Texas logged its first robotaxi fatality — the Waymo was the third vehicle in the sequence

Waymo reported a fatal collision in Dallas to NHTSA. According to the report, the Waymo vehicle was slowing for a pedestrian in the roadway when an SUV struck that pedestrian, throwing them into the Waymo's rear left side. Juan Guzman, 71, was taken to hospital and died. The SUV driver remained at the scene; police have announced no arrests or charges and the investigation continues. It is the only Texas fatality recorded in NHTSA's robotaxi crash database since June 2025. Texas robotaxi operators separately reported 21 new crashes to federal regulators in the latest period, with Austin logging eight as fleets there expanded.

Why it matters

"Texas records first fatal autonomous vehicle collision" is accurate and tells you almost nothing. A fatality occurred; an autonomous vehicle was involved; and on the reported facts the autonomous vehicle was slowing for the pedestrian a human driver then hit. Whether that is a story about robotaxi safety depends entirely on a causal order the phrase "involved in" erases.

This will keep happening, and it is worth naming the mechanism now rather than after the fifth instance. Robotaxis file crash reports with NHTSA as a condition of operating. Human-driven vehicles mostly do not, except where state law requires it and someone follows through. So a growing fleet of scrupulously-reporting vehicles operating in cities full of unreported human crashes will generate a rising count of "AV-involved" incidents regardless of how safely it drives — and that count is the only series anyone can cite, because it is the only one that exists. The reporting asymmetry manufactures a trend.

None of which makes the disclosure regime wrong. It is the right design, and we have argued the opposite failure — that Tesla's telematics reached reporters before investigators in two freeway deaths — is the more serious problem. A regime that produces awkward headlines while working is preferable to one that produces clean headlines because nobody files. But readers, regulators and this newsroom should be reading the 21 crashes and the one fatality as reports filed, not as harm caused, until a report says otherwise. In this case Waymo's own filing says otherwise.

What to watch

Whether police attribute fault in the Dallas case, whether NHTSA's database ever carries a comparable per-mile denominator, and whether any jurisdiction requires equivalent reporting from human-driven fleets.