Did you read the SA article? It's virtually impossible to hack a car in any kind of meaningful way. The Jeep in the piece that you linked to was owned by the hackers themselves and fucked about with for a year to enable them to do it.
If you can get hold of a certain type of car for long enough and install all kinds of software, you can cut it's transmission if it's connected to the internet. But you can't anymore because the manufacturer has fixed the bug.
Not much of a story really is it?
Might be worth going to the source.
Or if you find videos of nerds enthusing to other nerds about stuff that they've hacked too annoying to tolerate, here are their papers.
https://sm.asisonline.org/ASIS SM Documents/remote attack surfaces.pdf
https://securityzap.com/files/Remote Car Hacking.pdf
Basically the reason they need physical access to a sample of the car in question is to reverse-engineer the internal protocols by which the multitude of computers inside them communicate.
This is so they can gain access via one of these control units which is doing something relatively innocuous and pivot from there to a different control unit from which they can e.g. switch off the brakes.
They don't need to be able to install anything on the target vehicle to do the attacks in question, they just need to spend time playing around with a sample until they figure out how to fuck with it.
The particular conjunction of vulnerabilities and configuration issues and difficulty of exploiting them might change from model to model, but the security engineering economics of consumer products mean that in general, any determined group with appropriate skills is going to find a way to do this if they see a benefits case.
http://www.cl.cam.ac.uk/~rja14/Papers/moore-anderson-infoeconsurvey2011.pdf
As manufacturers add more and more features of this kind, the attack surface gets bigger. Indeed a couple of the more lethal exploits actually originate with features that are marketed as increasing safety by e.g. making braking more 'intelligent'.
Their discussion of what happened when they tried to communicate with the manufacturers does little to support breezy confidence in there being a timely response (unless there is an immediate threat that the exploit will be on TV.)
I think it's pretty clear from the detail in their papers that they were resource-limited, e.g. they had to find workarounds rather than spend a few thousand on specialised test gear for proprietary components.
Buying cars to reverse-engineer is likely not an issue for anyone who has a black budget to play with. Indeed, the spooks already have form for getting backdoors installed by the manufacturers into things they're interested in and there are a few chips from e.g. Texas Instruments that come up over and over again in this space.