A very broad question. Focusing the question down to just brakes and steering, for every innovation, to promote safety beyond what the average driver is capable of achieving, we create new vulnerabilities. These need to be addressed pro-actively.
One example. We used to feel secure in the knowledge that the hydraulic brake system was completely isolated from other car control systems, and that nothing could prevent braking in an emergency. Given that brakes easily overpower engines in cars, that was a decent safety feature. Same applied to the steering wheel. A mechanical link to the front wheels, nothing can get in the way of a driver's steering commands.
Neither of those comfortable truisms hold anymore. Now, to prevent panicked drivers from locking the wheels, we have ABS. Which means, there's something out of our control that prevents the brakes from working. Yes, it's supposed to prevent brakes from working only as long as the wheels are stopped and the car is still traveling, but nevertheless, a new failure mode exists.
Same with steering. As a means of stability control, either the steering mechanism itself, or brakes automatically applied at individual wheels, will override the driver's steering commands. Another vulnerability introduced.
Car companies have not historically been software companies, but they are finding it necessary to introduce those skills, big time. Innovations required to improve safety, because the vast majority of drivers are anything but expert, bring on new vulnerabilities. There is no free lunch.
Just like anything, the first ones are not the zenith. Though they may be very good, there is still much work to do. Regarding being secure, that is the most vexing challenge.
In my mind the best security is a system for which there are no current models. Hardware that does not exist otherwise, paired with an OS running on a new machine language other than 0s and 1s, designed just for that hardware, creating, in effect, a Black Box. For it is quite difficult to crack that which is unknown, has not existed prior and which one cannot visualize in any way.
Having a smart and safe & secure vehicle seems to be out of reach based on the current situations.
For example, we could ensure that the car secure enough to not move if the key was stolen by someone else ( secure and smart), but what will happen if the driver is sick and not able to drive and we need to take him to the hospital using his car, in this situation the car will not move ( secure, stupied and not safe).
we could argue that we can design a vehicle which ensures each one of these characteristics, but having all of them together need more work and research, intensive collaboration between the various involved parties, new methods and models, and the main thing is equality in the recognition of the importance of these characteristics.
Till now, people are more concern about the safety. They still believe that the security is a product which could be installed before shipping the vehicle!