Every so often, I hear from one source or another that self-driving cars are the way of the future, and that in a few decades, most, if not all cars will probably be self-driving. This has tremendous implications, for the economy, the environment, and for our lives. No more drunk drivers, no more accidents caused by distracted or drowsy drivers, and shipping costs look to go way down, as drivers will no longer be needed to move goods around.

There are, however, some serious drawbacks: for one thing, whole sectors of the economy may be reshaped as many driving jobs are eliminated. For another, self-driving cars require some very sophisticated hardware (which means that personal cars will probably go the way of the dodo thanks to the expense and the fact that services like Uber will be early adopters) and software. The algorithms that control the cars will have to account for a great many eventualities, including choosing who lives and dies in the event of a crash.

That’s not what bothers me, though. Those are legitimate issues, but my present concern is with security.  In order for the self-driving cars to operate, they must be capable of ‘seeing’ everything around them, whether through sonar, lidar, or some similar system. That includes other cars, naturally, though it would be advantageous if the cars could talk to one another via an internet of some sort. That’s where the problem comes in. Anyone who has been paying attention lately knows about the so-called “internet of things” where common household appliances are connected to the internet for…reasons. The problem with this, of course,  is that if it’s connected to the internet, somebody’s gonna hack it, and the security for these appliances is decidedly sub-par.

Now, the stakes are pretty low if your smart-fridge gets hacked-some hacker might learn what groceries you buy. For a self driving car, however? The stakes are much higher. If a hacker gained access to a car’s source code, they could do some truly terrifying things: deliberately crash cars into one another, hijack cars and ransom their passengers’ safety, or kidnapping via redirecting the car.

It’s possible that the self-driving cars will not be connected to the internet, or that the navigation system will be isolated in some way from the wider internet, in which case the problems I described will be less acute. It’s also possible that the cars will have much better security, since each car is a huge investment, and there is so much more at stake. But given the track record of these sorts of technologies, I wouldn’t bet on it happening at first. It will take either a terrorist attack using these cars, or a series of kidnappings or accidents to happen, before the corporations involved move to improve their security.  Until then, I think I’ll stick with my old-fashioned, non-smart car.