Waymo is pushing back against claims that its robotaxis are secretly driven by remote operators. After testimony from a company executive went viral during a Senate hearing, Waymo's head of global operations Ryan McNamara sent a detailed letter to Sen. Ed Markey (D-MA) on Tuesday spelling out exactly what its remote assistance team does - and doesn't do. The disclosure reveals the company employs roughly 70 remote agents split between the US and the Philippines, but insists they only provide guidance when the autonomous system explicitly asks for help.
Waymo found itself in damage control mode this week after a Senate hearing sparked widespread confusion about how its self-driving taxis actually work. The Google-owned autonomous vehicle company revealed it maintains a team of 70 remote assistance agents - with half based overseas in the Philippines - but insists these workers aren't steering cars from computer screens.
The controversy erupted when Ryan McNamara, Waymo's head of global operations, testified before the Senate about the company's remote assistance operations. Video clips from the hearing quickly went viral on social media, with many interpreting the testimony as proof that Waymo's supposedly autonomous vehicles rely heavily on human intervention.
In a letter sent to Sen. Ed Markey (D-MA) on Tuesday, McNamara attempted to set the record straight. The remote agents "provide advice only when requested by the automated driving system on an event-driven basis," the letter states. They don't take control of the vehicle or make real-time driving decisions - instead, they help the AI navigate ambiguous situations it can't resolve on its own.
The distinction matters. Critics have long questioned whether robotaxi companies oversell their autonomous capabilities while quietly relying on human backup. But Waymo argues its system is fundamentally different from remote driving services where humans actively control vehicles. When a Waymo vehicle encounters something unusual - construction zones with unclear lane markings, confusing traffic patterns, or blocked routes - it stops and asks for high-level guidance rather than transferring control to a remote operator.
