The article does not make it particularly clear, but the malware in question is delivered through official first-party OTA updates on cheap Chinese aftermarket head units that happen to run Android. It cannot self-propagate to any Android-based head unit, nor does it affect Android Auto which is a "dumb" screen mirroring protocol with the bulk of the software running on the connected phone rather than the head unit. This seems to be a very similar situation to that of cheap generic Android TV streaming boxes, which often come pre-infected from the factory with residential proxies and other malware as well; most of the infrastructure is likely shared.
Why do they gloss right over how this was distributed? Barring details of any other kind of exploit we would have to assume the vendor's update server was compromised? If so why don't they just say so.
Indeed this is an odd disclosure and I am not familiar with past posts by them.
Moreover, no CVE is associated with this claimed vulnerability. It's not even stated which Android version or automotive head-unit variant version is affected.
It’s astonishing how this relatively long article contains not a shred of actual evidence that any of this is true. It’s all “alleged”, “raised concerns”, “sources said” etc.
Wikipedia's source policy makes it nearly impossible to refer to anything that is not in the media, and any sensitive article has to use weasel words like this. Are you just noting the issue with the article, or actually doubting that Kaspersky Labs is a de-facto FSB branch since at least 2015?
The article is about a controversy involving allegations. There is plenty of evidence presented that the controversy and the allegations exist. (And if you dig into the links, there is plenty of evidence that the allegations are not without basis.)
It cannot self-propagate to any Android-based head unit
Remember that not that long ago viruses spread through floppy disks.
Today, people share USB sticks full of music from one car to another all the time. They also bring their music from their home car to a rental car and back.
I’ve never met anyone irl who used USB sticks full of music. I know the capability is there in most cars, just never seen it. It seems like Bluetooth capability and Spotify/Apple Music landed in mainstream cars too soon after “play MP3s from USB” was added, for that to catch on.
> nor does it affect Android Auto which is a "dumb" screen mirroring protocol with the bulk of the software running on the connected phone rather than the head unit
Huh, how does that work anyway? And while we're at it, Apple CarPlay as well? Both can run wirelessly via Bluetooth, but BT is nowhere near capable enough to stream full bandwidth video?
It uses Bluetooth to stream audio, but everything else happens through a WiFi connection exposed by the car that the phone automatically pairs with after the Bluetooth handshake.
They didn't run over BT. BT is used to initiate communication and share the password to a wifi-network. It then uses that Wi-Fi network for most communication, keeping the BT channel strictly for telephony.
I thought the latest Bluetooth protocols were basically designed to hand off to an ad-hoc Wi-Fi connection between the two devices after the initial handshake. (Might be an oversimplification of the real protocol)
They actually run over WiFi (WiFi direct IIRC) - Bluetooth is mostly just used as a setup handshake and to help the head unit decide which phone in the car should be the one connected.
Headline really quite clearly implies it, though. I think the correction is apt.
Bottom line is that lots of HN commenters here, as is our wont, will see this as a platform bug with a hated rival and not a bad third party integration that introduced vulnerabilities.
Like, if it was a Linux-based edge system from some fly-by-night contractor, would you be OK with a headline like "Malware infects Debian based refrigerators"? What'd Debian do?
It’s no different than how the old Ford Sync or something else could have been compromised.
The two big things here in my mind are:
1. Android Automotive has gotten very popular since it provides so much and writing your own OS is very very hard and expensive as so many car makers found out
2. Aftermarket head units often use it (see #1) so it’s likely far easier to get out there than if you had to compromise Ford/VW/Volvo/whoever
This is not Android Auto though, which is an entirely different product suite designed to connect a OEM infotainment system to an Android device owned by the vehicle operator. That protocol is proprietary, Google-owned and managed, not part of AOSP, and not available to the integrator of the software in question.
The actually vulnerable system is a custom vehicle head unit that merely happens to be running a software stack based on AOSP. It's not even "Android" in a product marketing sense.
Again, it's like blaming Debian because some loon stuffed it in a wifi NAS or whatever and put a backdoor into their UI. It's insane.
> Since a head unit typically holds nothing of value to an attacker, one of the more likely attack scenarios using “classic” Android malware is infecting the device to recruit it into a botnet
People do pair them with their phones, though. I could imagine a future version of malware like this propagating laterally.
Almost, though I understand I am the exception rather than the rule: Personally I have an aftermarket android head unit since the standard one was incredibly basic, no real time navigation updates, updating maps was a pain in the ass and so on. Initially I did pair it with my phone but since it is an aftermarket unit from a company which apparently does not exist anymore, newer phones cannot be paired with it. So my only option was to go the opposite route and use my phone as a wireless hotspot(almost - there's a raspberry pi with openwrt between the two). And since I self-host everything, I had no choice but to hook it up to my vpn. That said, I understand the implications of doing this so ultimately the network access it gets is incredibly limited: everything that is not my music server and the maps provider has been cut off completely. The downside is that every now and then I get a "can't connect to google services" notification though that is technically reassuring from a security perspective.
Keep in mind however that the 4G SIM is not there for the driver's benefit, but for Nissan's. It collects extremely invasive telemetry that is then sold to data brokers and consumed by car insurers and government agencies, among others. This is why I won't drive a car that I own without first removing the onboard modem.
"Pairing" with a head unit is not an open socket to dump anything you care to down the wire. That would require finding a rather remarkable vulnerability in one of the audio/address book/screen mirroring APIs the devices use.
That's scary from a user perspective, but harder to monetise at scale as an attacker. Proxy endpoints are just another commodity (and offer recurring revenue).
If you infect tens of vehicles that's not that valuable. But if you infect ten thousand vehicles, convinced a trusted member of one of the bigger black hat forums it's real and have him vouch for your marketplace post, there should be some buyers for full movement profiles, call logs and address books of ten thousand people
And doing that doesn't really interfere with also setting up and selling proxy endpoints
They're called data brokers because they have a buy side, too. That might be peanuts to you, but to an AliExpress seller, it could be most of their profit.
Some head units (working with a 1st party one atm) have two networks: OEM-paid (unlimited data) and user-paid. A 3rd party apk would be consuming all bought traffic quite soon.
Also typical Android permissions still apply. The user would need to grant the malicious app contacts, call logs, etc permissions.
It seems like this exploit is targeting those that keep their phones tethered for connectivity outwards or hooked a USB modem or a SIM card into a cell-equipped headunit.
The only valuable thing there is the relatively 'clean' mobile connection... and this malware's dropping a residential proxy endpoint on the headunit to take advantage of it. Bonus points if the headunit is always connected and always powered up to a +12v rail in the car, that's free and always-on real estate!
Head units aren’t always-on. Typically they go into a low power standby 2-5 minutes after ignition / accessory mode turns off, and go completely power-off 30-ish minutes later.
Otherwise any car sitting unused for a week or two would have a dead battery.
A lot of older cars didn't turn off their OBD port, have their headunits go into standby, or even turn off the cigarette lighter port. Early OBD ports connected to dealer computers for a few minutes, not an always on dongle. Plain headunits just play music, what could they possibly accomplish by staying on when you turn off the car? It was a convenience having the cigarette outlet left powered so you could light a cigarette without turning on the car. Other than maybe a bag phone, what would you possibly plug into that?
No you wouldn’t, because then you always have a cold boot of the headunit, even if you just accidentally hit the ignition. Users want the head unit to resume within a few seconds. Just like their phone.
There are a lot of cars out there where the head unit has connection to the CAN bus. Which means this malware vector could be used to directly cause crashes. E.g. https://news.ycombinator.com/item?id=19751872
The car hacker's handbook [1] has a chapter on just using the infotainment system to access the CAN. Specifically mentions "attacking through the update system".
An aftermarket head unit connects to the CAN bus? The aftermarket head unit I installed certainly doesn't. Are you sure what you are saying, which is true for OEM units, applies to aftermarket ones?
Can't vouch for all car architectures, but in most cases the head unit is QM and safety domains are usually segmented from each other. So even if the head unit talks CAN (it needs to get car data somehow) it will only communicate with the rest of the car through a gateway that will not allow it to take any dangerous actions.
Many do - the one I was looking at for my vehicle in particular uses it to restore the steering wheel controls (which are broadcast over the CAN-B low speed bus)
For whatever reason, the idea of this being in my car is relatively scarier for me than if this was just my phone ?
I think partially as my mental model of both android auto and CarPlay is that they operate as a passthrough of my device rather than as an separate installation of the OS entirely (I wasn’t aware the head unit itself had the ability to install APKs independently).
Also, feel like John Gruber is going to have a field day with this one
So I can use android auto on an android automotive head unit - got it but also this seems needlessly confusing naming structure. Apple TV comparison is apt lol
Android Connect instead of Android Auto, and any word other than recycling the android app communication nomenclature "Intent." But hey I'm no branding genius so let's workshop this to close the loop.
I don’t believe this is Android auto running from a phone, but a situation where the manufacturer have used Android Automotive as operating system for the built in head unit. As e.g. on Volvo’s.
I hope we see "de-smartification" conversion kits that replace the electronics with more straightforward (and repairable) offline equivalents. The ultimate AV.
"How has the automotive industry adapted to decades of computing best practices?"
- Head units connected to CAN bus with bluetooth vulnerabilities allowing attacker to remotely activate locks and windows and sometimes even driving controls
- Unsecured CAN bus cables everywhere allowing cars to be stolen through headlights and behind mud guard flaps
- Keyless entry basically a shit show of faraday pouches
- OBD port allowing thieves to clone a full key in seconds
- Even cars in decent neighbourhoods have to use steering locks
Sorry but this is a fucking joke and the automotive industry is cancer.
At least Tesla actually bothers with user updates and production improvements, most other manufacturers just shit out the same model 5 years in a row with an extra cup holder and USB port (probably rootable) if you're lucky. That said, Tesla's insistence that everything be done by touch screen is dog shit.
All this and still for 99% of cars my iPhone stuck to the dashboard provides better maps and entertainment and yet they can't even make a fucking phone holder standard, not even a fucking mounting point so I don't have to block an air vent.
Can't be safer than the non-entertainment system from WV Up! that's just a built-in head mount for your phone. Grab one with a large screen and it's the safest thing you can get. Android still has an auto mode for this where it controls the car's audio system through headless bindings, not sure this malware would target this, but just by being a simpler system chances are it's safer too
The duality of cybersecurity is interesting. Sometimes the high bar is cleared just to enable a low bar to go lower. Those PLCs monitoring water were ignored for a very long time because they couldn't click on ads. It took a war for them to become a target.
One more reason cars should not be internet connected. They last for decades and manufactures don't want to support their cars that long. Always proxy to a phone and the attack surface is limited to things that are updated.
Norton AntiVirus for your car ECU's. Protect your carfor just $220.95/month *
* Cars without subscription causes acceleration to be restricted to 60mph.
After discovering the new OLED televisions come with antivirus, I'm done with thinking technology will ever be secure.
Even from this perspective, it's pretty easy to make things more secure by having less technology. Have the infotainment system just be a blank canvas for Carplay or AA to display on (there does need to be a bit back and fourth, phone needs to send audio to car, car needs to send GPS, speed and state of charge to phone (not strictly necessary, but there are user benefits from the phone having this information). The car itself doesn't need a whole internet-connected general purpose computer attached to it, but doing that is an easy way for the manufacturer to supposedly add value.
Similarly, the LG kerfuffle could be solved by their monitors just being monitors, and not throwing in pointless extras that just broadens their attack surface. Monitors don't need to be general purpose computing devices either. I shouldn't have to worry about general computing problems, like getting infected with malware, outside of computers that obviously are general purpose (i.e. phone, desktop, laptop, and anything else I intentionally set up with foreknowledge of it being general purpose and internet-connected, like a Raspberry Pi).
Moreover, no CVE is associated with this claimed vulnerability. It's not even stated which Android version or automotive head-unit variant version is affected.
> “sources said”
Yes, that's how Wikipedia works. https://en.wikipedia.org/wiki/Wikipedia:Neutral_point_of_vie...
Remember that not that long ago viruses spread through floppy disks.
Today, people share USB sticks full of music from one car to another all the time. They also bring their music from their home car to a rental car and back.
Huh, how does that work anyway? And while we're at it, Apple CarPlay as well? Both can run wirelessly via Bluetooth, but BT is nowhere near capable enough to stream full bandwidth video?
Article does not say that.
Bottom line is that lots of HN commenters here, as is our wont, will see this as a platform bug with a hated rival and not a bad third party integration that introduced vulnerabilities.
Like, if it was a Linux-based edge system from some fly-by-night contractor, would you be OK with a headline like "Malware infects Debian based refrigerators"? What'd Debian do?
The two big things here in my mind are:
1. Android Automotive has gotten very popular since it provides so much and writing your own OS is very very hard and expensive as so many car makers found out
2. Aftermarket head units often use it (see #1) so it’s likely far easier to get out there than if you had to compromise Ford/VW/Volvo/whoever
The actually vulnerable system is a custom vehicle head unit that merely happens to be running a software stack based on AOSP. It's not even "Android" in a product marketing sense.
Again, it's like blaming Debian because some loon stuffed it in a wifi NAS or whatever and put a backdoor into their UI. It's insane.
People do pair them with their phones, though. I could imagine a future version of malware like this propagating laterally.
Just off the top of my head.
And doing that doesn't really interfere with also setting up and selling proxy endpoints
Also typical Android permissions still apply. The user would need to grant the malicious app contacts, call logs, etc permissions.
The only valuable thing there is the relatively 'clean' mobile connection... and this malware's dropping a residential proxy endpoint on the headunit to take advantage of it. Bonus points if the headunit is always connected and always powered up to a +12v rail in the car, that's free and always-on real estate!
Otherwise any car sitting unused for a week or two would have a dead battery.
[1]https://opengarages.org/handbook/ebook/ (chapter 9)
I believe the connection exists because the steering wheel buttons/iDrive talk to the original head unit over CAN.
This is available on standard OBD-II. Maybe, it is accessible over CAN?
I think partially as my mental model of both android auto and CarPlay is that they operate as a passthrough of my device rather than as an separate installation of the OS entirely (I wasn’t aware the head unit itself had the ability to install APKs independently).
Also, feel like John Gruber is going to have a field day with this one
Android Auto is the Google equivalent of CarPlay and runs on your phone.
It’s easy to confuse. Like watching Apple TV on your Apple TV in Apple’s TV app.
Unfortunately Android Auto already existed. So it’s confusing.
- Head units connected to CAN bus with bluetooth vulnerabilities allowing attacker to remotely activate locks and windows and sometimes even driving controls
- Unsecured CAN bus cables everywhere allowing cars to be stolen through headlights and behind mud guard flaps
- Keyless entry basically a shit show of faraday pouches
- OBD port allowing thieves to clone a full key in seconds
- Even cars in decent neighbourhoods have to use steering locks
Sorry but this is a fucking joke and the automotive industry is cancer.
At least Tesla actually bothers with user updates and production improvements, most other manufacturers just shit out the same model 5 years in a row with an extra cup holder and USB port (probably rootable) if you're lucky. That said, Tesla's insistence that everything be done by touch screen is dog shit.
All this and still for 99% of cars my iPhone stuck to the dashboard provides better maps and entertainment and yet they can't even make a fucking phone holder standard, not even a fucking mounting point so I don't have to block an air vent.
Simple. It hasn't.
The duality of cybersecurity is interesting. Sometimes the high bar is cleared just to enable a low bar to go lower. Those PLCs monitoring water were ignored for a very long time because they couldn't click on ads. It took a war for them to become a target.
Somehow I doubt it. They're ripe for ransomware attack.
Similarly, the LG kerfuffle could be solved by their monitors just being monitors, and not throwing in pointless extras that just broadens their attack surface. Monitors don't need to be general purpose computing devices either. I shouldn't have to worry about general computing problems, like getting infected with malware, outside of computers that obviously are general purpose (i.e. phone, desktop, laptop, and anything else I intentionally set up with foreknowledge of it being general purpose and internet-connected, like a Raspberry Pi).