Who’s Liable When a Self-Driving Car Breaks the Law?

Among the issues raised by the spread of self-driving vehicles, and still to be resolved, is what happens in the event of an accident or a fine. Who is liable: the occupant of the driving seat or the manufacturer?

In China, where the largest number of autonomous cars is to be found, no statutes have yet been decided. There are proposals to amend the Road Traffic Safety Law, which has been submitted for its first reading and is currently under public consultation; this amendment includes for the first time, a chapter, Articles 95 to 103, specifically dedicated to the issue of liability.

The proposed amendment distinguishes between driver assistance and autonomy. A vehicle is defined as autonomous if it is capable of consistently performing the entire task of driving under specific operating conditions. If a vehicle provides only driving assistance, liability remains with the human driver. This distinction is important because it prevents marketing from portraying advanced driver-assistance systems as “autonomous driving”. This is similar to the case of the United Kingdom: a clearly defined legal distinction between autonomous driving and driver assistance.

The most significant aspect? The transfer of liability in the event of traffic violations. If autonomous driving is activated and the vehicle commits a violation such as running a red light or exceeding the speed limit, the party held administratively liable will no longer be the driver, but rather the manufacturer or importer. If the company claims that such conduct was not caused by the autonomous system, the burden of proof to demonstrate this will fall on the manufacturer. This reversal of the burden of proof is significant because users no longer have to prove the functioning of closed algorithms over which they have no control. In any case, the liability is administrative, not civil or criminal.

In the event of an accident, the police will be able to access the vehicle’s data, request technical analyses and require the manufacturer’s cooperation; the manufacturer must provide the data relevant to the investigation while complying with privacy, cybersecurity and data protection laws. This means that the black box is no longer merely an accessory but becomes an essential part of the rule of law: it will be necessary to determine what software was installed, whether the function was activated, within which operational design domain it was operating, what command the system issued, and whether or not it requested human intervention. Complete, auditable, and independent records are fundamental to determining liability.

But the Chinese proposal goes beyond simply deciding who pays a fine: authorities must prevent autonomous driving from being activated outside their authorized conditions, suspend unsafe systems, order repairs or recalls, train the owner when necessary, and clearly explain the system’s limitations — and manufacturers must comply with these orders. It will be prohibited to commercially promote capabilities that the vehicle does not actually possess. In addition, specific mandatory insurance for accident liability will be established, the details of which must be set by the government. At the same time, Beijing is already experimenting with products that adapt traditional insurance to the risks posed by software, sensors and updates.

Why now? Because 70.5% of new Chinese passenger cars already incorporate Level 2 combined driver assistance, and more than 57,000 kilometers of roads have been designated for testing and demonstrations, accumulating a total of 220 million kilometers. Actual autonomous driving is still limited to specific sections of Beijing and Chongqing, with speeds capped at 50 and 80 km/h, which gives us an idea of just how far we still are from widespread adoption.

In March, a fleet of 100 robotaxis were grounded in Wuhan due to a system failure, leaving passengers stranded in the middle of expressways, although no one was injured. The lesson is not that autonomous cars are inherently dangerous, but rather that automation introduces problems that can affect entire fleets, meaning that regulations must take these kinds of possibilities into account.

From an international perspective, China lags behind countries such as Germany, the United Kingdom, the European Union and Spain, which has had a national program for testing autonomous vehicles since 2025. What sets China apart is not that it is the first, but that it integrates licensing, traffic violations, data, insurance, advertising, and oversight into the national traffic law of the world’s largest automotive market.

In California, as of July, it is already possible to take action against Waymo and other providers for traffic violations, but not through the conventional fine a driver would receive: the police officer issues a formal notice against the autonomous vehicle, and the case is forwarded to the manufacturer and the DMV, whose truly coercive tool is the operating permit. Before July 2026, there was a legal loophole: in September 2025, the San Bruno police stopped a Waymo vehicle after an illegal U-turn, but were unable to issue a traffic ticket because there was no human driver to penalize; all they could do was report the incident to Waymo. Parking violations are simpler because they are tied to the vehicle and its registered owner, not necessarily to a driver. San Francisco issued 589 parking tickets to Waymo vehicles totaling $65,065 in 2024, and Waymo paid them without dispute.

In Arizona, the law expressly states that the automated driving system is considered the driver for the purposes of assessing compliance with the regulations, while the person who registered the autonomous operation may receive the fine or penalty if the vehicle violates the law. Typically, that person would be Waymo, never its passenger. Texas goes even further: its updated law states that the automated system is the operator, but any fine must be issued to the vehicle owner, or in the case of commercial operations, to the holder of the state permit. Therefore, in Austin, a violation by Waymo is directed at the corporate entity or the holder of the corresponding Waymo permit. Parking tickets are placed on the vehicle in the usual manner; traffic violations require an additional process against the registered corporate owner.

The proposed reform in China does not guarantee that autonomous driving will be safe, nor that it will be deployed on a massive scale any time soon. In general, it is recognized that there is a lower risk in many situations, but also that there are significantly greater challenges when manuevering or where there is complicated directional lighting, at dawn or dusk.

In short, autonomous driving is no longer about technological possibilities; it is beginning to become a legal institution in its own right. As an innovation, we have now reached the point where autonomous driving is becoming a reality; it no longer requires a leap of faith and is readying to accept responsibilities. China’s proposal is clearer and more explicit because it assigns liability for a violation directly to the manufacturer or importer and, crucially, requires the company to prove that the autonomous system did not cause the violation if it wishes to avoid liability.

In the final analysis, everybody recognizes that somebody has to be responsible for an autonomous car, which means legislating for the “algorithmic driver.” Autonomy begins where the fiction of a driver ends, and we must understand that when we shift from the steering wheel to software, liability also changes hands. As they say in China, may you live in interesting times…

—

This post was previously published on MEDIUM.COM.

—

Subscribe to our Email Newsletter:

Why Subscribe? Because this conversation matters.
When you subscribe, you’re directly supporting independent, mission-driven journalism about masculinity, relationships, mental health, fatherhood, and social change. Your inbox becomes part of a movement that’s been challenging stereotypes and expanding what it means to be a good man since 2010. Plus, our newsletter is curated by humans, not algorithms — thoughtful context in an AI-flooded world.

The Good Men Project is a mission-driven men’s media platform. A major platform for stories about men, identity, fatherhood, and emotional intelligence—and a trusted home for the national conversation about masculinity.

The Good Men Project accepts paid guest posts and provides bulk guest post packages for SEO agencies and resellers. Email [email protected] for more information