Robotaxi Braking Injuries Surge: Waymo and Zoox Test Drivers Hurt as Autonomous Fleets Scale

TL;DR

• A TechCrunch analysis of federal OSHA filings found Waymo and Zoox test drivers suffered more than two dozen sprains, whiplash, and other pain injuries in 2024 and 2025 linked to hard braking and abrupt autonomous maneuvers.
• The injuries highlight a core robotaxi engineering challenge: autonomous systems are programmed to brake aggressively for perceived obstacles, phantom objects, and unpredictable traffic, jolting occupants inside.
• As both companies rapidly scale driverless operations to new cities, the findings raise fresh questions about in-vehicle safety, regulatory oversight, and the physical toll of refining autonomy with humans still in the loop.

The Human Toll Behind the Wheel

For years, the safety conversation around robotaxis has focused on what happens outside the vehicle — collisions with pedestrians, cyclists, and other cars. A new analysis of workplace safety data is shifting attention inside the cabin.

According to a TechCrunch review of Occupational Safety and Health Administration filings, safety drivers and test operators for Waymo and Zoox reported more than two dozen injuries over the past 18 months caused not by crashes, but by the vehicles themselves. The common thread: sudden, hard braking and sharp, unexpected maneuvers while the autonomous system was engaged.

The reports describe a pattern of soft-tissue injuries that rarely make headlines but are increasingly common as testing mileage explodes. While neither company has reported fatalities or serious collisions involving these specific incidents, the filings offer a rare, unfiltered look at the physical strain of being a human supervisor for an AI driver.

Inside the OSHA Data: What Two Dozen Injuries Reveal

TechCrunch's analysis focused on OSHA Form 300 logs and incident reports filed by Waymo and Zoox in California and other testing states through 2024 and into early 2025. In total, the filings document more than two dozen cases classified as sprains, strains, whiplash, neck and back pain, and contusions.

The narratives are strikingly similar. In multiple Waymo cases, operators described being thrown forward against seatbelts and harnesses when the vehicle executed an emergency stop for a perceived hazard. Several Zoox reports detailed neck strain and lower back pain after the purpose-built, bidirectional vehicles braked hard at intersections or during unplanned evasive maneuvers.

Most injuries were categorized as minor and resulted in days away from work or restricted duty rather than hospitalization. But the frequency is notable. The reports span both companies' major test hubs, including San Francisco, Phoenix, Los Angeles, and Las Vegas, and coincide with a period of intensive mileage accumulation and expanded rider-only operations. Safety experts note that OSHA filings likely undercount the true total, as they only cover employees and contractors, not passengers in fully driverless rides.

Why Robotaxis Are Braking So Hard

The abrupt stops are not a malfunction in the traditional sense. They are a feature of how autonomous driving systems are currently tuned to handle uncertainty.

Autonomous vehicles are designed to be hyper-cautious. Unlike a human driver who might gently slow down or nudge around an ambiguous object, an AV's planning stack will often choose the most conservative action: maximum braking. Engineers call this a "safety-first" or "defensive" policy, and it is intentional.

Several factors trigger these events:

Phantom Braking for Ghost Objects: Lidar, radar, and camera systems can misclassify steam, shadows, plastic bags, or reflections as solid obstacles, prompting an instant stop.

Unprotected Turns and Cut-Ins: In dense urban traffic, the system may predict that an oncoming car, jaywalking pedestrian, or double-parked delivery van could enter its path, initiating a hard brake even if the human supervisor sees no immediate threat.

Rear-End Avoidance Trade-Offs: To avoid any possibility of a forward collision — the most heavily penalized event in AV safety scoring — the system accepts a higher rate of harsh deceleration, even if it increases discomfort for occupants and risk of being rear-ended.

Former AV engineers say the problem is exacerbated as fleets move from well-mapped, sunny suburbs to chaotic downtown cores. More edge cases mean more conservative decisions, and more conservative decisions mean more jolts.

The Scale-Up Problem: More Miles, More Jolts, More Risk

The timing of the injury surge matters. Both Waymo and Zoox are in the midst of their most aggressive expansion yet.

Waymo now provides fully driverless paid rides in San Francisco, Los Angeles, Phoenix, and Austin, with testing underway in additional metros, and reports hundreds of thousands of trips per week. Zoox, owned by Amazon, has moved from closed-course testing to public operations in San Francisco and Las Vegas with its custom, steering-wheel-free robotaxi, and is preparing for a broader commercial launch.

Every new city, weather condition, and traffic culture introduces new scenarios that require tuning, which in turn requires more supervised testing miles with humans on board to monitor and take over if needed.

That creates a difficult loop for the industry. To make robotaxis smoother, companies need more real-world data. To get more data, they need more test drivers enduring the current, jerkier version of the technology. And as driverless services scale to the general public — including elderly riders, children, and passengers not wearing shoulder harnesses correctly — the tolerance for abrupt braking drops dramatically. A hard stop that causes a trained test driver's neck strain could cause a more serious injury to an unprepared passenger holding coffee or not fully braced.

Regulators and labor advocates are taking notice. While the National Highway Traffic Safety Administration tracks crashes involving autonomous systems, in-cabin injuries from non-collision braking events fall into a gray area between traffic safety and workplace safety, often escaping public scrutiny until OSHA logs are reviewed.

What Happens Next

Neither Waymo nor Zoox has disputed the filings. In statements to TechCrunch, both companies said safety — for the public and for employees — remains their top priority, and that they continuously refine their driving behavior to reduce harsh braking while maintaining collision avoidance.

Waymo has previously said it measures "ride comfort" metrics including hard braking and lateral acceleration, and uses them to train smoother driving models. Zoox has emphasized the safety benefits of its vehicle design, which includes four-point harness-like seatbelts and carriage-style seating designed to better handle sudden stops.

But smoothing the ride without compromising safety is one of the hardest problems in autonomy. Make the car too hesitant and it freezes at intersections. Make it too assertive and it risks a collision. Make it too comfortable and it may brake too late.

For now, the OSHA data provides a crucial reality check: even as robotaxis log millions of autonomous miles and tout superhuman crash-avoidance records, the technology is still rough enough to physically hurt the very people tasked with perfecting it. How quickly Waymo, Zoox, and their competitors can solve the hard-braking problem may determine not just worker safety, but whether the general public embraces driverless rides as truly ready for everyday use.

AndroGuider Team

Articles written by the AndroGuider team. We try to make them thorough and informational while being easy to read.