Get way more money on your Waymo cases

Critical aspects of litigation in autonomous-driving cases

2026 July

You’re sitting in a car at a red light, and a car stops next to you. You look over and then have to do a double take – there’s no driver. There are plenty of cameras and gadgets, but an empty driver’s seat. Then you realize, autonomous driving is here. Now. 

Litigating against Waymo or any company that offers autonomous driving is not simply “another vehicle case.” Forum selection, available causes of action, evidence preservation, and discovery all differ from traditional automobile litigation. This article addresses critical aspects of litigation in autonomous driving cases including, the forum to bring the claim, types of litigation – causes of action, and the discovery process.

Where can we bring a claim? 

Court vs. arbitration

As of March 22, 2026, when I first signed up for my Waymo user account, the Waymo Terms of Use includes a limitation on the forum in which a party can bring a claim. Where an injured party can bring their claim will be dependent upon who the party is. 

If the injured party is an account holder and user of Waymo – welcome to arbitration! This is an enforceable provision of the clickwrap agreement between a user and Waymo. Here, an account holder explicitly consents to the terms and conditions, including the provision to arbitrate their claims for personal injury against Waymo. Notably, the arbitration provision is presented clearly, assent requires an affirmative act (clicking the “I agree”) and is conspicuously disclosed. The arbitration will be held in the county of the user’s billing address. Note, in other automobile cases with arbitration provisions (think rideshare), bringing a claim against the individual driver in state court while arbitrating against the entity was an option. Here, there is no driver to bring a claim against. 

For an individual injured as a guest in a ride (a passenger in a ride requested by an account holder, who is not an account holder themselves), the venue may change. The agreement to arbitrate is a binding provision of the agreement between an account holder and Waymo. If an injured passenger does not have an account, there is no written contract between them and Waymo. But Waymo’s terms of service provide that they are enforceable against all users of their services, “including as a guest in a ride requested by another user.” 

Whether this provision is enforceable then becomes fact specific. There is no affirmative act accepting this term. However, this will hinge on a few factors, including whether the guest had the opportunity to review the term, knew of the term, separately acknowledged the term, repeatedly rode as a guest, or was a minor. 

If the injured party was not inside the Waymo at the time of the collision (think pedestrian, cyclist, occupant of another vehicle, property owner, etc.) and does not have a Waymo account, they are not bound by Waymo’s Terms of Use and have the court system available to bring their claim. 

Which court?

If the injured party is not a user and does not have an account, the next step is determining which court they can bring their claim in.

Waymo’s Terms of Use designate state and federal courts in Santa Clara County, California, for court proceedings not subject to arbitration. However, these venue provisions do not bind a non-user without an account. Thus, an injured party may pursue venue based on where the incident occurred, where Waymo’s operations take place (Santa Clara), or any statutory-specific venue rules. If an injured party can satisfy diversity jurisdiction requirements, set forth a class claim, or allege a substantial federal question (think about claims that can be tied to autonomous-vehicle regulation), then the federal courts will be available. Otherwise, your injured client will be able to bring their claim in state court. 

Types of litigation: What are the causes of action? 

Autonomous-vehicle cases are not likely to fit into a single theory of liability. California law recognizes that a single incident can give rise to multiple causes of action. Successful pleadings should assert several theories of liability. 

Negligence

Unlike human-driver cases, negligence is evaluated against system behavior, not individual judgment. For a negligence cause of action, the pleading should focus on system operations. Specifically, acts that are part of the implementation of the autonomous driving system. A few examples include: 

• Operation of vehicles in unsafe conditions
• Failure to adequately supervise or monitor remote-assistance systems
• Inadequate or failure to conduct a safety validation inspection prior to public deployment of an autonomous driving vehicle
• Ignoring system limitations or prior incidents
• Failure to implement reasonable safeguards

The idea in a negligence cause of action is the need to focus specifically on the conduct necessary for the operation of an autonomous driving system through its agents, not whether the autonomous driving system itself was defective. Even if the autonomous driving system was functioning as designed, agents can be negligent in how the system is used, deployed, or managed. By doing this, the narrative is conduct-based and similar to a traditional negligence claim rather than focusing on a broken system. Practice tip: Avoid pleading language such that “there was nothing wrong with the system” or “the system worked fine.” (It’s unnecessary.) 

Product defect

Product-defect claims are available and should be included in your pleading. However, these are harder to prove, narrower, and strategically different theories than the negligence claim. The key is understanding what the actual “product” is, who the manufacturer is, and which defect theory actually fits autonomous-driving software. A product-defect cause of action may consist of claims that the autonomous-driving system is: 

• Defectively designed (software and sensor shortcomings)
• Defectively manufactured (hardware or sensor failures)
• Defectively warned (known limitations not adequately disclosed)

The product here can be pled as the autonomous-driving system as a whole, or the software and sensor suite controlling the vehicle operation, or the integration into a specific vehicle platform. Typically, the pleading should avoid defining the product as “the ride” or “the service” as that has a risk of morphing into a negligence-based theory. 

In California, a product is defectively designed if it fails either the consumer-expectation test or the risk-utility test (with burden shifting). Under the consumer-expectation test, a product is considered defective if it is unreasonably dangerous beyond what an ordinary consumer would anticipate, given their knowledge of the product and its intended use. 

This is why understanding what “the product” is will shape the focus of the argument. For example, if the product is the autonomous-driving system, one may argue that the system cannot reliably detect or respond to certain foreseeable roadway conditions. However, if the product is the software and sensor suite that controls the vehicle operations, one can argue the software should implement more safety responses. 

For a claim of manufacturing defect, the focus is on the product departing from its intended design. This could mean a sensor was miscalibrated, software was improperly corrupted or updated, the vehicle deviated from fleet standards. Here, the focus is not proving the design itself was bad, rather that this vehicle was not what was intended. The key here is the defect is tied directly to incident-specific facts of your client’s claim. 

A failure-to-warn claim means the product can be defective even if it works exactly as designed, if risks were not adequately disclosed. Typically, this means a failure to warn the public or regulators about known system limitations. In the autonomous driving context, this could mean overstating the capabilities of the product, failure to instruct on foreseeable misuses of the product, or failure to adequately disclose when the autonomous vehicle should and should not be on the road. The United States Department of Transportation regulates autonomous driving; if the crux of your claim is a failure to warn regulators, this is where you may find your federal claim. 

The central battlefield: Discovery

The case begins immediately: Preservation of evidence

Waymo vehicles utilize LiDAR (light detection and ranging), imaging radar, and sensors as part of a system that fuses together to create a high-resolution 3D map of their surroundings. For good measure, these vehicles also utilize cameras (a lot of cameras). This is monitored remotely and logged. A preservation- of-evidence letter should be sent immediately and include evidence from the vehicle (on-board sensor data, camera videos) and the remote assistance logs. 

This system requires ongoing and routine inspections, maintenance, and investigations – your preservation-of-evidence letter should also include investigation reports. Be sure to specify both safety and investigation reports. Note, the safety reports are routinely completed as part of the ordinary course of business as part of regulatory compliance. Use this to oppose any claims that these reports are privileged and protected from disclosure. 

Scope 

The scope of discovery should be broad, but specific enough to include requests of information and evidence from the actual vehicle, Waymo system, and remote assistance. This could include saved data obtained before a collision – specify the sensory data compiled during the specific ride. System design documents – product-testing specifications. Or training documents on assistance monitoring – even more specific, the assistance logs for the actual trip. Even broader, regulatory submissions.

Anticipate objections based on privilege, vague and ambiguous, and claims of proprietary information. Privilege is not valid, but the objection is coming; ask for a privilege log and explanation to evaluate the claim. Vague and ambiguous – identify a specific category of documents. The real one with teeth is the claim for proprietary information. Waymo owns over 100 patents, and relevant information for purposes of litigation is not included in their patent filings. Thus, a protective order will be necessary. Note, a claim of trade secret cannot be a discovery shield where information is central to liability. Push back on overbroad language of the protective order. 

Person-Most-Qualified depositions

These depositions will be crucial to liability, notably to plug any gaps that need to be filled from the document production. For these, topics should again focus on conduct outside of the autonomous-driving system (safety inspections, system monitoring, operations of the system, etc.), the autonomous-driving system, data from the vehicle, and regulation compliance standards. Consider some of the following topics:

• Autonomous system – function, form, design, structure, organization, software updates, etc.
• Incident-response procedures
• Remote-assistance protocols and training
• Safety inspections and investigations
• Vehicle maintenance
• Data retention policies
• Regulatory compliance practices

Litigating against an autonomous-driving company fundamentally alters nearly every assumption that traditionally underpins motor-vehicle-injury practice. In a conventional auto case, liability analysis begins with human conduct: perception, reaction time, distraction, intoxication, or simple inattentiveness. Evidence is comparatively finite, discovery is familiar, and the central dispute often comes down to credibility. 

Autonomous-vehicle cases invert each of these principles. Human judgment is replaced by automated systems; fault lies not with a single decision-maker but with a complex network of hardware, software, remote operators, policies, and corporate choices; and the most critical evidence exists almost entirely in digital form, controlled by a sophisticated corporate defendant with strong incentives to limit disclosure.

Not just another car case 

If you are representing an injured party, recognizing that this is not “just another car case” is the most important first step – and one that must be taken immediately. Autonomous-vehicle litigation exists at the intersection of tort law, product liability, regulatory compliance, and emerging technology. Successfully navigating this landscape requires reframing how we think about venue, causes of action, discovery, and ultimately – the case valuation. 

Autonomous-vehicle cases represent the natural evolution of trial advocacy. They require attorneys to translate complex technologies into accessible narratives, to connect abstract system failures to real human harm, and to confront powerful corporate defendants on equal footing. While the learning curve is steep, the opportunity is substantial. 

As autonomous vehicles become more prevalent, these cases will only increase in frequency and complexity. Understanding autonomous systems, regulatory frameworks, and data-driven discovery will be critical. By mastering forum strategy, pleading layered theories of liability, and commanding the discovery process, we can not only secure meaningful recovery on behalf of injury victims – but also play a critical role in shaping the future of roadway safety. 

Autonomous driving is here. Our advocacy must rise to meet it. 

Ryan Cowan is a senior attorney at Wilshire Law Firm, working in the firm’s Personal Injury Litigation department. He represents injured individuals that have suffered catastrophic injuries as a result of motor vehicle collisions, premises liability, dog bites.