Texas Regulators Aren’t Told When Robotaxis Crash

This is an opinion piece, and is intended to provide analysis and commentary on current events and issues facing personal injury law. While resources are provided to support statements, opinion statements should not be misconstrued as fact, evidence, or argument in favor of or against any particular legislation. This is not legal advice.

Waymo opened fully driverless rides to the public in 2020 in the desert city of Phoenix, Arizona. Their service in Austin didn’t begin until around five years later. Yet federal crash reports list 199 crashes involving driverless vehicles in Austin from June 2025 through August 2026, compared with 112 in Phoenix. Only San Francisco and Los Angeles have more.

This summer, the Texas Senate told its Transportation Committee to quantify the impact of autonomous vehicles on traffic collisions. In August, the agency that licenses those vehicles testified before that committee that state law doesn’t require the companies to tell it when one crashes.

The regulator that’s responsible for regulating autonomous vehicle crashes can’t even tell you how many of those crashes have happened. How do you regulate crashes you aren’t counting?

Texas Became the Test Track

According to National Highway Traffic Safety Administration data, roughly one out of every five driver-less vehicle crashes reported nationwide happened in Texas between June 2025 and August 2026 (321 of 1,538). Only California has a bigger share of the wrecks over that span.

Worse, our share keeps climbing. Texas logged 26 of these crashes in the third quarter of 2025 and 92 in the second quarter of 2026. A year ago, about one in eight U.S. robotaxi crashes happened here. Now it’s about one in four.

Texas has also become a launch pad for driverless vehicles, and the state has courted the role. TxDOT keeps a welcome page for autonomous vehicle companies that calls Texas “an ideal setting for AV technologies to thrive.” Aurora began running driverless trucks between Dallas and Houston in May 2025. Tesla launched its robotaxi service in Austin that June. Waymo, already in Austin, began service in Dallas, Houston, and San Antonio in February 2026.

Some companies are testing almost nowhere else. Thirty-five of Tesla’s 36 reported robotaxi crashes happened in Texas. So did all 81 of Avride’s. Fifteen companies now hold state authorization to run driverless vehicles commercially, including freight carriers. Eleven of the 18 driverless heavy-truck crashes in the federal data happened on Texas roads, and those bring their own evidence problems for injured families.

Austin has tried to fight back on behalf of its residents. City records show at least 236 complaints about driverless cars from July 2023 through mid-March 2026. But state law bars Texas cities from regulating these vehicles on their own. Cities can count the wrecks, but only the state government can act.

The Only Detailed Record Belongs to the Company

Almost everything the public knows about these crashes comes from NHTSA’s Standing General Order, which requires companies to report crashes involving their automated systems. The companies fill out the reports and write the crash descriptions. They can also ask NHTSA to withhold some details as confidential business information.

In 35 years of practice, almost every crash case I’ve taken to court usually started with the police report and built from that information. In Texas robotaxi crashes, there usually isn’t one.

A police agency is listed as investigating only 47 of the 321 Texas crashes, less than one in six. A police report was available in 15 of those 47. By contrast, the self-driving car’s operating company had video of 315 of 321 crashes, and vehicle data for 281 of them.

For most robotaxi crashes in Texas, no officer took a statement or drew a diagram. The detailed account sits on a server owned by the company whose vehicle was involved.

Texas Has a Checkbox, Not a Count

Texas crash reports can flag a driverless vehicle, and the state deserves credit for that. Since April 2023, the police crash report form, the CR-3, has included fields for autonomous vehicles and the automation level engaged. Anyone can search those fields in TxDOT’s public crash database.

However, that checkbox only works when an officer works the crash. If police investigate fewer than one in six robotaxi crashes, the state’s database starts out blind for most of them. It also depends on officers recognizing the vehicle and coding it correctly. Officers are human and make mistakes accordingly, and nobody has published an audit of how often these details are recorded accurately.

That puts the Senate committee in a bind. Its charge asks lawmakers to quantify how these vehicles affect collisions. The state’s own records can’t do it.

The Regulator Learns From the News

SB 2807 put the Texas Department of Motor Vehicles in charge of authorizing driverless commercial operation. TxDMV can restrict, suspend, or revoke that authorization when a vehicle’s operation has caused or is likely to cause serious bodily injury.

Neither the statute nor TxDMV’s rules requires a company to tell the agency when that happens, though. As reported by NBC 5, TxDMV’s enforcement director told senators “The current statute as written does not have a mandate for reporting to the department.”

Cities asked for more. Dallas wanted operators’ crash histories. Austin asked which crash data, including federal reports, companies must share with cities. TxDMV refused both as outside its rulemaking authority. State law, it wrote, does “not allow the department to require” authorization holders to submit crash history.

In August, TxDMV learned from media reports that a Waymo had been involved in a fatal crash in Dallas. Police say an SUV struck the pedestrian first and threw him into the Waymo, and the investigation remains open. Waymo says it called the agency the next morning. Nothing required it to.

TxDMV can pull an authorization over serious injury. Serious-injury crashes are exactly the ones no one has to report. What is the agency supposed to act on if they don’t? What’s stopping a company facing potentially serious liability from conveniently choosing not to inform regulators of what happened?

As of the hearing, TxDMV had never restricted, suspended, or revoked an authorization. Its staff told senators they hadn’t identified any needed legislative changes. Respectfully, I’ve identified one.

The Case Against Another Report

To a certain degree, federal reporting already exists. NHTSA requires a five-day report when a crash involves a death, a hospital trip, an airbag deployment, a tow, or a struck pedestrian or cyclist.

The reports that are created by this limited requirement would still not be sent to the agency that can act. The companies write the reports, NHTSA posts them monthly, and TxDMV, the only agency that can pull a Texas authorization, receives none of them directly.

Sending TxDMV a copy of a report a company already files costs almost nothing, and one form sent to two places is not a patchwork. California’s new rules, finalized this spring, take that approach and align state crash reporting with the federal format.

Self-driving vehicle companies don’t want a different rule in every state; compliance gets far more difficult at that point.

However, Texas Rep. Craig Goldman called the current mix a “chaotic web” of state laws at a February markup. Congress seems to have a similar opinion and is working on a national system through the SELF DRIVE Act, which would create a federal crash-data repository for automated vehicles. This bill cleared a House subcommittee 12 to 11 in February and has not moved since. As introduced, it would bar states from requiring their own crash reports and give companies at least 30 days to report a serious crash, compared with five today.

This not only doesn’t address some of Texas’s major problems in reporting, but it actually makes a few mechanisms worse. Reporting gets slower, and Texas loses the power to fix it entirely. This is why it’s critical that Texas lawmakers act now.

What Texas Should Do This Session

I’ll add one point in their favor: if driverless technology delivers what its backers promise, fewer families will face the devastation that a drunk or distracted driver can cause. I’ve spent my career representing the people those drivers hurt. Their stories are heartbreaking.

Texas invited this technology onto its roads, and it’s easy to see a future where more of these vehicles means a future with less crashes. But you can’t referee a game nobody is scoring.

The interim committees are expected to release their reports this winter, and lawmakers can start filing bills on November 9. Here is what I’d ask them to file:

• Require reporting to TxDMV. Any company holding SB 2807 authorization should send TxDMV a copy of every federal report on a Texas crash, on the same deadlines. Make it a condition of keeping the authorization.
• Make the Texas record public. The state’s crash database can’t reliably say how often driverless vehicles crash. Automation levels are easy to miscode, commercial service isn’t recorded, and crashes that aren’t documented by police never appear. TxDMV should publish what the companies report instead.
• Make the checkbox count. TxDOT should publish an annual count of crashes flagged as involving autonomous vehicles and compare it with what the companies report. If the numbers don’t match, lawmakers will want to know why.

None of this slows deployment. Each company sends one more copy of a report it already writes. The Senate asked how these vehicles affect collisions on Texas roads. TxDMV can’t answer that until someone tells it about the crashes.