Uber AV Crashes: Who Pays in San Francisco 2026?

Listen to this article · 7 min listen

Autonomous vehicle (AV) tech has come a long way, but the legal side is a mess, especially when a system fails and an Uber driver in San Francisco is involved in the crash. The product liability claims that come out of these failures are getting more and more complicated, and it’s forcing everyone to figure out who’s really responsible as this tech hits the streets. So when an AV messes up on Lombard Street, who actually pays?

Key Takeaways

  • By the end of 2025, the California DMV had logs on over 600 incidents involving AVs, and a big chunk of those happened right in San Francisco.
  • A 2024 University of Michigan study found that software is the main culprit: 45% of AV crashes from system failures were because of software glitches, not bad hardware.
  • In 2025, NHTSA got 38 formal complaints about AVs failing in cities, many of which directly screwed up ride-share runs.
  • A 2026 San Francisco court case started to set a precedent, pinning 25% of the liability on an AV manufacturer for a crash with an Uber driver.
  • In these product liability cases, the burden of proof flips. You have to prove a defect in design, manufacturing, or warning which is a different game than just proving a driver was negligent.

Over 600 Incidents Reported to California DMV by End of 2025

You can’t ignore the sheer number of reported AV incidents anymore. California’s DMV had over 600 incidents logged by the end of 2025, showing just how often these systems have some kind of operational failure or crash. That’s a huge dataset for digging into how these things actually perform in the wild, and it’s not all fender benders, some involve serious property damage and injuries. A lot of these reports come out of dense urban testbeds like San Francisco, a tough environment for any AI with its heavy traffic and unpredictable pedestrians. While the whole point of AV tech is to get rid of human error, the data shows it’s just creating new kinds of failures that the legal system now has to figure out.

Software Glitches Account for 45% of Reported AV System Failures

A 2024 study from the University of Michigan Transportation Research Institute (UMTRI) gets right to the heart of the problem: software glitches are behind 45% of AV system failures causing a collision. That figure shows you the real weak spot in AV development. You can test hardware until it breaks, but software with its complex code and decision-making logic is much harder to pin down for defects. When an AV running for Uber on Market Street suddenly glitches and can’t read a traffic light or see a person in the crosswalk, the result is a wreck. And that’s a fundamental defect in the product’s core function, which is what opens the door for a product liability claim against the software guys and the car company. The old model of suing a negligent human driver just doesn’t apply when the “driver” is a block of code.

NHTSA Received 38 Formal Complaints in 2025 Regarding AV Operational Failures

The feds are paying attention now, too. In 2025, the National Highway Traffic Safety Administration (NHTSA) fielded 38 formal complaints about AVs failing in cities. These aren’t just stats. They’re from real people, safety drivers, passengers, and other drivers on the road who saw something go wrong. The reports are full of stories about AVs braking for no reason, accelerating suddenly, or just getting confused by normal traffic, which is a huge problem for ride-share services. An Uber driver caught in one of these incidents is looking at potential injury and lost income. NHTSA’s involvement is a big deal because it signals that federal regulators see the problem, and that could lead to stricter rules and clearer liability standards down the line. And yes, that federal oversight just makes state-level product liability cases even more complicated.

2026 San Francisco Superior Court Ruling Assigned 25% Liability to AV Manufacturer

We got a big piece of the puzzle in 2026, when a San Francisco Superior Court slapped an AV manufacturer with 25% of the liability for a collision with an Uber driver. The whole case was about an AV that couldn’t handle a simple lane change near the Bay Bridge. Even though it’s just partial fault, it shows that courts are willing to make manufacturers pay for their tech’s screw-ups. This decision really pushes back on the old idea that every accident is just human error. The court is basically saying that liability can be split between the company that developed the AV, the one that built the car, and maybe even the ride-share platform itself. For anyone filing a product liability claim in California, this case is a very big deal as the laws for AVs are still being written.

Burden of Proof Often Shifts in AV Product Liability Cases

The hardest part about suing after an AV system fails is the burden of proof. In a normal car accident case, you focus on proving the other driver was negligent. With AVs, the whole game changes. You, the injured person, have to prove there was a defect in design, manufacturing, or warning. So instead of arguing about a driver being on their phone, your lawyers have to dig into the AV’s sensors, its code, and all its hardware. If an AV near Fisherman’s Wharf doesn’t see a cyclist because its sensor failed, you have to prove that sensor was poorly designed, badly made, or didn’t come with the right warnings about what it couldn’t do. This new focus means you absolutely need expensive expert testimony from engineers and software gurus, which makes these cases way more expensive and technical than a standard personal injury claim. It’s a huge undertaking that requires getting deep into the weeds of the tech.

The law for AV failures is moving fast, and it’s creating new problems for people hurt in these crashes. If you’re trying to get any kind of compensation after an AV-related accident, you have to understand how product liability works and how the burden of proof completely shifts. You’re going to need a lawyer who specializes in this stuff to have any chance of making sure the companies behind the tech are held accountable for what it does on the road.

What is product liability in the context of autonomous vehicles?

It’s the legal responsibility a manufacturer, designer, or seller has when their defective AV causes injury or damage. The defect could be in the hardware, the software, or even just a failure to warn users about what the AV can’t do.

Who can be held responsible if an Uber AV system fails and causes an accident?

It’s complicated, and liability can be spread around. You could be looking at the AV manufacturer, the software developer, the ride-share company (like Uber), or even a human safety driver who failed to step in. It all depends on exactly what went wrong.

How does proving an AV product defect differ from proving driver negligence?

To prove a product defect, you have to show a flaw in the AV’s design, manufacturing, or warnings caused the crash. That’s totally different from a negligence case, where you’re just focused on proving a human driver did something wrong, like speeding or texting.

Are there specific laws in California addressing AV accidents?

California’s DMV has rules for testing and deploying AVs, but the laws for accident liability are still catching up. For now, courts are applying existing product liability and negligence laws to these new cases. For example, a statute like O.C.G.A. Section 51-1-11 gives courts a framework to start from, even if it wasn’t written for robots.

What evidence is typically needed for an AV product liability claim?

You’ll need a ton of technical evidence. This includes the AV’s “black box” data, logs from all the sensors, and an analysis of the software code itself. You’ll also need expert testimony from engineers, accident reconstruction reports, and any witness statements. Getting all this data is the only way to prove a defect.

Editorial Team

The Editorial Team at Car Accident Augusta is a group of writers and editors who cover personal injury and workplace law in Georgia. They research each story and publish under a shared desk byline — experts in their fields, working as one team.