A. Tesla Autopilot Litigation Overview

(1) The Maldonado Case: Partial Recognition of FSD Liability

On the night of April 10, 2019, on a quiet road in Key Largo, Florida, 22-year-old Naibel Benavides Leon was stargazing with her boyfriend, Dylan Angulo. They were standing beside their SUV, parked on the roadside.

Stars filled the sky. At that moment, a Tesla Model S was barreling toward the T-intersection at over 100 kilometers per hour.

The driver, George McGee, had Autopilot engaged.

He dropped his cellphone. It would only take a second, he thought. He leaned down to pick it up. "I trusted the car to stop on its own," he later told the court. The car did not stop. It failed to detect the stop sign, the flashing red light, or the vehicle ahead. The Tesla slammed into the SUV where Benavides was standing.

She died instantly at the scene. Angulo survived with severe injuries.

A legal term enters the picture here: product liability. Put simply, it works like this. When a toaster doesn't just burn bread but burns down the house, you ask the manufacturer, "Why did you build it that way?" In autonomous vehicle litigation, product liability deals not with metal and plastic but with sensors and algorithms. The defect is not a loose screw but a failed recognition. The warning is not a sticker but a marketing slogan and a CEO's tweet.

On August 1, 2025, a Miami federal jury delivered a historic verdict. They found Tesla 33% at fault. Driver McGee was assigned 67% of the fault. Total damages came to $329 million. Of that, $200 million was punitive damages. By the numbers alone, the driver bears more responsibility. But the legal community interpreted this verdict as "the first crack in Tesla's shield of invincibility." This case was the first involving a third-party death in a Tesla Autopilot incident to reach a jury verdict.

In earlier California lawsuits, Tesla won every time. In those cases, the plaintiffs were Tesla drivers themselves, people injured while operating their own vehicles. Juries concluded that "the driver should have been paying attention." But the Benavides case was different. The victims were a young couple watching the stars. They were not driving a Tesla. They were not riding in one. They were simply standing by the road.

Plaintiff's attorney Brett Schreiber told the court: "Tesla designed Autopilot exclusively for controlled highway use. Yet it deliberately chose not to restrict its use on other roads. Elon Musk promoted Autopilot as safer than a human driver. Words matter. When someone plays games with words, they are playing games with facts and information."

Tesla's attorney Joel Smith countered. Tesla warned drivers to watch the road and keep their hands on the wheel. McGee chose not to do so. He compounded the danger by reaching for his cellphone. The cause of the crash was not Autopilot but the driver's negligence.

The jurors heard both arguments. And they decided both had merit. The driver was careless. True. But the system was also defective. Also true.

33 to 67. This is the new fault-allocation formula of the autonomous driving era. The critical issue was the $200 million in punitive damages. Punitive damages are amounts imposed to "punish" a company beyond actual harm. For these to be awarded, the plaintiff must prove not mere negligence but conduct that was knowing and willful. The jurors concluded that Tesla knew the system's limitations and continued selling it anyway. Tesla immediately announced an appeal, issuing a statement that "today's verdict is wrong and undermines automotive safety." But legal analyst Dan Ives saw it differently. "That's a big number. It will send shockwaves through the entire industry. This is not a good day for Tesla."

There is a telling detail. Months before the trial, the plaintiff's side offered Tesla a settlement of $60 million. Tesla rejected it. To be precise, Tesla failed to respond within 30 days, resulting in an automatic rejection. The consequence: Tesla now faced a liability more than four times that amount. This could be a miscalculation, or it could be the price of believing "we are never at fault."

During the same period, Tesla was quietly settling other fatal crash lawsuits.

The 2018 case of Apple engineer Walter Huang, who died when his Model X struck a highway barrier. The 2019 case of 15-year-old Jovani Maldonado Garcia, killed by a Model 3. Both cases were resolved through confidential settlements just before trial. The settlement amounts were never disclosed. But it was clear Tesla wanted to avoid the courtroom. The Miami verdict showed why. Standing before a jury, they could lose.

(2) California DMV Administrative Action

Administrative courts have no juries. Instead, regulators sit in judgment. They ask a single question: "Could this language mislead consumers?" If the answer is "yes," the company has a problem.

In 2022, the California Department of Motor Vehicles formally charged Tesla with false advertising. The dispute centered on two words: "Autopilot" and "Full Self-Driving." The DMV's argument was straightforward.

These words imply the car drives itself. But Tesla's technology does not do that. It remains a Level 2 driver-assistance system, requiring the driver to watch the road and intervene at any moment. The name does not match reality. This is deception.

Tesla pushed back. "Autopilot" is a term borrowed from aviation. Aircraft autopilot also requires continuous pilot monitoring. Tesla stated on its website that "the driver must remain in control." In small print? Yes, but it was stated.

After three years of proceedings, on December 16, 2025, California Administrative Law Judge Juliet Cox issued her ruling. Tesla's marketing is deceptive. Suspend its manufacturing license and dealer license for 30 days each.

This was a nuclear-grade sanction. It meant Tesla could neither sell nor manufacture cars in California. Tesla's Fremont factory is located in California. A significant share of Tesla's U.S. sales occur in that state.

DMV Director Steve Gordon said at a press conference: "The DMV has decided to give Tesla one more chance to make things right." The DMV softened the penalty. The manufacturing license suspension was stayed indefinitely. The dealer license suspension was stayed for 60 days. Within that window, Tesla could either discontinue use of the problematic terms or revise its marketing to eliminate consumer confusion. Failure to comply would trigger the 30-day sales suspension. Tesla issued a statement: "This is a 'consumer protection' order regarding the use of the term 'Autopilot.' Not a single customer has filed a complaint. California sales will continue without interruption." An interesting line of argument: if no customer complained, it isn't deception. But the DMV did not bring its case based on customer complaints. It determined that the terminology itself was deceptive.

What does this administrative action mean? It represents pressure on a different plane than a civil lawsuit. Civil suits address damages from individual crashes: who was at fault, how much to pay. Administrative action targets the company's right to do business. Even without a crash, even without a victim, regulators can say, "Your words are the problem."

The precedent matters more. The fact that the DMV ruled "Autopilot" a deceptive term becomes a record that plaintiffs can invoke in civil suits. "As the regulator itself acknowledged, Tesla's marketing misled consumers." That single sentence can sway a jury.

Tesla has already begun changing its terminology. "Full Self-Driving Capability" was renamed "Full Self-Driving (Supervised)." The word "supervised" was added. The technology didn't change; the label did. But a change in labeling can translate into a change in legal liability. Words matter.

(3) Analysis of Jury Verdict Trends

Jury verdicts function as traffic lights for the industry. When the light turns green, similar claims follow in a line.

Through 2023, Tesla's light was always green. In the Micah Lee wrongful death trial held in Riverside County, California, the jury found Tesla not liable. The conclusion: "The vehicle had no defect at the time of the crash, and the driver's inattention was the cause." Other trials during the same period reached identical results. Tesla's defense attorneys brimmed with confidence. "Autopilot is nothing more than a driver-assistance device. The responsibility for driving always rests with the driver."

But in 2025, the signal began to change.

The Miami Benavides verdict was the first yellow light. Tesla liability at 33%, damages of $243 million. Once these numbers became public, plaintiff's attorneys nationwide started mobilizing.

According to Reuters, lawyers handling other Autopilot-related lawsuits requested permission to share Tesla's internal engineering and design documents across cases. Their reasoning: many of these suits address similar technical issues. Sharing evidence would increase efficiency and enable stronger claims.

Tesla objected. The judge denied the request for unrestricted sharing of trade-secret materials. But legal experts predict a significant increase in Autopilot lawsuits. When a large jury verdict is recorded, other plaintiffs become more likely to choose trial over settlement.

Several factors explain the shift in juror attitudes.

First, time. In 2019, crash victims could say, "I barely knew what Autopilot was." Jurors in 2025 are different. They've seen Tesla crashes in the news multiple times. They know "Full Self-Driving" is not yet fully autonomous. Once the illusion fades around the technology, expectations placed on the company rise.

Second, the accumulation of evidence. In early lawsuits, plaintiffs struggled to obtain Tesla's internal documents. But as cases piled up, so did the paperwork. Internal emails from engineers. Minutes from executive meetings. Evidence that the company knew the system had blind spots in certain conditions. When these documents are laid before a jury, the defense of "we didn't know" stops working.

Third, the type of victim. When a Tesla driver is injured, jurors find it easy to say, "You bear some responsibility too." You bought the car. You turned on the feature. But when an innocent third party is harmed, the calculus changes. A young couple watching stars. A pedestrian crossing at a crosswalk. They never chose Tesla. They were simply in the wrong place at the wrong time. Juror sympathy tilts toward them.

According to NHTSA data, of 392 autonomous-driving-related crashes reported between July 2021 and May 2022, 273 (roughly 70%) involved Tesla vehicles. Whether this statistic reflects Tesla's high market share or a systemic flaw in its technology is debatable. But when this number is presented to a jury, it leaves an impression regardless of interpretation.

In the U.S. District Court for the Northern District of California, another type of lawsuit is underway: a class action over FSD false-autonomy claims. It covers drivers who purchased or leased a Tesla in California between October 2016 and July 2024. In August 2025, the court certified the class. This lawsuit can proceed without a death or serious injury. Its core claim is that "the expectation sold for money was false." If Tesla loses this class action, it may owe compensation to millions of owners.

Financial analysts have begun reassessing Tesla's legal risk. The insurance industry has done the same. Insurance premiums for Tesla vehicles are rising. Product liability insurance costs are climbing too. A courtroom loss is not just a payout. It rewrites the price tag for the entire system.

B. The Cruise (GM) Incident

(1) The San Francisco Pedestrian Dragging Accident

On the night of October 2, 2023, at the intersection of 5th Street and Market Street in San Francisco, a woman was crossing at the crosswalk. The signal was green for pedestrians.

That's when a human-driven car struck her. It was a hit-and-run. She was thrown into the adjacent lane.

Unfortunately, a Cruise driverless robotaxi happened to be there.

Cruise is General Motors' autonomous driving subsidiary. In August 2023, the California Public Utilities Commission (CPUC) granted Cruise a permit to operate paid robotaxi services 24 hours a day across San Francisco. At the time, roughly 950 Cruise vehicles were operating nationwide. It had been less than two months since the permit was issued.

The robotaxi braked hard. But it could not avoid the victim. The vehicle hit her and stopped.

Up to this point, there was room to call it unavoidable. The first collision was caused by another vehicle, and Cruise had no time to react.

The problem was what happened next.

Cruise's algorithm assessed the situation. A collision has occurred. The vehicle is stopped in the road. It must move to a safe location. The car began moving again. It headed toward the curb. The problem was that the victim was still trapped beneath the vehicle. The car traveled roughly 6 meters (20 feet), dragging her along the road. The speed was about 11 kilometers per hour. Slow. But fast enough to drag a human body across asphalt.

Would a human driver have done this? After a crash, a person gets out of the car. They check the scene. They look to see if anyone is hurt. But a robotaxi has no one to get out. It only has an algorithm. And that algorithm followed a rule: "after a collision, move to a safe location." The rule was correct. The situational awareness was wrong.

The victim suffered horrific injuries. A fractured pelvis, broken legs, and countless abrasions from being dragged across pavement. She required extended hospital treatment.

This accident was a technical failure. It was also a moral one.

Immediately after the incident, Cruise executives showed regulators and the press video footage of the crash. But the footage they showed only included the moment the vehicle came to a stop. The part where it started moving again and dragged the victim was missing. According to the California DMV, the full video was not provided until nine days after the accident.

A Cruise spokesperson pushed back. "We showed the full video the day after the accident." Who is right remains disputed. But what is clear is that regulators were unhappy with Cruise's initial account. DMV officials later said they learned the victim had been dragged only through conversations with other government agencies.

A technical error might be forgivable. Software is not perfect. There are bugs, there are edge cases. But behavior that looks like an attempt to hide information is different. That breaks trust. Once trust is broken, it does not recover easily.

(2) Permit Suspension and Settlement

The response was immediate and severe.

Three weeks after the accident, the California DMV suspended Cruise's autonomous vehicle deployment and testing permits entirely. The reason was explicit: "The vehicles pose an unreasonable risk to public safety." The DMV's suspension notice also stated that Cruise had failed to fully disclose the severity of the accident to regulators.

Cruise halted operations not only in San Francisco but across the entire United States. All 950 robotaxis returned to the garage overnight. The company recalled its entire fleet and issued a software update.

But that was only the beginning.

In November 2023, Cruise co-founder and CEO Kyle Vogt resigned. Nine senior executives were fired or stepped down. Roughly 24 percent of the workforce, about 900 people, lost their jobs. GM cut its investment in the Cruise division in half, from $2 billion to $1 billion per year.

The U.S. Department of Justice (DOJ) and the Securities and Exchange Commission (SEC) also launched investigations. NHTSA had already been investigating multiple incidents in which Cruise vehicles collided with or came dangerously close to pedestrians.

In May 2024, Cruise reached a settlement with the victim. The amount was reported to be between $8 million and $12 million. The exact figure is confidential. The victim has been discharged from the hospital and is said to be recovering.

In October 2024, Cruise agreed to a consent order with NHTSA. The terms: a $1.5 million civil penalty, regular reporting on autonomous operations, and quarterly meetings with NHTSA officials. The agreement resolved allegations that Cruise had violated accident reporting requirements by failing to properly disclose crash information.

A Bloomberg article called the incident "a clear lesson for the autonomous driving industry." "An algorithm's mistake may be forgivable. But a cover-up by human executives is not."

By late 2024, GM made a bigger decision. It would shut down or drastically scale back Cruise's robotaxi business. GM's total investment in Cruise was estimated at roughly $10 billion. Much of that money was gone. The robotaxi dream had yet to generate revenue, and with regulators and the public having lost confidence, there was no justification for continued investment.

In February 2025, GM announced it would acquire full ownership of Cruise. But this was not expansion; it was consolidation. Cruise would no longer develop Level 4 fully autonomous driving. Instead, it would focus on GM's Level 2 driver-assistance system, Super Cruise. A project built on $10 billion and years of effort was effectively over.

The Cruise case left behind a question. When a driverless vehicle causes an accident, who is responsible? If the driver's seat is empty, the traditional framework of "driver negligence" does not apply. So where does liability go? To the manufacturer. The software developer. The fleet operator. The remote monitoring center. It disperses across every entity that built the safety management system.

A robotaxi is both a "product" and a "service." If you sell a toaster, that toaster does not change. But a robotaxi is updated daily. The software changes, the maps are refreshed, the algorithms learn. In this context, what constitutes a "defect"? A defect at the time of shipment, or a defect introduced after an update? The law still has no clear answer to this question.

Cruise's downfall became a lesson for competitors. How you respond after an accident may matter more than the accident itself. Disclosing information transparently, cooperating with regulators, acknowledging responsibility. These are the conditions for survival.

C. Waymo's Current Status

(1) Accident Statistics and Safety Data

While Cruise collapsed and Tesla struggled in court, Waymo was quietly driving. Starting as Google's self-driving project and becoming an Alphabet subsidiary, the company's strategy was: "We will persuade with data."

As of September 2025, Waymo had logged over 127 million miles (approximately 200 million kilometers) of fully autonomous driving. "Fully autonomous" means no one in the driver's seat. Miles driven with nobody behind the wheel. This is equivalent to one person spending 150 lifetimes doing nothing but driving.

Waymo uses this data to argue it is safer than humans.

A peer-reviewed study published in May 2025 found that Waymo vehicles had an 85% lower rate of serious injury crashes compared to human drivers. Intersection-related injury crashes were 96% lower. A joint study with Swiss Re, the insurer, found that over 25 million miles of driving, Waymo vehicles reduced insurance claims by more than 90% compared to human drivers.

How could these numbers be used in court? If a Waymo vehicle caused an accident, Waymo's attorneys could argue: "Our system is safer than humans. The data proves it. This accident was an unavoidable exception." The plaintiff's side could counter: "You advertised that you're safer than humans, so you shouldn't make mistakes that humans would make. The standard should be higher."

Waymo is not perfect, of course. According to data reported to NHTSA, between July 2021 and November 2025, Waymo vehicles were involved in 1,429 crashes. 117 people were injured and 2 died. These numbers look large in isolation. But context is needed. In most crashes, the Waymo vehicle was not at fault. Other vehicles rear-ended the Waymo or ran red lights. One analyst reviewed 38 serious incidents (involving injury or airbag deployment) between July 2024 and February 2025.

His conclusion: only 1 of those was clearly Waymo's fault. Three were difficult to judge, and the remaining 34 were mostly or entirely the fault of the other party.

Still, accidents are accidents. In May 2024, a Waymo vehicle in Phoenix struck a wooden utility pole while maneuvering at low speed. It was a structure not in the map data. No one was injured, but Waymo issued a voluntary recall (software update) for 672 vehicles. In May 2025, 1,212 vehicles were recalled for a software issue that could cause minor collisions with objects like roadside barriers, gates, and chains.

In October 2024, a Waymo vehicle killed a dog in San Francisco. A neighborhood cat named "KitKat" was also struck and killed by a Waymo. These incidents drew anger from animal welfare groups and local residents. Technically, detecting small objects, especially animals that move unpredictably, is a difficult problem. But emotionally, the reaction is: "A robot killed our family member."

Waymo's response is data and transparency. They report every accident under NHTSA's reporting system. Even minor ones. Including contact incidents so slight that a human driver would never have reported them. This conservative reporting approach inflates the crash count but builds trust. "We are not hiding anything."

Waymo also operates a Safety Data Hub. On a publicly accessible website, anyone can review Waymo's crash data and mileage figures. Researchers can verify and reproduce the data. David Zuby, chief research officer at the Insurance Institute for Highway Safety (IIHS), said: "By making detailed crash and mileage information publicly accessible, Waymo's transparency will support independent research and promote public trust. We hope other companies developing and deploying autonomous systems will follow suit."

(2) Expansion Plans and Regulatory Response

Waymo's expansion is not about moving down roads; it is about moving through law.

By the end of 2024, Waymo was operating commercial services in San Francisco, Los Angeles, Phoenix, and Austin. It provides over one million paid rides per month. In 2025, it expanded to Atlanta. Through a partnership with Uber, riders in Atlanta can now hail a Waymo robotaxi.

2026 will be more ambitious. Waymo plans to expand to 15 cities: Dallas, Houston, San Antonio, Miami, Orlando, Detroit, Denver, Las Vegas, Nashville, San Diego, Washington D.C., Minneapolis, Tampa, New Orleans. It even plans to launch its first international service in London.

Minneapolis matters. Winters here are brutal. Snow, ice, sub-zero temperatures. For years, these conditions were the Achilles' heel of autonomous driving technology. Camera sensors get blocked by snow, and road markings disappear beneath it. Waymo's decision to begin testing in Minneapolis signals confidence that it has solved the winter problem.

But entering a new city is not just a matter of technology. Every city and every state has different regulations. Permit conditions, data reporting formats, emergency vehicle response protocols, and insurance requirements all vary. Waymo works closely with local regulators before entering each market. It establishes response protocols with fire departments, police departments, and emergency services. This takes time. But remembering how Cruise was criticized for blocking emergency vehicles, it is a necessary investment.

In November 2025, the California DMV approved a major permit expansion for Waymo. The coverage area extends to nearly all of Southern California down to the Mexican border, the entire Bay Area, and Sacramento. This is far larger than Waymo's current operating territory. In August 2025, New York City approved Waymo for the city's first autonomous vehicle testing permit. It can operate up to eight vehicles in Manhattan. Waymo's regulatory strategy is 'cooperation, not confrontation.' When NHTSA opened a preliminary investigation into some of Waymo's crashes, Waymo did not take a defensive posture. It proactively shared data and cooperated with the investigation. In 2025, NHTSA closed its investigation into Waymo, concluding that software updates had resolved the issues. There were no fines. This case demonstrates that preemptive recalls and cooperation can reduce legal liability.

While Tesla argues that 'regulation stifles innovation,' Waymo is quietly submitting passing report cards against regulatory standards. Which approach will go further is something only time will tell.

D. Industry-Wide Trends (From 288 to 544 Cases)

(1) Surge in Autonomous Driving Litigation (288 to 544 Cases)

Numbers are written on the whiteboard in a law firm conference room. 288. 544. Nearly double. These figures represent the change in autonomous driving-related lawsuits between 2023 and 2024.

How should we read these numbers? The interpretation that 'the technology got worse' is superficial. A more accurate reading is that 'the technology spread.' When autonomous vehicles leave the laboratory and enter public roads, accidents follow. When accidents happen, lawsuits come next. This is a natural progression.

NHTSA has imposed reporting requirements for crashes involving Automated Driving Systems (ADS) and Level 2 Advanced Driver Assistance Systems (ADAS) since 2021.

This is called the 'Standing General Order.' Manufacturers must report certain types of crashes to NHTSA. Thanks to this rule, accidents that previously went unrecorded are now documented. More records mean more potential lawsuits.

The types of litigation are also diversifying.

First, personal injury and wrongful death lawsuits. The Tesla Benavides case is a prime example. Victims injured or killed in crashes (or their families) seek damages from the manufacturer. The central issues are product defects, whether in design, manufacturing, or warnings, and whether those defects caused the accident.

Second, false advertising and consumer protection lawsuits. The California FSD class action falls into this category. No accident is required. The claim is that 'the feature I paid for does not match the advertising.' Consumer protection statutes, prohibitions on deceptive marketing, and contractual warranty obligations provide the legal basis.

Third, shareholder lawsuits. When executives exaggerate the progress of autonomous driving technology to inflate stock prices, and then the stock crashes after an accident or recall, shareholders allege securities fraud. The core allegation is that executives 'knowingly lied.' Fourth, intellectual property disputes. These are patent infringement actions between companies over lidar sensor technology, AI training data, and autonomous driving algorithms. The trade secret dispute between Waymo and Uber is a notable past example.

Fifth, lawsuits and administrative actions related to regulatory violations. The California DMV's administrative action against Tesla and NHTSA's consent order against Cruise fall into this category. These are not civil lawsuits, but they directly affect operating rights and reputation.

Attorneys welcome the growth of this market. One lawyer put it this way: 'Autonomous driving accidents are complex. You need data scientists, software engineers, and human factors experts. The costs are high. But so are the verdicts.' Law firms are assembling dedicated autonomous driving teams. They are hiring data scientists who analyze Event Data Recorders (EDR) and vehicle log data.

Defense attorneys are busy too. Manufacturers like Tesla, GM, Ford, Hyundai, BMW, and Mercedes-Benz are increasing their legal budgets in anticipation of autonomous driving litigation. Insurers are redesigning their product liability insurance offerings.

The surge in litigation also affects the regulatory environment. As case law accumulates, trends emerge in how courts rule. Legislators observe these trends and draft new laws. Regulators adjust their enforcement standards. The insurance industry recalculates premiums. Litigation is not just a mechanism for resolving disputes. It is the process of rewriting an industry's rules.

(2) The Product Liability vs. Driver Liability Debate

For a hundred years, car accidents were 'the driver's fault.' Speeding, drunk driving, inattention. The law held the person behind the wheel responsible.

But what happens when no one is behind the wheel?

This is the legal question that autonomous driving poses. And right now, courts and legislators are searching for the answer.

Driver liability is the old door. When a traffic accident occurs, police arrive. They investigate who ran the red light, who was speeding, who was distracted. The driver at fault bears responsibility. The insurance company pays. This is the familiar system.

Product liability is the new door next to it. Walk through this door and the questions change. Instead of 'What did the driver do wrong?' the question becomes 'How was the car built?' Was there a design defect? Were adequate warnings provided? Was the advertising deceptive? Answering these questions requires examining the manufacturer's engineering documents, internal emails, and test records.

Tesla's Autopilot and FSD are currently classified as Level 2 systems. Level 2 means 'the driver must always watch the road and be ready to intervene at any time.' Legally, the ultimate responsibility for driving still rests with the driver. So Tesla's attorneys say in court: 'Autopilot is merely an assistive device. The driver should have been paying attention.'

But jurors are now asking a different question. 'Why wasn't the driver paying attention?' Didn't Tesla name it 'Full Self-Driving,' didn't the CEO tweet that it was 'safer than humans,' didn't the advertisements show drivers taking their hands off the wheel? If so, isn't the driver's complacency partly Tesla's fault? The August 2025 jury verdict in the Benavides case accepted this logic. Driver responsibility: 67%. Manufacturer responsibility: 33%. Both were at fault. This is the new formula.

At Level 3 and above, the situation becomes clearer.

Level 3 means 'the system handles driving under specific conditions, and the driver intervenes only when the system requests it.' Mercedes-Benz's Drive Pilot falls into this category. Mercedes has declared that it will accept manufacturer liability for accidents that occur while Drive Pilot is active. Volvo has taken a similar position.

At Level 4 (Waymo, Cruise), there is no person in the driver's seat, so the concept of 'driver liability' itself becomes difficult to apply.

California has already enacted relevant legislation. Under AB 1777 (enacted September 2024), liability for traffic violations that occur while autonomous mode is active can be assigned to the manufacturer rather than the driver. Texas and Arizona are moving in a similar direction.

The insurance industry is also being reshaped. Personal auto insurance covers 'driver error.' As autonomous vehicles proliferate, driver errors will decrease, and personal insurance premiums may fall. On the other hand, the product liability insurance that manufacturers must carry will become more expensive. The cost of accidents is shifting from individuals' wallets to corporate balance sheets.

A RAND Corporation study predicted the following:

'As autonomous vehicles become widespread, the focus of liability will shift from individual drivers to manufacturers.' The Insurance Information Institute shares this view: 'In autonomous vehicle crashes, liability will shift from the at-fault driver to manufacturers and component suppliers, triggering product liability claims rather than standard automotive negligence claims.'

This is a fundamental shift in the legal framework. And this shift has already begun. Tesla's $329 million verdict is the opening signal. More signals will follow, from courtrooms, from legislatures, from the desks of insurance actuaries. The future of autonomous driving will not be determined by technology alone. Where the law draws the line of liability will shape it equally.