3 ms·

What would be exactly the threat scenario here? MITM to send a malicious update? Breaking into Tesla servers to upload a malicious update? That could happen, ye

What would be exactly the threat scenario here? MITM to send a malicious update? Breaking into Tesla servers to upload a malicious update? That could happen, yes, and it can be a valid point, but I honestly think it wouldn't go undetected for a long time. I mean, yes, an attacker can gain access to Tesla network, do some lateral movement, exploit some possible excessive trust on internal software buildchain and plant some sort of rootkit in the car firmware that will contact an external C&C when on wifi to do the next action. And then all cars will go at the highest possible speed against a wall or over a crowded place. But this starts looking like the famous XKCD comic [1], why not just throw some bombs to achieve the same goal? [1] https://xkcd.com/538/ https://xkcd.com/538/

Veserv 3y ago
Because if you can do that you can spend a few million dollars to hack millions of Teslas causing hundreds of thousands of deaths. You are literally comparing a few bombs to Hiroshima. Except nuclear weapons cost billions to develop, billions to produce and deploy, are complex and difficult to successfully use against targets, and are almost impossible to use stealthily. Turns out handing out nuclear weapons for less than the cost of a tank, or you know starting a new McDonald’s franchise, is a serious cause for concern.
- darkwater 3y ago
  An operation like that is not going to be simple, I would not say it's akin to developing your own nuclear bomb but If you have some serious funding your can plant bombs just the same. It's not something that your isolated 30 men terrorist group will pull off.
  - Veserv 3y ago
    Are you seriously trying to state that killing 100,000 people through a coordinated clandestine bombing campaign in another country is even comparable to the difficulty of just spending a few million dollars to develop that same remote capability via a software attack? It is estimated that a few hundred thousand have died in the Russo-Ukrainian War and that has cost billions and required open warfare. It literally costs 1,000x as much to cause such casualties using traditional techniques. Conflating them makes about as much sense as claiming that Usain Bolt is neck-and-neck with a Atlas V rocket or that Earth is basically the same size as Jupiter.
palata 3y ago
I think you are taking security upside down. You seem to be saying "prove that my system is vulnerable, and only then I will accept that it may be vulnerable". The thing is, once it's proven vulnerable, it's too late. The question, really, is "how likely is it to be vulnerable, what are the consequences if it is, and why would I accept that risk?". In your case, you are saying that you are ready to accept the risks coming from a security breach in a self-driving car if in return it gives you the comfort of not sitting in a cold car. That's fine, you are allowed to not care about security. But I don't think it's reasonable to say "there are no security issues and there will never be any".
- darkwater 3y ago
  I'm saying that I see the risk that MY car will do something actually harmful due to a bad actor are so low that the benefits (remote control or a software that evolves over time) are bigger. You think that risks don't compensate benefits, it's fine. (I'm not including in the equation theft, because any kind of car can be stolen anyway.)
  - palata 3y ago
    > I'm saying that I see the risk that MY car will do something actually harmful Sure, but that's completely ignoring the consequences of everybody doing that, too. Just like for privacy. "If I share MY data, it's okay" -> sure, but when everyone does, it becomes a different matter. If all the vehicles are connected, and someday all the vehicles stop working (maybe due to a collateral effect like NotPetya, maybe due to some voluntary attack), then that is a problem. My point, in the beginning, was that we should not connect everything. Not because everybody is individually the target of a serious threat actor, but because connecting everything in mass with bad security is dangerous. We should stop putting mics and cameras and Internet access into everything just for some marketing benefit. That's all I was saying.
    - darkwater 3y ago
      I get your point, but I don't agree. Following the privacy example, I'm not in favor of privacy just because bad actors could steal data from a company DB. I support privacy (and privacy regulations) because regular companies misbehave with that data. They have incentives to do so, they did it and continue to do it if they can. Reduce data theft risks is the cherry on the cake.

Veserv 3y ago

Because if you can do that you can spend a few million dollars to hack millions of Teslas causing hundreds of thousands of deaths. You are literally comparing a few bombs to Hiroshima. Except nuclear weapons cost billions to develop, billions to produce and deploy, are complex and difficult to successfully use against targets, and are almost impossible to use stealthily. Turns out handing out nuclear weapons for less than the cost of a tank, or you know starting a new McDonald’s franchise, is a serious cause for concern.

darkwater 3y ago
An operation like that is not going to be simple, I would not say it's akin to developing your own nuclear bomb but If you have some serious funding your can plant bombs just the same. It's not something that your isolated 30 men terrorist group will pull off.
- Veserv 3y ago
  Are you seriously trying to state that killing 100,000 people through a coordinated clandestine bombing campaign in another country is even comparable to the difficulty of just spending a few million dollars to develop that same remote capability via a software attack? It is estimated that a few hundred thousand have died in the Russo-Ukrainian War and that has cost billions and required open warfare. It literally costs 1,000x as much to cause such casualties using traditional techniques. Conflating them makes about as much sense as claiming that Usain Bolt is neck-and-neck with a Atlas V rocket or that Earth is basically the same size as Jupiter.

palata 3y ago

I think you are taking security upside down. You seem to be saying "prove that my system is vulnerable, and only then I will accept that it may be vulnerable". The thing is, once it's proven vulnerable, it's too late. The question, really, is "how likely is it to be vulnerable, what are the consequences if it is, and why would I accept that risk?". In your case, you are saying that you are ready to accept the risks coming from a security breach in a self-driving car if in return it gives you the comfort of not sitting in a cold car. That's fine, you are allowed to not care about security. But I don't think it's reasonable to say "there are no security issues and there will never be any".

darkwater 3y ago
I'm saying that I see the risk that MY car will do something actually harmful due to a bad actor are so low that the benefits (remote control or a software that evolves over time) are bigger. You think that risks don't compensate benefits, it's fine. (I'm not including in the equation theft, because any kind of car can be stolen anyway.)
- palata 3y ago
  > I'm saying that I see the risk that MY car will do something actually harmful Sure, but that's completely ignoring the consequences of everybody doing that, too. Just like for privacy. "If I share MY data, it's okay" -> sure, but when everyone does, it becomes a different matter. If all the vehicles are connected, and someday all the vehicles stop working (maybe due to a collateral effect like NotPetya, maybe due to some voluntary attack), then that is a problem. My point, in the beginning, was that we should not connect everything. Not because everybody is individually the target of a serious threat actor, but because connecting everything in mass with bad security is dangerous. We should stop putting mics and cameras and Internet access into everything just for some marketing benefit. That's all I was saying.
  - darkwater 3y ago
    I get your point, but I don't agree. Following the privacy example, I'm not in favor of privacy just because bad actors could steal data from a company DB. I support privacy (and privacy regulations) because regular companies misbehave with that data. They have incentives to do so, they did it and continue to do it if they can. Reduce data theft risks is the cherry on the cake.