Two stories dropped this week that belong in the same sentence. Waymo pulled over and called police on riders carrying a ghost gun in San Francisco, the robotaxi acting as both vehicle and informant. And separately, The Verge's Andrew Hawkins detailed how conventional cars already sell driver data to insurers, dealers, and data brokers without most owners realizing. Read them back to back and the picture that emerges is not about crime or privacy in isolation. It is about who owns the relationship between you and your machine.
The Car as Ambient Witness
The Waymo incident is framed as a safety win, and maybe it is. But the mechanism that makes it possible, always-on sensing, cloud-connected judgment, real-time communication with authorities, is identical to the mechanism that makes your Honda quietly log your hard-braking events and sell them to LexisNexis. The hardware is different. The logic is the same. A 2023 paper in the International Journal of Information Management by Nili et al. found that consumers dramatically underestimate the scope of vehicle data collection, rating it as less invasive than smartphone tracking despite cars collecting location, biometric, behavioral, and passenger data continuously. That gap between perception and reality is where the business model lives.
Obama's AI Plan Won't Touch This
Obama urged Democrats this week to make AI regulation a central agenda item, calling for a clear plan on safeguards. But the car-as-surveillance-node problem predates large language models by a decade. It is not an AI safety issue in the sense that Washington means when it says AI safety. It is a data-ownership issue, a consent architecture issue, a question of whether the thing you paid sixty thousand dollars for works for you or for the platform that built it. The robotaxi just makes the power dynamic legible by making it dramatic. Your car was already watching. It just hadn't pulled over yet.