Driverless cars face the safety test

Overview

Updated 7 days ago

Waymo says its driverless cars were involved in 95% fewer crashes that caused serious injury or death than human drivers, across about 270 million miles driven without a person at the wheel through June 2026. The company's September report covers Atlanta, Austin, Los Angeles, Phoenix and San Francisco, and puts injury-causing crashes 82% below the human rate.

Independent numbers arrived in July, when the Insurance Institute for Highway Safety found 68% fewer police-reportable crashes for Waymo than for human drivers in four cities. Not every measure points the same way: Los Angeles County crash reports rose as the fleet grew, and a June 2025 federal reporting change removed some minor collisions from the counts.

Why it matters

Car crashes kill about 40,000 Americans a year. This is the first independent proof that driverless cars might cut that toll.

Questions about this story

Free account needed to ask — your question is kept and asked for you right after sign-up. Answers are public.

No questions yet — be the first to ask.

Key Indicators

95%

Fewer serious-injury crashes

Waymo's comparison with human drivers across about 270 million driverless miles through June 2026.

82%

Fewer injury crashes

In Waymo's September 2026 analysis, equal to 841 fewer injury-causing crashes.

68%

Fewer crashes, outside study

The Insurance Institute for Highway Safety's July 2026 comparison of police-reportable crashes in four cities.

270M

Driverless miles analyzed

Waymo's safety data through the end of June 2026, 50 million more than its previous report.

41%

Reported LA rate drop

Reported Waymo crashes per million miles in Los Angeles County fell from 9.4 to 5.5; a 2025 federal reporting change accounts for much of the decline.

Voices

Curated perspectives — historical figures and your fellow readers.

People Involved

David Harkey

President, Insurance Institute for Highway Safety

Presented the study's findings

Tekedra Mawakana

Co-CEO, Waymo

Pointing to 270 million driverless miles of safety data as Waymo expands

Eric Teoh

Director of Statistical Services, Insurance Institute for Highway Safety

Lead author of the July 2026 Waymo crash study

Organizations Involved

Insurance Institute for Highway Safety

Nonprofit research group

Author of the independent safety study

An insurer-funded research group whose crash tests and safety ratings shape which vehicles carry industry awards.

Waymo

Autonomous vehicle company

Reports 95% fewer serious-injury crashes across about 270 million driverless miles

The Alphabet-owned company running the largest driverless taxi service in the United States.

National Highway Traffic Safety Administration (NHTSA)

Federal Agency

Investigating Waymo incidents

The federal regulator responsible for vehicle safety standards and crash investigations.

Timeline

October 2020 October 2026

10 events Latest: October 15th, 2026 · 2 weeks ago

Tap a bar to jump to that date

October 2026
Reported LA crash rate falls after federal rule change
Latest Analysis

Waymo's reported crash rate in Los Angeles County fell from 9.4 to 5.5 per million miles year over year, a 41% drop that The Basin attributes largely to NHTSA's June 2025 decision to stop requiring reports of some minor collisions.
October 15th, 2026
LA County Waymo crash reports triple in a year
Analysis

A Los Angeles Times review of federal crash data counts 69 Waymo crashes in the county in 2024 and 219 in 2025, with 186 more reported so far in 2026. More than 80% of this year's crashes caused no apparent injury.
October 5th, 2026
September 2026
Waymo claims 95% fewer serious-injury crashes
Study

Waymo's analysis of about 270 million driverless miles through June 2026 finds 95% fewer crashes involving serious injury or death than human drivers in five cities, and 82% fewer injury-causing crashes.
September 24th, 2026
July 2026
IIHS study finds Waymo 68% safer
Study

The Insurance Institute for Highway Safety reports Waymo cars were involved in 68% fewer crashes than human drivers.
July 23rd, 2026
Feds set emergency-response deadline
Regulatory

After dozens of Waymos stalled on July 4, NHTSA gives robotaxi firms until month's end to fix responder failures.
July 13th, 2026
June 2026
Waymo recalls about 3,900 robotaxis
Recall

Waymo issues a software recall after 13 cases of robotaxis entering active freeway construction zones.
June 18th, 2026
January 2026
NHTSA probes school-bus pass
Investigation

Federal regulators open an investigation after a Waymo robotaxi illegally passes a stopped school bus.
January 15th, 2026
October 2023
California pulls Cruise's permits
Regulatory

The California DMV suspends Cruise's driverless permits, and the company later winds down its robotaxi unit.
October 24th, 2023
Cruise robotaxi drags a pedestrian
Incident

A Cruise driverless car in San Francisco runs over and drags a pedestrian who was first struck by a human driver.
October 2nd, 2023
October 2020
Waymo opens driverless rides to the public
Milestone

Waymo begins offering fully driverless taxi rides to the public in Phoenix, the first such service in the U.S.
October 8th, 2020

Scenarios

Federal regulators write the first rules for driverless cars

Possible Resolves by End of 2027

Discussed by: Streetsblog USA, safety researchers citing the IIHS report

The study exposes a gap: the U.S. has no performance standard built for cars with no driver. If the data pushes Congress or NHTSA to act, the agency could open a formal rulemaking that sets crash-reporting or performance requirements for driverless fleets. Recent stalls and recalls add pressure to move.

Waymo passes one million weekly rides

Likely Resolves by End of 2026

Discussed by: Waymo, TechCrunch, robotaxi trackers

Waymo reported 500,000 paid weekly trips in March 2026 and plans to add roughly 20 cities this year, including Tokyo and London. If expansion holds and the safety story helps win city approvals, weekly volume could double. Recalls or a serious incident could slow the rollout.

A Waymo crash triggers a permit suspension

Unlikely Resolves by Jul 23, 2027

Discussed by: Safety advocates recalling the Cruise shutdown

The study counts 89 crashes and one Austin market where Waymo did worse than humans. A single high-profile injury or death, as with Cruise in 2023, could prompt a state regulator to pull Waymo's driverless permit in a major market and stall the safety narrative.

A second independent study confirms the safety edge

Possible Resolves by End of 2027

Discussed by: IIHS, insurance researchers

The IIHS itself flagged the small sample and called for better federal monitoring. As Waymo logs more miles, a follow-up study, from IIHS or another independent body, could test whether the safety gap holds at larger scale. A confirming result would strengthen the case for wider deployment.

NHTSA tightens robotaxi crash reporting

Possible Resolves by Q2 2027

Discussed by: IIHS, The Basin, Los Angeles Times

The IIHS used its July study to call for a standardized national system for automated-vehicle crashes and mileage, saying current data cannot support close monitoring of a much larger fleet. The Basin's October analysis found that a June 2025 rule change removed many minor collisions from the counts, making year-over-year rates hard to compare. A new reporting requirement would put company safety claims back on a common footing.

Historical Context

3 moments from history that rhyme with this story — and how they unfolded.

October 2023

Cruise pedestrian dragging (2023)

A human driver struck a pedestrian in San Francisco, throwing her into the path of a Cruise robotaxi. The driverless car ran her over and dragged her about 20 feet before stopping.

Then

California suspended Cruise's driverless permits within weeks. Executives resigned and operations halted.

Now

General Motors cut funding and wound down the Cruise robotaxi business in 2024, leaving Waymo as the clear U.S. leader.

Why this matters now

It shows how fast one incident can erase public trust. Waymo's strong crash data is the counterweight, but a single bad crash could still reverse the story.

2004-2011

Electronic stability control mandate (2007)

IIHS and federal researchers found electronic stability control cut fatal single-vehicle crashes by more than half. The technology automatically brakes individual wheels to stop skids.

Then

NHTSA issued a rule in 2007 requiring the system on new vehicles.

Now

Stability control became standard on all new U.S. passenger vehicles by the 2012 model year, preventing thousands of deaths.

Why this matters now

It is the clearest case of independent crash data driving a federal mandate. The Waymo study could start a similar path from evidence to rules.

2021-2024

Tesla Autopilot investigations (2021)

NHTSA opened probes into crashes involving Tesla's Autopilot driver-assist system, including cars striking parked emergency vehicles. Tesla marketed the feature as advanced but kept a human responsible.

Then

Tesla recalled about 2 million vehicles in 2023 to add driver-attention safeguards.

Now

The scrutiny sharpened the line between driver-assist systems, where a person must watch, and fully driverless cars like Waymo's.

Why this matters now

It shows regulators weighing safety claims against real crash data. The IIHS study gives the driverless side of that debate independent numbers.