US documents detail Tesla

Tesla’s robotaxi hype and Waymo’s carefully staged autonomy both lean on the same unglamorous backbone: people. Newly released US government documents reviewed by Wired detail how “self-driving” services are quietly propped up by remote operators and human “bystanders” who intervene when the software gets confused, stuck, or simply too cautious.

According to Wired, the records show that both companies have built operational pipelines for human assistance—sometimes in real time, sometimes asynchronously—despite marketing that implies the vehicle is the driver. The public story is AI; the safety story is escalation protocols.

The documents underscore a mundane but politically important reality: autonomy is not a binary. It’s a stack of automation plus a call center. When the system encounters edge cases—construction zones, ambiguous right-of-way, blocked lanes, emergency vehicles, unusual pedestrian behavior—the “robot” can request help, be guided, or be instructed to pull over. This shifts the real engineering constraint away from neural networks and toward latency, network reliability, operator tooling, and decision authority.

That, in turn, raises the question regulators keep dodging: who is responsible when a remote human touches the steering wheel by proxy? If the vehicle is in a mixed-control state—software executing, human advising, company supervising—liability becomes a shell game. The state’s preference, as usual, is paperwork that makes blame legible after the fact rather than rules that prevent the ambiguity.

Wired’s reporting also highlights how incident reporting can become a narrative-management exercise. Companies can categorize events as “minor,” “expected,” or “no safety impact,” while the public only sees the glossy aggregate: miles driven, disengagement rates, and selective footage. A remote intervention that prevents a crash is, depending on the PR need, either evidence of robust safety systems or an irrelevant non-event.

The promised future—cars that drive themselves so humans don’t have to—arrives first as a new class of low-visibility operators tasked with babysitting fleets at scale. Autonomy doesn’t remove labor; it relocates it, often into centralized operations where workers absorb risk and responsibility without the dignity of being called the driver.

The takeaway isn’t that autonomy is fake. It’s that the real system is more decentralized than the slogans admit: software, sensors, telecom infrastructure, human judgment, and corporate policy all share control. If regulators want to protect the public, they should stop treating “AI” as a magical legal category and start demanding clear chains of command, auditable intervention logs, and honest disclosure about when a human is effectively driving—just from a desk.

Until then, “robotaxi” will remain what it already is: a product demo with a hidden staffing plan.