Hacking Traffic Lights Barely Increases Crash Risk — Your Fellow Drivers Do

Self-driving and connected cars can't always "see" a traffic light the way you do. Many rely on wireless messages from the intersection telling them exactly when the light will turn red. This study asked a simple but scary question: what happens if someone jams that wireless channel on purpose?

To find out, researchers built a computer simulation of a real signalized intersection, mixing human-driven cars with automated ones at different levels (10% to 90% automated). They also modeled four driver personalities, from cautious to aggressive, and tested two ways an automated car can react when it loses its signal: pull over safely (called a minimal-risk maneuver) or just keep driving using cruise control.

The headline result is reassuring in one way and unsettling in another. Aggressive human drivers did not make the attack dramatically worse. They only notably affected one measure: hard-braking episodes forced on nearby drivers jumped from nearly zero to about 8 per 1,000 vehicle-seconds — a modest bump. Crucially, driver temperament on its own was the real danger. Going from cautious to aggressive surroundings changed baseline risk 13 to 18 times over, a far bigger effect than the attack itself.

One finding looks like good news but isn't. Aggressive driving appeared to reduce rear-end collisions during the attack, but only because the traffic jam cleared the queues those drivers normally build up. That's a traffic-flow benefit, not a safety gain. The practical takeaway: you can't design safety systems assuming everyone around you drives politely, and you can't assume the internet-connected car next to you will always get its signal.