There's ongoing work in adversarial examples for neural networks. This could conceivably be used to exploit self-driving cars, making it see things that aren't

There's ongoing work in adversarial examples for neural networks. This could conceivably be used to exploit self-driving cars, making it see things that aren't there.