This is a known attack vector for autonomously-driving vehicles. A 2018 paper titled "Robust Physical-World Attacks on Deep Learning Visual Classification" [1]

This is a known attack vector for autonomously-driving vehicles. A 2018 paper titled "Robust Physical-World Attacks on Deep Learning Visual Classification" [1] tested what it called "adversarial perturbations" on the most important of all traffic signs: the STOP sign. By slapping black and white stickers a couple square inches in size on the STOP sign, the authors managed to achieve a 100% misclassification of the STOP sign into 45 MPH speed limit one. To the human eye, the STOP sign is clearly visible and perceivable. In theory, people owning autonomously-driving vehicles could be assassinated this way and nobody would be the wiser. 1. https://arxiv.org/pdf/1707.08945.pdf https://arxiv.org/pdf/1707.08945.pdf