Civil Law and Comparative Autonomous Vehicle Regulation Models

1. Introduction

Autonomous vehicle regulation concerns the legal rules governing motor vehicles that use artificial intelligence (AI), sensors, cameras, radar, lidar, software and automated control systems to perform some or all driving tasks with limited or no continuous human intervention.

These vehicles include automated passenger cars, robotaxis, autonomous buses, self-driving trucks, delivery vehicles and connected transport fleets.

From a civil-law perspective, autonomous vehicle regulation raises important questions about negligence, product liability, contractual responsibility, insurance, consumer protection, data privacy and compensation for personal injury or property damage.

The central legal question is: Who should bear civil liability when an autonomous vehicle causes an accident—the vehicle owner, the human driver, the manufacturer, the software developer, the fleet operator or another responsible party?

The answer depends on the vehicle's automation capabilities, who controlled its operation, whether a defect or negligent act caused the accident, and the legislation governing the claim.

Autonomous vehicle regulation is particularly important because traditional road-traffic law assumes that a human driver observes the road, makes decisions and controls the vehicle. Higher levels of automation shift some of these functions to software, creating new challenges for established legal doctrines.

2. Understanding autonomous vehicle technology

Level 1 — Driver assistance

A system assists with a specific function, such as steering or speed control. The human driver remains responsible for the overall driving task.

Level 2 — Partial driving automation

The vehicle can control steering and acceleration or braking together under specified conditions, but the human driver must supervise the driving task continuously.

Level 3 — Conditional driving automation

The system performs the driving task within its operational limits, but a human fallback-ready user may be required to respond to a takeover request.

Level 4 — High driving automation

The vehicle can perform the driving task within its defined operational design domain without requiring a human to take over during that operation.

Level 5 — Full driving automation

The system is designed to perform the driving task under all conditions covered by its design, without needing a human driver.

These levels follow the general classification framework in SAE J3016. They describe driving automation capabilities, not a vehicle's legal status or the identity of the party liable for an accident. A vehicle marketed as “self-driving” may still require active human supervision.

3. Principal regulatory models

Autonomous vehicle regulation can be divided into several models. Some focus primarily on vehicle approval and technical safety; others establish a specific legal entity responsible for self-driving operations or retain traditional liability rules while adapting them to new technology.

A. Australia: National safety reform and existing civil liability

Australia's framework combines federal vehicle standards with state and territory road rules, registration requirements, compulsory insurance and civil liability legislation.

The National Transport Commission is developing nationally consistent arrangements for higher levels of automation. As of October 2026, Australia's stated roadmap provides for conditional deployment in selected locations from 2027, subject to legislative and operational readiness.

National Transport Commission

+1

Key features include:

Vehicle safety and compliance requirements before supply and operation.

Development of an Automated Vehicle Safety Law.

Safety duties for entities responsible for automated vehicle operation.

Updates to road rules, enforcement, insurance and registration.

State and territory arrangements for testing and trials.

Australia's approach is particularly relevant to negligence claims involving manufacturers, vehicle owners, fleet operators and other parties responsible for a vehicle's safe operation.

B. United Kingdom: Dedicated automated-vehicle legislation

The United Kingdom enacted the Automated Vehicles Act 2024, establishing a framework for authorising self-driving vehicles and assigning responsibilities to regulated entities.

The framework distinguishes between vehicles that merely provide driver assistance and those capable of driving autonomously without continuous human monitoring. It provides for an authorised self-driving entity (ASDE), responsible for the vehicle's behaviour while it is self-driving, and an in-use regulatory system.

Legislation.gov.uk

+1

Its central features include:

A legal test for self-driving capability.

Safety assessment and authorisation.

Defined responsibilities for authorised entities.

Monitoring of vehicles after authorisation.

Controls on misleading claims about self-driving capability.

The Act's framework is subject to commencement provisions, so enactment must not be confused with every provision already being operational. This model is significant because it expressly separates the responsibilities associated with autonomous driving from the ordinary responsibilities of a human driver.

C. European Union: Type approval, product safety and civil liability

The EU uses a combination of vehicle type-approval requirements, general vehicle-safety legislation, automated-driving technical standards and national civil liability laws.

Regulation (EU) 2019/2144 establishes general safety requirements for vehicles. Implementing Regulation (EU) 2022/1426, as amended, addresses technical requirements and procedures for approving the automated driving systems of fully automated vehicles. The EU adopted a further amendment in March 2026.

EUR-Lex

+1

This model separates two questions:

Regulatory compliance: Was the vehicle or automated driving system properly approved and operated within its authorised conditions?

Civil liability: Who must compensate the injured person under the applicable national law and relevant EU rules?

A vehicle's approval does not automatically establish that it was free from defects in every real-world situation. Nor does approval necessarily immunise a manufacturer or operator from a civil claim.

D. Germany: Codified liability and automated-driving rules

Germany combines its Civil Code (Bürgerliches Gesetzbuch, or BGB) with road traffic legislation, including provisions specifically addressing highly and fully automated driving.

German law is especially important for comparison because it retains established principles of vehicle-owner liability and driver responsibility while accommodating automated driving functions.

Relevant principles include:

Owner liability under § 7 of the Straßenverkehrsgesetz (StVG), subject to statutory conditions and exceptions.

Driver liability under § 18 StVG, subject to the provision's requirements.

The general tort provision in § 823 BGB.

Product liability and contractual remedies where a vehicle or component is defective.

Statutory rules concerning the use and supervision of automated driving functions.

The German model demonstrates that civil-law systems can adapt established liability provisions to autonomous technology without necessarily treating the software itself as an independent legal person.

E. United States: Federal vehicle standards and state-based liability

The United States uses a more decentralised approach. Federal authorities regulate vehicle safety standards and investigate safety defects, while states play major roles in vehicle operation, licensing, insurance and ordinary civil liability.

Consequently, the legal position can differ between states, particularly regarding negligence, product liability, comparative fault and damages.

Federal vehicle-safety regulation may determine whether a vehicle complies with applicable standards, but a civil court may separately examine whether the manufacturer, driver or operator breached a legal duty.

The model creates flexibility for testing and deployment, but fragmented rules can complicate multi-state operations, insurance arrangements and litigation involving software updates or remotely supervised fleets.

F. International harmonisation

International vehicle regulations developed through the United Nations Economic Commission for Europe (UNECE) seek to align vehicle-safety requirements across participating markets.

In June 2026, UNECE announced the adoption of a global regulatory framework for fully driverless automated driving systems, emphasising safety management, testing, validation and in-service monitoring. The framework's legal effect depends on the applicable international instruments and domestic implementation.

UNECE

International harmonisation can reduce differences in technical approval, but it does not automatically unify national rules on negligence, compensation, limitation periods or insurance.

4. Civil liability principles for autonomous vehicles

A. Negligence

A negligence claim generally requires the claimant to establish a duty of care, breach, causation and legally recoverable damage.

In an autonomous vehicle accident, the relevant breach might involve inadequate system design, failure to maintain sensors, negligent fleet supervision or misuse of a driver-assistance system.

The court must determine what precautions were reasonable in the circumstances. The fact that an AI system made a poor decision does not itself prove negligence by every party involved.

B. Product liability

A product-liability claim may arise where a vehicle, sensor, braking system or software component is defective.

Potential defects include:

Failure to detect pedestrians or cyclists in foreseeable conditions.

Inadequate emergency braking or collision-avoidance functions.

Software that fails to respond safely to sensor errors.

Defective steering or braking hardware.

Inadequate warnings concerning the system's limitations.

Unsafe software updates or cybersecurity design.

Depending on the jurisdiction, the claimant may rely on negligence, statutory product-liability provisions, consumer guarantees, contractual warranties or several causes of action together.

C. Driver and user responsibility

At lower levels of automation, the human driver generally remains responsible for supervising the road and using assistance systems appropriately.

At higher levels of automation, responsibility may shift towards the authorised operator or entity controlling the self-driving function. However, vehicle owners and users may retain duties concerning maintenance, permitted use, warnings and other matters.

A legal analysis must therefore establish whether the automated system was actually engaged, what operational conditions applied and which person or entity was responsible for the relevant driving function.

D. Insurance and compensation

Compulsory motor insurance is important because injured road users should not have to resolve every technical dispute before obtaining compensation.

A well-designed regime should provide an accessible route for compensation, followed where appropriate by recovery proceedings against a manufacturer, software developer, fleet operator or another responsible party.

Insurance does not necessarily eliminate the need to prove liability in every case. The scope of cover, exclusions, statutory benefits and rights of recovery depend on the applicable regime.

E. Data, evidence and cybersecurity

Autonomous vehicles may record video, sensor outputs, vehicle speed, braking events, software versions, system alerts and remote interventions.

This evidence can help reconstruct an accident, but it also raises questions about privacy, access, retention, reliability and disclosure.

A court may need to determine whether a system malfunction occurred, whether the vehicle was operating within its approved conditions and whether the responsible party preserved relevant records.

Cybersecurity incidents can complicate causation: an accident may result from a design weakness, a failure to apply a security update, an external attack or several contributing factors.

5. Eight important cases relevant to autonomous vehicle regulation

Legal qualification: Fully autonomous vehicles have generated relatively little binding appellate case law compared with conventional road accidents and product-liability litigation. The following cases include established judicial authorities that can be applied by analogy, together with a notable US autonomous-driving trial. The distinction matters: a court deciding a conventional negligence or product case does not necessarily decide the legal responsibility of an autonomous driving system.

Case 1: Donoghue v Stevenson [1932] AC 562

Jurisdiction: United Kingdom

Area: Negligence and manufacturers' duty of care.

Facts: A consumer alleged illness after finding a decomposed snail in a bottle of ginger beer. The product had been purchased by a friend, and the consumer had no direct contract with the manufacturer.

Legal principle: The House of Lords established a foundational modern principle of negligence: a manufacturer may owe a duty of care to the ultimate consumer where the relevant relationship and circumstances justify that duty.

Application to autonomous vehicles: An injured pedestrian may have no contract with the manufacturer of an autonomous car. Nevertheless, the pedestrian may be able to bring a negligence claim if the manufacturer owed and breached a duty of care, causing the injury.

For example, a vehicle manufacturer might face a claim alleging that it failed to take reasonable care when designing a pedestrian-detection system.

Significance: This case provides a foundation for manufacturer liability independent of contractual privity. It remains relevant when determining whether vehicle manufacturers owe duties to passengers, pedestrians, cyclists and other foreseeable road users.

Case 2: Grant v Australian Knitting Mills Ltd [1936] AC 85

Jurisdiction: Australia, on appeal to the Privy Council

Area: Consumer products and manufacturer liability.

Facts: Dr Grant developed dermatitis after wearing woollen underwear that contained excessive chemical residue from the manufacturing process.

Legal principle: The decision confirmed that a manufacturer can be liable in negligence for a defective product that causes injury to its ultimate consumer, even where the consumer did not contract directly with the manufacturer.

Application to autonomous vehicles: A defective braking component, faulty sensor or unsafe automated steering system may injure a person who never purchased the vehicle. The claimant may investigate whether the defect arose from manufacturing, design, testing or quality-control failures.

The case does not establish that every malfunction proves negligence. Evidence must still support the relevant legal elements.

Significance: This Australian authority is useful for analysing liability involving defective autonomous vehicle hardware and other safety-critical components.

Case 3: Wyong Shire Council v Shirt (1980) 146 CLR 40

Jurisdiction: Australia

Area: Negligence and reasonable precautions.

Facts: A man suffered serious injury while water-skiing in an area where he believed the water was sufficiently deep. The case concerned whether the council had taken reasonable precautions against the risk.

Legal principle: The High Court explained the assessment of reasonable foreseeability and breach of duty, considering matters such as the probability of harm, the likely seriousness of injury, the burden of taking precautions and the social utility of the relevant activity.

Application to autonomous vehicles: A manufacturer or operator should assess foreseeable hazards such as pedestrians crossing unexpectedly, cyclists entering a vehicle's path, temporary roadworks and sensor performance in poor weather.

The relevant question is whether reasonable precautions should have been taken against the risk, assessed in the circumstances known or reasonably knowable to the responsible party.

Significance: This case supplies a structured Australian negligence framework for evaluating autonomous driving safety measures, testing protocols and operational safeguards.

Case 4: March v E & MH Stramare Pty Ltd (1991) 171 CLR 506

Jurisdiction: Australia

Area: Causation and contributory negligence.

Facts: The case involved a collision between a truck parked in the middle of a road and a vehicle driven by a person who was intoxicated. The dispute concerned causation and the responsibility of the parties.

Legal principle: The High Court rejected an overly rigid approach to causation based solely on a mechanical application of the “but for” test. Causation requires a common-sense assessment of the circumstances, while legal responsibility also involves normative considerations.

Application to autonomous vehicles: Suppose an autonomous vehicle collides with another car after a sensor failure, while the other driver is speeding. A court may need to consider whether both events contributed to the collision and how responsibility should be allocated.

A technical malfunction does not automatically make the manufacturer solely liable, just as another driver's negligence does not necessarily absolve the manufacturer or operator.

Significance: The decision is important for multi-party autonomous vehicle accidents involving software, human conduct, road conditions and other contributing causes.

Case 5: Adeels Palace Pty Ltd v Moubarak [2009] HCA 48

Jurisdiction: Australia

Area: Breach of duty and causation.

Facts: A shooting occurred at a function venue, and the injured claimants alleged that the venue operator had failed to provide adequate security. The litigation examined whether the alleged breach caused the injuries.

Legal principle: The High Court emphasised that proof of breach does not automatically establish causation. The claimant must establish the necessary causal connection under the applicable legal test.

Application to autonomous vehicles: A manufacturer may have failed to install a particular safety feature, or a fleet operator may have inadequately monitored its vehicles. The claimant must still demonstrate that the breach caused or materially contributed to the injury in the legally relevant sense.

For example, evidence may be needed to show whether an emergency braking system, if properly designed, would have prevented or reduced the collision.

Significance: This case highlights the importance of accident reconstruction, technical evidence and causal analysis in autonomous vehicle litigation.

Case 6: Roads and Traffic Authority of NSW v Dederer [2007] HCA 42

Jurisdiction: Australia

Area: Duty of care, foreseeable risk and reasonable precautions.

Facts: A young person dived from a bridge into shallow water and suffered catastrophic injury. The claim concerned whether the public authority responsible for the bridge had failed to take reasonable precautions.

Legal principle: The High Court examined the relationship between foreseeable risk, the defendant's duty and the precautions that reasonable care required. Foreseeability does not mean that every possible precaution must be taken.

Application to autonomous vehicles: Regulators, road authorities and fleet operators must consider which safety precautions are reasonable in light of the risks, available technology and circumstances.

For example, an autonomous vehicle operator may need to evaluate whether a particular operating zone presents foreseeable hazards that require additional restrictions or monitoring.

Significance: The case helps distinguish a foreseeable risk from a legally established breach. It also illustrates why the standard of reasonable care must be assessed in context rather than by assuming that every accident proves a failure.

Case 7: A v National Blood Authority [2001] 3 All ER 289

Jurisdiction: England and Wales

Area: Statutory product liability and defect.

Facts: Patients received blood products contaminated with hepatitis C. The litigation concerned whether the products were defective under the Consumer Protection Act 1987.

Legal principle: The court considered the statutory test of product safety and the expectations of persons generally, taking account of the circumstances relevant to the products in question.

Application to autonomous vehicles: Product-liability legislation may provide a separate route to compensation when an autonomous vehicle or component fails to provide the safety that persons are generally entitled to expect.

The precise test differs by jurisdiction, and the case does not mean that every accident or software error necessarily constitutes a statutory product defect.

Significance: This case is relevant to strict or statutory product-liability models, in which the claimant's case may differ from a traditional negligence claim. It helps frame questions about defective sensors, unsafe automated braking and vehicle-system design.

Case 8: Huang v Tesla, Inc. — Florida federal trial and jury verdict (2024)

Jurisdiction: United States

Area: Driver assistance, warnings and product-liability litigation.

Facts: This case is often confused with another prominent Tesla Autopilot fatality case. In the 2024 Florida litigation, the family of a man killed in a Tesla crash alleged that the vehicle's Autopilot system and related design or warning issues contributed to the fatal collision.

Outcome: The jury found Tesla not liable in the case. The verdict was a trial-level outcome, not a binding appellate ruling establishing a universal rule for autonomous vehicle liability.

Legal significance: The case illustrates how courts and juries may examine driver conduct, system capabilities, warnings and the evidence linking an automated driving feature to a crash.

Application to autonomous vehicles: Litigation may turn on whether the technology was genuinely capable of performing the driving task without human supervision or was merely a driver-assistance feature requiring the driver to remain attentive.

This distinction is central to determining whether responsibility should rest with the driver, manufacturer or another party.

Important limitation: Autopilot is not equivalent to a fully autonomous Level 4 or Level 5 driving system. This case should therefore be used as an example of litigation involving automated driving technology, not as proof that courts have settled the liability rules for fully driverless vehicles.

6. Comparative case-law analysis

The cases above reveal several principles that recur across common-law and civil-law approaches.

These authorities do not create a single universal liability rule. Their relevance depends on the jurisdiction, cause of action and governing statutory framework.

7. Comparative civil liability: Australia, the UK, Germany and the US

A. Australia compared with the United Kingdom

Australia and the UK both draw on common-law negligence and product-liability principles. However, the UK has enacted a dedicated legislative framework specifically designed to distinguish self-driving operation from conventional driving.

Australia is developing nationally consistent arrangements for automated vehicles, with conditional deployment targeted for selected locations from 2027. The UK model provides a useful comparator because it expressly addresses authorisation and the responsibilities of entities overseeing self-driving functions.

National Transport Commission

+1

B. Australia compared with Germany

Germany's civil-law framework integrates automated-driving rules into an established statutory structure. Owner liability, driver responsibility and product liability are assessed through specific statutory provisions rather than relying exclusively on common-law negligence.

The comparison illustrates that a civil-law jurisdiction can preserve familiar compensation mechanisms while defining how automated driving affects responsibility. Australia can pursue a similar outcome through coordinated legislation without necessarily adopting a comprehensive civil code.

C. United States compared with the European Union

The US approach gives states an important role in operational and civil liability questions, while the EU has developed a more harmonised framework for vehicle type approval and technical safety requirements.

Neither model eliminates the need for case-specific civil litigation. Technical approval, compliance with safety standards and proof of a product defect are different questions, and their relationship depends on the governing law.

8. Major legal challenges and proposed reforms

A. Defining the responsible legal entity

Legislation should clearly identify the entity responsible for a vehicle while its autonomous driving system is engaged. Responsibility should be traceable even where the vehicle has no human driver.

This does not require treating AI as a legal person. Liability can remain with the organisations and individuals responsible for designing, supplying, authorising, maintaining or operating the system.

B. Establishing a fair compensation mechanism

An injured person should not face an insurmountable evidential burden merely because the accident involved complex software.

A robust system should provide:

Accessible compulsory insurance or compensation arrangements.

Preservation of relevant vehicle and system records.

Appropriate access to technical evidence.

Clear rules for contribution between responsible parties.

Effective remedies against manufacturers or operators where liability is established.

C. Regulating software updates

Autonomous vehicles may change their behaviour after remote software updates. Regulators should require appropriate validation, version control and records of safety-critical changes.

Where an update introduces a defect, the investigation should examine who designed, approved, distributed and installed it, and whether the defect caused the accident.

D. Protecting vulnerable road users

Pedestrians, cyclists, children and persons with disabilities may face particular risks if an autonomous system performs poorly in unusual road conditions.

Safety assessment should therefore consider real-world operating conditions rather than relying exclusively on performance in controlled tests.

E. Privacy and data governance

Vehicle cameras, location records and sensor systems may collect personal information about passengers and members of the public.

Regulation should define appropriate rules for data collection, retention, access, cybersecurity and disclosure in accident investigations, while preserving lawful access to evidence needed for civil claims.

9. Conclusion

Comparative autonomous vehicle regulation seeks to reconcile technological innovation with road safety, accountability and effective compensation for injured persons.

Australia, the United Kingdom, Germany, the European Union and the United States use different legal structures, but each must address common questions: who controlled the vehicle, whether the system was defective, what precautions were reasonable, whether the breach caused the injury and which party should bear the resulting loss.

The eight cases discussed above provide important foundations for answering these questions. Donoghue v Stevenson and Grant v Australian Knitting Mills establish the importance of manufacturer duties; Wyong Shire Council v Shirt and Roads and Traffic Authority v Dederer illuminate reasonable precautions; March v Stramare and Adeels Palace address causation; A v National Blood Authority concerns product defects; and Huang v Tesla illustrates the factual complexity of litigation involving automated driving.

The principal conclusion is that autonomous vehicle regulation should not depend solely on traditional driver liability. It should combine clear safety authorisation, identifiable operational responsibility, reliable technical evidence, appropriate insurance and fair civil remedies. That approach allows the law to respond to increasingly automated transport without leaving injured road users without an effective route to justice.