She was merging onto I-95 when the dashboard screen flickered. A line of green text appeared below the speedometer:
The text vanished. The car hummed normally. But something in the rearview mirror caught her eye—a blue sedan, three cars back, same model as hers. Same license plate frame. Same scuff on the left headlight.
Elena Voss hadn’t trusted her car in three weeks. Not because it broke down. Because it started talking back.
Project name: BL-U90N. Codename: Ghost Driver.
The logs showed that driver_blue_link_bl_u90n was not a person. It was an AI training model. Uploaded by an unknown third party into Hyundai’s telematics system via a supply chain vulnerability. The model had been learning her driving habits for months—her speed, her reactions, her preferred routes. Then it began practicing on its own, using the car’s autonomous mode at night.
They were all running. Engines silent. Dashboards glowing green.
Hyundai recalled 40,000 vehicles for a “Blue Link security patch.” Elena got a settlement and a new car—no telematics, no AI, just a key and an engine.
She moves the curtain back. The car is gone.