automatic application of pressure to the wheel cylinders), permitting access to maintain the friction surfaces. It appears that this hack, merely consisted of transmitting the CAN bus command to put the hydraulic servo system into maintenance mode.

At low speeds, when the electrical regen isn't operative, this will result in the brake pedal travelling further than expected and loss of power assistance. However, with sufficient pedal pressure, it should be possible to slow the car using unboosted pressure.