INTRODUCTION
The notion of vehicles that could function without human control is no longer simply a product of imagination but is also a reality of technology. The rapid advancements in computing and artificial intelligence have significantly accelerated the development of unmanned vehicles. Though automation was present in earlier times in aviation technology in the form of autopilots, it was always within a structured and controlled environment and involved human supervision at all times. The autopilots of even highly advanced aircraft cannot function without human supervision because of legal and safety requirements.
On the other hand, the idea of a completely autonomous road vehicle assumes a system that can successfully operate within a dynamic and changing traffic environment without the need for real-time human intervention. According to the classification system developed by the Society of Automotive Engineers (SAE), the more advanced forms of vehicle automation, specifically Levels 4 and 5, transfer control from the human driver to the vehicle’s onboard systems. At Level 5, the vehicle is assumed to be able to perform all driving tasks under all conditions without the need for a human driver to be present and attentive. Even though this system of completely autonomous vehicles is still under development, this model of vehicle automation represents the popular idea of the “self-driving car”: a vehicle that can perceive the world around it, make decisions regarding vehicle operation, and perform these operations without the need for human intervention, apart from setting the vehicle’s final destination.
The proponents of autonomous vehicles believe that such vehicles are likely to bring numerous safety and efficiency advantages. For instance, it is estimated that human error is a major cause of road accidents. Therefore, it is expected that replacing human drivers with algorithms will reduce accidents resulting from human errors such as fatigue, distraction, intoxication, and poor judgment. Another reason why autonomous vehicles are likely to be a major innovation in modern transportation policy is that it is expected that such vehicles will be able to improve traffic flow and allow users to be productive during travel.
Despite these expected advantages, the use of self-driving cars is also surrounded by complex issues concerning regulation and responsibility. The regulation of road traffic has historically been based on the idea that a human driver is in control and responsible for the operation of the vehicle. This basic assumption is undermined if control is transferred to autonomous technology. The question of legal status is the first question in this context: are these vehicles merely tools used by humans, or are they independently operating technology, the malfunctioning of which gives rise to a different kind of legal responsibility?
The pertinent issue, from a legal perspective, then becomes the attribution of liability in the event of an accident. When there is an accident involving an autonomous vehicle, there are several potential liable parties, including the occupant of the vehicle, the manufacturer, the software developer, the party responsible for updating the system, etc. Traditional negligence standards are based on the driver’s failure to act with reasonable care. However, where there are no human judgment inputs, there are concerns regarding the ability to apply these standards. This has, in turn, led scholars to consider whether there are alternative liability schemes, including those based on product liability, which can be used to deal with the consequences of autonomous vehicle use.1 2
The development of autonomous mobility, therefore, does not only pose a technological issue but also a doctrinal issue. With the advancement of automation, there is a need to reassess the concept of responsibility, allocation, and enforcement in an environment where human involvement is minimised.
WHAT ARE SELF-DRIVING CARS?
An autonomous vehicle can be defined as a self-driving car. A self-driving car is a vehicle that can drive without human intervention. Unlike ordinary cars that rely on a human being to perceive the environment, make decisions, and make the vehicle move, an autonomous vehicle makes use of various technologies to make its own decisions. Such technologies include cameras, radars, lidars, ultrasonic sensors, and computing systems. All these technologies work together to help the vehicle perceive its environment as well as navigate. For instance, they can perceive road conditions, traffic lights, obstacles, as well as other road users.3
Essentially, a self-driving car is one where human judgment is replaced by algorithmic decision-making. Once a destination is input and/or a route is selected, it analyzes information in real-time and compares it with pre-mapped information and decides on the safest and most efficient way of travel. In effect, it is not only carrying out its mechanical functions of moving and turning but is also carrying out cognitive functions, i.e., those of a human driver.
The level of automation of self-driving cars is classified under a framework set out by the Society of Automotive Engineers (SAE), which categorizes there are six distinct levels of driving automation:
| Level | Description |
|---|---|
| Level 0 | No automation; the human driver performs all driving tasks. |
| Level 1 | Driver assistance for either steering or speed control, but not both simultaneously (e.g., cruise control). |
| Level 2 | Partial automation; the vehicle can control steering and acceleration/deceleration concurrently, though the driver must continuously supervise. |
| Level 3 | Conditional automation; the system performs driving tasks under specific conditions, but the human driver must be prepared to intervene when prompted. |
| Level 4 | High automation; the vehicle can operate independently in defined environments or operational domains without human input. |
| Level 5 | Full automation; the vehicle is capable of performing all driving functions under all conditions without any human involvement. |
This graduated model demonstrates that, in fact, there is a range of degrees of automation, not a simple dichotomy between human-driven and autonomous vehicles. The vast majority of vehicles currently in public use are at Levels 1 and 2, in which human intervention is both legally and functionally necessary.
The main purpose of autonomous vehicle technology is to make road transportation safer, more efficient, and more convenient. Because a significant cause of traffic accidents is human error, autonomous technology, if designed and tested appropriately, could potentially reduce accidents resulting from distraction, fatigue, drunkenness, and misjudgment. Furthermore, autonomous traffic flow control and vehicle-to-vehicle communication could make vehicles more fuel-efficient, reduce traffic jams, and optimize route planning.4
The operation of autonomous cars is based on sophisticated technology integration. Computer vision is responsible for object and marking detection. Machine learning is responsible for pattern recognition. Sensor fusion integrates multiple detection technologies for improved accuracy. Lastly, there is real-time operation for timely reaction to dynamic situations. These technologies integrate and create a sophisticated technology stack for executing vehicle operations such as lane change, merging, braking, and obstacle avoidance with a high degree of accuracy.
However, as there is a shift from human operation to autonomous operation, traditional concepts of liability need reevaluation. Traditionally, liability is assigned based on drivers and vehicle owners. However, as there is a shift towards autonomous operation, liability is extended beyond drivers and owners of vehicles. Some scholars have argued that traditional tort principles, especially those based on negligence, may not be sufficient for addressing algorithm-based decisions.5
In this respect, the legal landscape surrounding autonomous vehicles is still in its developmental stages. In this context, there are challenges that the relevant authorities must balance in their attempts to foster innovation in the technology and, at the same time, ensure safety in public use. It is important to establish the balance of liabilities among different stakeholders in the use of autonomous vehicles, including drivers, manufacturers, insurance underwriters, and developers of the technology. As different jurisdictions test different models of legislation and regulation, the development of autonomous vehicle law will depend on the interaction of different stakeholders in the field and in the legal arena.
TECHNOLOGICAL COMPONENTS AND CAPABILITIES
The operational capacity of autonomous cars is made possible through the integration of several interdependent technologies. Currently, autonomous cars are equipped with multi-layer sensing technologies, which allow them to perceive their environment. Some of these technologies include LiDAR, radar, high-definition cameras, and ultrasonic sensors. LiDAR, radar, and ultrasonic sensors allow the vehicle to create a three-dimensional spatial map of the environment. High-definition cameras, on the other hand, are used to interpret road signs and lane markings. These technologies provide real-time data regarding the surrounding environment, including other cars, pedestrians, road signs, and weather conditions. This data is then sent to the vehicle’s onboard computer, which processes and integrates the data to enable movement.
Autonomous cars are also equipped with sophisticated control technologies, which enable them to take action based on the data received. These technologies include steering, acceleration, and braking, which are controlled electronically. These technologies are very sensitive, as even the slightest computation can have significant safety implications.
At the core of this autonomous functionality is artificial intelligence, such as machine learning models that have been trained on vast amounts of data. These models process the information that is fed to them, learn patterns, predict the actions of other users, and react accordingly to the driving situation. This is different from the traditional method of programming, which does not change with time, thus improving the functionality of the vehicle in different scenarios. However, this adaptability has brought some challenges, especially in the process of verifying the safety compliance of the system in different scenarios.6
High-definition maps and localisation functionality improve the reliability of the system, with the ability to achieve centimeter-level positioning accuracy. This is done by using the GPS system, which combines the satellite navigation system with the environmental maps that the vehicle has been pre-programmed with, thus creating the ability to locate the exact position within the network. The connectivity functionality, which includes vehicle-to-vehicle (V2V) and vehicle-to-infrastructure (V2I), provides the vehicle with the ability to receive information from the cloud, such as updates, traffic information, and diagnostics, which improves the situational awareness but has brought some challenges, such as cybersecurity issues.
Safety assurance of such vehicles is also heavily reliant on aspects of redundancy and fail-safe design. The safety of such vehicles is ensured through comprehensive safety standards that place significant emphasis on systematic safety analysis and testing. The development of safety frameworks for fully autonomous vehicles is also indicative of the integration of safety and engineering design.
The different technological aspects of autonomous vehicles, such as perception systems, artificial intelligence systems, control systems, mapping systems, connectivity systems, and safety redundancies, collectively form a comprehensive ecosystem that enables fully autonomous vehicles to function in complex and dynamic road environments. The complexity of such systems is indicative of a major leap forward in terms of transportation technology; however, it also simultaneously requires safety standards and regulations to ensure that such innovation does not compromise safety and welfare.
Difference between Self-Driving Vehicles and Autonomous Vehicles
There is a need to make a conceptual distinction between vehicles described as “self-driving” and those which are, in fact, autonomous in the context of high-level automation standards. The reality is that, in commercial terms, the description used by the manufacturer may not reflect the true technical capabilities of the technology installed in the vehicles available for public purchase. The implication may be that the vehicle is capable of autonomous operation in all conditions, whereas in fact, the technology may only be a high-level driver-assistance system.
Driver-assistance technologies, often referred to as “Autopilot” or similar terms, are typically at Levels 1 and 2 in the SAE scale. At these levels, the technology may control both steering and acceleration at the same time or, alternatively, provide lane-keeping and adaptive cruise control functions. Nevertheless, ongoing and continuous human supervision is both legally and functionally required. The human driver must remain vigilant, situationally aware, and ready to intervene at all times.
For instance, Tesla has stated that its Autopilot feature is intended for fully attentive driving and doesn’t remove the driver’s obligation of monitoring the road and remaining prepared to take control of the vehicle when needed.7 Additionally, it demands that the driver keep his hands on the steering wheel and react quickly if situations arise beyond its operation design limits. Thus, it is not categorized under Level 5, where there is no need for monitoring.
The blurring line between partial and full automation creates safety and legal issues. For instance, there is a risk that people using driver-assistance technology will grow complacent and lose active attention to the road, leading to accidents. In terms of legal issues, there is a risk that the misunderstanding of the capabilities of technology will make it difficult to determine who is at fault in the event of an accident, particularly if the driver believes that the car can operate independently and the law requires that there be human oversight.
In spite of the challenges that have arisen in the development and deployment of partially automated vehicles, the deployment of this technology in the market has gained traction in recent times. Large automobile manufacturers, including Tesla, BMW, General Motors, and Volvo, have all adopted this technology in their production vehicles. In addition, technology companies, including Google through its subsidiary dedicated to autonomous vehicles and Uber, have all made investments in the development of this technology as well.
It is therefore important to grasp the technical and legal distinction between assisted driving and autonomy. The former is designed to assist, while the latter is designed to replace, human control. Unless and until vehicles are widely available and certified at Levels 4 and 5, human drivers remain at the heart of the chain of responsibility in most road traffic scenarios.
LIABILITY IN THE CONTEXT OF AUTONOMOUS VEHICLES
For conventional motor accidents, the compensatory framework is typically designed in accordance with traditional principles of tort law, such as negligence-based (fault-based) liability and, in some circumstances, strict liability. The framework of conventional motor accidents is founded upon the presumption that a human actor is in control and has effective control over the motor vehicle, thus making them the primary source of liability. The concept of autonomous motor vehicles, therefore, presents a challenge to this fundamental presumption. The shift in control from the human actor to the technological framework is more complex and presents novel questions in determining liability.
For conventional motor accidents, the framework is typically founded upon the concept of determining whether the motorist breached their duty of care by failing to take reasonable action in the circumstances. The concept of autonomous motor vehicles, therefore, potentially extends the scope of liability from the motorist to the manufacturer, programmer, integrator, and/or provider.
A. Fault-Based Liability (Negligence Theory)
Negligence principles require that there be evidence of duty, breach, and causation. In usual cases of road traffic accident claims, the courts will assess whether the driver would have acted as a reasonably prudent man in the circumstances. This includes compliance with road signs, driving at a reasonable speed, being vigilant, and not being intoxicated.
Autonomous vehicle technology complicates the analysis because different levels of human intervention exist. For SAE Levels 1 to 3 of autonomous vehicle technology, there is still driver control. In this situation, there will be evidence of negligence if the driver fails to take control of the vehicle after being prompted to do so. The courts will assess whether the driver’s inaction was reasonable.8
Nevertheless, as the automation levels get closer to Level 4 and Level 5, where the vehicle can perform all driving functions within defined operational domains, the basis on which driver negligence is held becomes less valid. Since the human on board does not have the ability to control or influence vehicle movement, the basis on which negligence is held becomes less real.
B. Strict Liability Frameworks
The application of strict liability eliminates the burden of proof, and the basis for such liability is the occurrence of harm resulting from the activities or products in question. The application of strict liability in the context of autonomous vehicles is a pragmatic solution to the problem of evidentiary asymmetry. Victims of accidents resulting from autonomous vehicles may not have access to proprietary data and information, which may be necessary in determining negligence.
The basis for the application of strict liability in autonomous vehicles is grounded in the theoretical justification, which is based on the allocation of risk. The rationale is that those involved in the production, design, and distribution of autonomous vehicles are in the best position to mitigate defects, provide adequate insurance, and apportion risk.9 The application of this theory may encourage the testing and safety of autonomous vehicles, but this has to be carefully balanced in order not to undermine innovation.
C. Product Liability
On the other hand, there are more particularistic measures that can be taken in the context of product liability with respect to harm caused by autonomous vehicles. These can be grouped under three different categories:
1. Design Defects – Where the design of the vehicle or the programming of the autonomous systems makes them unreasonably dangerous.
2. Manufacturing Defects – Where there was a flaw in the production of the autonomous vehicles.
3. Failure to Warn – Where there was a failure to adequately warn consumers about the risks of driver assistance systems.
With more and more automation on the road, accidents will not be as readily attributable to driver error, but to perception systems, decision-making, and validation testing. It is possible that there will be an evaluation of whether or not the autonomous vehicles performed in a reasonable fashion.
Software developers will also be subject to liability in the context of autonomous vehicles, particularly in light of the fact that there is so much reliance on machine learning and data processing in autonomous vehicles. Post-sale updates will be a critical area of evaluation in this context.
D. Operator Negligence and Residual Human Responsibility
Notwithstanding the advances in automation, the majority of existing vehicles on the road require some level of human intervention. Regulating bodies, such as those in California and other American states, require that there be the capacity for human intervention in the event of an emergency. Thus, negligence actions can arise where there is misuse of driver-assistance technologies, where there is failure to heed the technologies, and where there is non-compliance with vehicle maintenance recommendations.
Defences in negligence actions are likely to involve showing compliance with operational recommendations, as well as showing that there was no unreasonable failure to intervene. Given the presence of human intervention, there is likely to be a continuation of hybrid models of liability during the transition phase.
E. Shared Liability Models
However, some countries prefer the distributed responsibility approach, which considers the possibility that the accident could be caused by a combination of technological failure and oversight error. The allocation of responsibility may be to the manufacturers, owners, software developers, and, in some cases, the users of the vehicle, depending on the circumstances, which may be attributed to the fact that the autonomous vehicle operates within a larger environment that includes maintenance, updates, and compliance with regulatory requirements.
Studies comparing the various approaches to autonomous vehicle liability allocation show that the allocation of responsibility differs significantly from one jurisdiction to another, depending on various factors such as policy and the level of technological development in the different countries. The lack of international standards is another factor that has contributed to the confusion regarding the allocation of responsibility.
F. No-Fault Insurance Systems
In jurisdictions with no-fault insurance systems, the victim of the accident receives compensation from the insurance companies without the burden of establishing liability for the damage. In these systems, the liability issue may be resolved not as a tort case, but as an insurance claim case between the insurance companies and the manufacturer. Even though the no-fault system makes it easier for the victim to receive compensation, the issue of liability among the concerned parties still lingers.
G. Data Privacy and Cybersecurity Considerations
This is because autonomous vehicles produce large amounts of operational data, which is vital in accident scenario reconstruction and attributing blame. The operation of such data-intensive autonomous vehicles, however, poses major privacy and cybersecurity risks.
Any unauthorized access, data breach, or interference with the operation of the vehicle may become a source of liability in the future. Transportation safety regulatory agencies have issued guidelines that highlight the need to ensure the cybersecurity posture and transparency of autonomous vehicle operation. The insurers must not only consider the risks associated with physical collision but also those associated with cyber risks and regulatory compliance.
INSURANCE CONSIDERATIONS FOR SELF-DRIVING CARS
Traditional Auto Insurance and Product Liability Coverage
The emergence of autonomous vehicles thus calls for a review of the conventional motor insurance models. The conventional automobile insurance model is founded on the basis of allocating faults to drivers in road accidents. The conventional automobile insurance model thus covers third-party liability in cases of negligent driving, as well as personal injury and damage to property.
However, the emergence of autonomous technology, which shifts the control and operation of vehicles from the human mind to hardware and software technology, may thus alter the risk allocation from the insured individual to the manufacturer and developer of the technology. Thus, in cases where accidents happen due to errors in the algorithms and technology, the conventional automobile insurance model may thus not provide adequate coverage.
The emergence of autonomous technology thus calls for the assumption of greater significance of product liability insurance models. Product liability insurance protects the manufacturer and developer against any claims for defects in the design, production, and failure to warn. Thus, in cases where the autonomous vehicle is involved in accidents due to defects in the technological structure, the risk allocation may thus shift from the motorist to the manufacturer and developer.
Challenges and Opportunities for the Insurance Industry
The integration of self-driving vehicles also offers structural challenges and strategic opportunities for insurers. On one hand, it could reduce the number of accidents caused by human error and therefore lower claims for traditional motor insurance policies. On the other hand, it could also increase the complexity and severity of claims in the event of liability for sophisticated technology systems, intellectual property rights, and cybersecurity breaches.
To insure self-driving vehicles, actuaries will need to develop models that assess the reliability of technology rather than human behavior. This will involve assessing factors such as software validation processes, update cycles of systems, cybersecurity systems, and safety performance of manufacturers. The risk assessment could also become even more data-dependent and involve monitoring systems via telematics and performance monitoring. At the same time, there is an emerging market for new and specific forms of insurance products. These could be:
• Coverage for cyber risks, addressing liability for hacking, breaches, and interference.
• Tech errors and omissions coverage for software developers and integrators.
• Business interruption coverage for operators of fleets and other commercial users who suffer losses due to malfunctioning systems.
• Data liability coverage for addressing privacy concerns related to the collection and storage of vehicle operation data.
Hence, it is evident that there is a reduction of traditional risks related to drivers, but there is an expansion of risk categories related to technology.
Emerging Insurance Models for Autonomous Vehicles
Several conceptual models have been proposed to adapt insurance mechanisms to autonomous mobility:
1. Usage-Based Insurance: Usage-based insurance (UBI) relies on telematics data to calculate premiums according to vehicle usage patterns. For example, the premium charged to autonomous vehicles may be based on the distance traveled, domain, system performance history, frequency of human intervention, and compliance with traffic rules, among other factors. The high volume of performance data associated with autonomous vehicles means that UBI models could potentially be more precise in their assessment than traditional premium models.
2. Manufacturer-Based Insurance: In the manufacturer-based model, the vehicle manufacturer is held primarily liable for providing liability insurance. It can be packaged with the vehicle purchase or lease, based on the manufacturer’s authority over system design and updates. It can be seen as aligning risk with technological authority, and there is the potential for simplified compensation schemes, eliminating the need to determine fault between drivers and producers. It can also encourage manufacturers to emphasize safety and have good quality control.
3. Government-Supported or Public Insurance Models: In some jurisdictions, there have been discussions on public insurance schemes, including government involvement, in the context of autonomous vehicle insurance. Public insurance schemes can ensure universal insurance, quick victim compensation, and possibly reduced insurance premiums through pooled risks. Government involvement can also ensure consistency in regulations and public acceptance of new technologies. However, public insurance schemes must be accompanied by proper financial planning and the establishment of recourse rights on manufacturers in cases of system defects.
The Need for Coordinated Regulatory Development
The insurance regulation of autonomous vehicles is jurisdiction-specific and subject to change. The determination of financial responsibility will depend on general legislative decisions on tort reform, product liability, and cybersecurity compliance. The insurance governance of autonomous vehicles will require meeting three primary needs as the technology develops:
1. Providing timely and sufficient compensation to accident victims;
2. Allocating risk in accordance with technological control; and
3. Managing innovation without undermining safety incentives.
The move from driver-centric to system-centric mobility is not merely a technological shift, but a paradigm shift in insurance law and risk management.
REAL-LIFE CASES INVOLVING AUTONOMOUS VEHICLES
A. Uber Autonomous Vehicle Fatality (2018)
In 2018, a test vehicle under the autonomous driving division of Uber was involved in an accident where the vehicle struck and killed a pedestrian in Tempe, Arizona. This was the first reported incident where there was a fatality involving a completely autonomous test vehicle on public roads. However, at the time of the incident, the vehicle was in autonomous mode, but there was a human safety driver on board, seated in the driver’s seat.
As expected, there was an investigation into the incident by the National Transportation Safety Board (NTSB), where several factors were analyzed, including the vehicle’s perception system, the design constraints of the autonomous vehicle, and the actions of the safety driver. It was found that the autonomous vehicle had detected the pedestrian but failed to take adequate action. However, the question of liability extends beyond the safety driver to the organization.
This is an example of the multi-level liability of autonomous vehicle testing. There are several entities that are likely at fault, including the organization, the safety driver, the vehicle manufacturer, and possibly others depending on the circumstances.
B. Waymo Self-Driving Vehicle Incidents (2016)
In 2016, Waymo, Google’s autonomous vehicle division, announced a series of minor accidents during public road testing. These accidents consisted of interactions with other cars, bicycles, and objects along the roadside, as well as a number of “disengagements,” where a human had to take over operation of the vehicle.
Despite these accidents not having any human fatalities, they have sparked public discussion regarding autonomous vehicle testing and safety validation. These disengagements highlight the need for human intervention, even at advanced stages of testing. Determining liability for accidents involving autonomous cars is based on specific circumstances and may involve different parties, such as the autonomous vehicle operator, safety driver, other drivers, and even vehicle technology.
The Waymo accidents highlight transitional states of automation, where there is shared human and vehicle operation, and the complexity of determining liability based on evidence. Analysis of vehicle and driver information is necessary for determining liability, demonstrating the increasing use of digital evidence in vehicle accidents.
C. Tesla Autopilot Crash and Criminal Proceedings
Another class of cases deals with advanced driver assistance systems rather than autonomous driving. In 2019, a Tesla Model S, while in “Autopilot” mode, was involved in a fatal accident. Charges of vehicular manslaughter have been filed against the driver, Kevin George Aziz Riad. This is one of the first cases where the driver has been charged in relation to an accident involving semi-autonomous driving systems.
Tesla has consistently claimed that “Autopilot” is not an autonomous driving system but rather an assist system, which requires continuous supervision. Thus, the issue in the trial was whether the driver was paying attention and exercising due care while the system was in use. This is an important distinction in the law, where the system is not autonomous but rather assistive, and the driver is held legally responsible.
Besides its immediate legal consequences, it also raised some concerns about technological over-reliance and consumer knowledge. This is because, while marketing emphasizes freedom and independence in terms of technology, its design and operations require supervision and control. This is a behavioral factor that affects legal determinations of responsibility and potential judicial approaches to reasonableness in terms of new technology.10
Broader Implications
These real-life scenarios show us that autonomous vehicle liability is not only not speculative, but rather it is being determined incrementally.11 Industry analysis and commentaries have emphasized the role of public trust as an essential component in the development of autonomous mobility.12 13 Ethical scholarship has also emphasized the role of transparency, accountability, and realistic representation.
As autonomous technologies continue to progress towards greater levels of automation, the emphasis of the law is likely to continue moving away from driver conduct, towards system design, validation, and regulation. The federal transportation regulators are likely to play a crucial role in establishing safety standards, which will, in turn, have implications on civil and criminal liability.
The development of the law on this issue can be seen as a transitional phase, with existing disputes relating to ‘hybrid systems’ where humans and machines are jointly in charge. Fully autonomous systems, once they are widely adopted, might further change the parameters of transportation law.
COMPARISON OF LEGAL SYSTEMS’ LIABILITY FRAMEWORKS
India
The legal framework of India is based on common laws such as negligence, strict liability, and product liability laws. However, as of today, there is no specific legal framework that deals with the regulation of autonomous vehicles. The regulation of road traffic is mainly governed by the Motor Vehicles Act of 1988.
Section 2(28) of the MVA states that “Motor Vehicle” means any mechanically propelled vehicle adapted for use upon roads, whether the propulsion of the vehicle is from an external or internal source;14 and it is mandated that it should be registered, and it should be driven by a licensed driver. The MVA provides consequences for traffic offenses and negligent acts on roads. The legal framework is based on continuous human supervision and holds the driver liable for all acts of the vehicle.
This structural assumption also generates a doctrinal problem within the autonomous setting. The MVA does not account for algorithmic decision-making, software-based navigation, or the delegation of control to artificial intelligence systems. If the autonomous vehicle violates a traffic law or causes harm, the occupant or registered owner would presumptively be liable unless rebutted. Such an approach can yield unfair outcomes when the harm is caused by system malfunctions rather than the occupant or owner.
For example, liability issues will arise when the vehicle malfunctions due to defective software updates or cybersecurity breaches that compromise vehicle control. The recent news of vehicle manufacturers recalling beta software due to safety concerns also demonstrates the reality of these risks.15 Similarly, cybersecurity breaches in connected vehicles pose the risk of third-party interference. The occupant will not have the requisite knowledge and control while still being presumptively liable under the existing traffic laws.16 Although there is no autonomous vehicle-specific legislation, other relevant legislative instruments may be applicable, such as:
• The Consumer Protection Act, 2019, which provides for product liability actions in cases involving defective goods or deficient service.
• The Information Technology Act, 2000, which regulates electronic records, cybersecurity offenses, and digital transactions.
However, these legislative instruments are peripheral to the allocation of civil and criminal liabilities pertaining to the operation of autonomous vehicles, and scholarly comment has focused on the need to create a regulatory framework that deals with the operational design domain, testing authorization, standards of criminal attribution, and mandatory insurance schemes, etc., to alleviate the uncertainty that may surround the operation and deployment of such technologies.17 18
United Kingdom
The United Kingdom has followed a more focused legislative strategy with the passing of the Automated and Electric Vehicles Act 2018 (AEVA), which states that if the vehicle has been listed as safe for autonomous driving and the vehicle is involved in an accident while operating independently at the time of the crash, the insurance provider, and not the driver, becomes directly liable for the damage or harm caused.
This system ensures the prompt compensation of the victim while also allowing the insurer to seek compensation from the manufacturer if the vehicle has been subjected to unauthorized modifications or if the necessary software upgrade has not been installed, which caused the accident. The UK system also acknowledges the reduced role of the driver and holds the insurer liable when the vehicle operates independently. The AEVA also enables regulatory control over the listing of the vehicle, safety certification, and operation, which makes it a more comprehensive system.
China
China has followed a model of regulation through administrative guidelines on road testing and operational safety. Documents like “Guidelines on Road Testing of Intelligent Connected Vehicles (Trial)” and “Safety Technical Requirements for Road Testing of Intelligent Connected Vehicles (Trial)” have defined intelligent connected vehicles as those “which can partially or completely replace human driving through integrated sensing, control, and decision-making systems.”
In this system, the party conducting the road test is held primarily liable in the event of damage during the road test activities, unless they can prove that the damage was caused by “force majeure” or “the intentional acts of a third party.” However, the party must ensure the presence of a safety driver, who can take over the vehicle, and install data recorders during the road test. It must also report the occurrence of an accident to the authorities.
This model is based on the concept of administrative accountability and road testing rather than reforming the tort laws.
Comparatives
This is evident upon conducting a comparative analysis of each jurisdiction:
• India follows traditional tort-based approaches centred on drivers, which creates a conceptual gap in areas of reduced human involvement.
• The United Kingdom follows a legislative approach centred on insurer liability and is tailored for autonomous operations.
• China focuses on administrative oversight and testing accountability via regulatory tools.
However, it is evident that a unified global framework does not exist regarding liability for autonomous vehicles. There are also different approaches to determining levels of automation, primary responsibility, and cybersecurity integration. Resolving disputes is also not well developed.
Further analysis of academic studies also suggests that liability frameworks are not only linked to economic determinants but also to perceptions of public trust and mental health regarding automated systems.19 Legal determinations are therefore economically and socially relevant.
Need for Reform and International Coordination
This comparative analysis shows that the regulation of autonomous vehicles is in a state of transition. The regulation has to find the right balance between technological development, the need to protect the public, the need to compensate those who suffer, and the need to be clear and certain so that manufacturers and users know what to expect. What might be the legislative response in India? For India, the legislative response might include the following:
• What levels of autonomy will be included in the law?
• What will be the nature of the liability if the system malfunctions?
• What will be the standards that must be implemented to protect the system from cyber threats?
• What will be the insurance requirements to allow autonomous operation?
What might be the international response? At the international level, greater coordination will be needed to achieve harmonized safety standards and international consistency in operation. The absence of such coordination might lead to regulatory confusion that could hinder the development of autonomous vehicles.
CONCLUSION
The rise of autonomous vehicles signals a paradigm shift in transportation, moving control from human drivers to machine algorithms. This shift has significant implications for the fundamental principles of motor vehicle law, which place responsibility squarely on the shoulders of the individual. As autonomous vehicles progress along the spectrum from assisted to full autonomy, the principles of negligence, strict liability, and product liability must be adjusted to fit this new reality.
At the lower tiers of autonomy, traditional principles of fault remain applicable, but as control shifts to machine algorithms, so does responsibility, falling more squarely on the shoulders of the manufacturer, programmer, and integrator. The shift from behavioural fault to technological performance is significant and has led to a need to rethink the rules and principles governing liability in motor vehicle accidents. The principles of product liability and strict liability may provide more rational outcomes in cases where accidents are the result of technological failure.
This is evident from a comparative analysis of different nations’ approaches. India is currently governed by driver-centric legal presumptions under the Motor Vehicles Act of 1988, which does not directly address autonomous operations. The United Kingdom has introduced a new framework through the Automated and Electric Vehicles Act of 2018, wherein primary responsibility is fixed on insurers during autonomous operations. China is heavily reliant on administrative regulations and controlled testing scenarios.
Insurance systems are also witnessing a paradigm shift. The change in liability from drivers to corporate entities is leading to a shift in underwriting considerations as well. The reliability of technology and cybersecurity are likely to become integral aspects of new insurance systems, as are evident from new forms of insurance systems that are usage-based, manufacturer-based, and state-supported.
In fact, there have been real-life cases of autonomous and semi-autonomous technologies, which have demonstrated that the issue of liability is no longer just a theoretical concept. These cases have demonstrated the complexity of the issue of liability, especially in the context of “hybrid” systems, where there is a combination of human intervention and autonomous technologies.
Ultimately, the integration of autonomous vehicle technologies into the law requires a balanced approach, which promotes innovation while ensuring public safety. It requires lawmakers to define the concept of autonomy, establish standards on civil and criminal liability, ensure cybersecurity, and create insurance standards. It also requires international coordination, especially to avoid “regulatory fragmentation” and ensure the smooth deployment of these technologies internationally.
The development of autonomous vehicle technologies is not just a technological development; rather, it is a jurisprudential development. As the concept of control shifts from humans to intelligent technologies, the law must also shift in the same way.
*****
Footnotes
1. John K. Gurney, Sue My Car Not Me: Products Liability and Accidents Involving Autonomous Vehicles, 2013 U. Ill. J.L. Tech. & Pol’y 247, 251 (2013).
2. David LeValley, Autonomous Vehicle Liability – Application of Common Carrier Liability, 36 Seattle U. L. Rev. 5 (2013).
3. Synopsys, What Is an Autonomous Car?, https://www.synopsys.com/automotive/what-is-autonomous-car.html. (Accessed 04-03-2026)
4. Car Insurance for Self-Driving Cars: Challenges and Opportunities, Shriram General Insurance (Dec. 17, 2025), https://www.shriramgi.com/article/car-insurance-for-self-driving-cars (accessed Mar. 4, 2026).
5. Kenneth S. Abraham & Robert L. Rabin, Automated Vehicles and Manufacturer Responsibility for Accidents: A New Legal Regime for a New Era, 105 Va. L. Rev. 127, 130 (2019).
6. Tesla Press Information: Autopilot (2020), https://www.tesla.com/en_EU/presskit/autopilot (accessed Mar. 4, 2026).
7. Nat’l Highway Traffic Safety Admin., Automated Vehicles for Safety, https://www.nhtsa.gov/vehicle-safety/automated-vehicles-safety (last visited Mar. 4, 2026).
8. UNC Sch. of Law, Who Bears the Blame When Self-Driving Cars Cause an Accident? (Oct. 12, 2020), https://ncjolt.org/blogs/who-bears-the-blame-when-self-driving-cars-cause-an-accident (last visited Mar. 4, 2026).
9. Maria Lubomira Kubica, Autonomous Vehicles and Liability Law, 70 Am. J. Comp. L. Supp. 1 (2022).
10. Kambria, The History and Evolution of Self-Driving Cars (2019), https://blog.kambria.io/the-history-and-evolution-of-self-driving-cars/ (last visited Mar. 4, 2026).
11. MIT Tech. Rev. Insights, Self-Driving Cars Take the Wheel (Feb. 15, 2019), https://www.technologyreview.com/2019/02/15/137381/self-driving-cars-take-the-wheel/ (last visited Mar. 4, 2026).
12. Patrick Lin, The Ethical Dilemma of Self-Driving Cars, YouTube (2015), https://www.youtube.com/watch?v=ixIoDYVfKA0 (last visited Mar. 4, 2026).
13. MIT Tech. Rev., Self-Driving Cars Take the Wheel (Feb. 15, 2019), https://www.technologyreview.com/2019/02/15/137381/self-driving-cars-take-the-wheel/ (last visited Mar. 4, 2026).
14. Motor Vehicles Act, No. 59 of 1988, § 2(28), India Code (1988).
15. BBC News, Tesla Withdraws Self-Driving Beta over Software Issues (Oct. 25, 2021), https://www.bbc.com/news/technology-59037344 (last visited Mar. 4, 2026).
16. Sascha Brodsky, How Self-Driving Cars Can Be Hacked, Lifewire (2021), https://www.lifewire.com/how-selfdriving-cars-can-be-hacked-5114337 (last visited Mar. 4, 2026).
17. Essenese Obhan & Shourya Paul, Driving into the Future: Regulating Autonomous Vehicles, Mondaq (2023), https://www.mondaq.com/india/patent/1239240/driving-into-the-future-regulating-autonomous-vehicles (last visited Mar. 4, 2026).
18. Rohit Ray, Liability for Self-Driving Vehicles: Is There Anyone to Blame?, LiveLaw (2022), https://www.livelaw.in/lawschoolcolumn/liability-for-self-driving-vehicles-automated-cars- (last visited Mar. 4, 2026).
19. Yineng Xiao & Zhao Liu, Accident Liability Determination of Autonomous Driving Systems Based on Artificial Intelligence Technology and Its Impact on Public Mental Health, 2022 J. Env’t & Pub. Health art. e2671968 (2022).