Complexity abounds for AV liability

By Michael Majstorovic*

The current framework in clarifying liability where a person has been directly injured from an automated driving systemis inadequate.

The framework creates a complex pathway to compensate people injured by an automated driving system, to the extent that they are worse off than a person who has been injured by a human-driven vehicle.

Fault-based compulsory third party (CTP) schemes are configured around traditional vehicle classes and an insured human driver, not an algorithm, being at fault. The liability issues have implications for insurers, as they can limit their ability to recover losses through subrogation under common law. Law reform, such as the expansive approach recommended by the National Transport Commission (NTC), is warranted to limit losses that might result for injured people and insurers.

Three liability concerns arise where an injury has been caused by an automated driving system. They can create a complex pathway to motor accident injury insurance (MAII) compensation for injured persons. The concerns also create difficulties for insurers in recovering costs through subrogation. That suggests legislative reform is warranted to prevent unfair losses for those parties.

MAII schemes do not contemplate AV crashes

There are two narrow threshold definitions across CTP schemes and national injury insurance schemes (NIIS) in Victoria and Queensland that may not accommodate injuries caused by automated driving systems.

The first is the definition of motor vehicle. In most jurisdictions, a motor vehicle can be an AV except for schemes like Qld, which defines a motor vehicle on the basis of whether it has been registered in accordance with Qld vehicle standard laws.

That suggests AVs that have not complied with those standards will not be covered by CTP schemes and NIIS.

Second, a personal injury is often required to be caused by a driver or the ‘driving’ of a motor vehicle. If an automated driving system inflicts an injury, it is not a person ‘driving’ a vehicle. Therefore, access to MAII schemes may be barred.

Reform is warranted as people injured by an automated driving system can be worse off than if they were injured by a human-driven vehicle. Some experts suggest definitions could be rephrased by focusing on the movement of the vehicle.

That might include replacing human-related criteria like ‘driving’ with AV-friendly words like ‘collision’ or ‘defect’. NTC has recommended an expansive approach to current MAII schemes, suggesting ‘driving’ in ‘the driving of the vehicle’ be replaced with technology neutral terms like ‘operation’.

Fault-based/hybrid CTP schemes – AVs are not legal entities

Even if definitions were reformed to accommodate AVs, fault-based and hybrid CTP schemes operate on the basis that fault is attributed to an insured person (owner or driver), not an algorithm with no legal presence.

For example, South Australia’s Motor Vehicles Act 1959 contemplates that the insured party must be a ‘person’ – the policy will ‘insure the owner … [or] any other person who at any times drives… [for] all liability that may be incurred by the owner or other person’.

Therefore, claims under fault-based CTP schemes/common law negligence will not be possible where the injury has been caused by an engaged automated driving system. Reform is warranted to ensure fault can be established and to ensure people do not need to recover losses through complex, expensive proceedings under other areas of law (eg, consumer law).

The most essential reform is ensuring there is always a legal entity responsible for an automated driving system when it is engaged. In NTC’s Changing driving laws to support automated vehicles policy paper, the majority of stakeholders favoured the view the vehicle’s manufacturer must hold accountability.

The stakeholders even consisted of AV manufacturers themselves, like Volvo, which said it was willing to accept liability for its highly and fully automated vehicles.

NTC recommended that, as a nationally consistent approach, the legal entity responsible for an automated driving system be the entity identified under a new safety assurance certification system (eg, the manufacturer). The entity will be known as the automated driving system entity (ADSE) and have a legal corporate presence for ease of liability.
 
Another key reform is clarifying which party is responsible at the various levels of automation. That is because complex causational issues can arise, especially with conditionally automated vehicles.

NTC has recommended that, for vehicles with conditional, high or full automation, the ADSE be taken to be in control of the vehicle, rather than the fall-back human.

However, ADSEs will need a rebuttable presumption where injuries have been sustained by something occurring outside the engagement of the dynamic driving task. For example, ADSEs should not be responsible for ensuring passengers wear seatbelts or for maintaining the vehicle and updating software.

Additional liability barriers – access to data

Enhanced access to telematic vehicle-generated data is crucial for determining liability for claims involving AVs.

An example of enhanced access can be seen in the July 2021-enacted German Road Traffic Act, which requires black-box data capturing technology to be fitted into AVs and has other provisions about who can access the data and how.

A key concern in enabling data access is that insurers must develop a comprehensive understanding of privacy and ethics issues involved with vehicle-generated data, like locational information. NTC has recommended balanced reform towards eased data access and noted additional benefits like lower premiums.

Further discussion on reform

Establishing liability for MAII compensation in fault-based jurisdictions will be difficult where an injury has been caused by an automated driving system.

Legislative reform is warranted to clarify liability and ensure MAII schemes are preserved.

NTC’s approach is to nationally expand current MAII schemes. This author also favours this approach because it would provide streamlined accountability and certainty in the short-to-medium term, which is appropriate, given the low presence of AVs in Australia’s total motor vehicle fleet.

There are concerns an expansive approach would create cost-shifting issues as the cost of an injured person caused by an automated driving system would not be borne by the negligent party (ADSE), as premiums are paid by vehicle owners and the cost of the scheme is borne by insurers, governments and taxpayers.

However, it has been argued that enhancing insurers’ recovery rights, rather than a national reinsurance pool, could assist. Those rights might be enhanced through adopting a WorkCover model of recovery or entitling MAII insurers to automatic recovery against ADSEs.

There are varied alternative approaches that can be adopted. One is to remove fault-based CTP liability schemes.

It is argued this would not be impractical as other schemes, like NIIS, are provided on a nationwide no-fault basis, which shows states and territories can work together to create consistency.

Other approaches, like radical overhauls to framework, as seen in the UK’s single insurer model, are appropriate but should be considered for the long term when Australia has more AVs in its fleet.

This second part of the research essay examined various constraints on establishing liability for MAII schemes and discussed whether reform was warranted. Existing MAII schemes are narrow and do not contemplate automated driving systems.

Fault-based schemes require fault to be attributed to an insured person (owner or driver), which can create complex pathways to compensation if the injury has been caused by a non-human automated driving system. A nationally consistent approach, such as NTC’s recommendations, will play an important part in ensuring injured people and insurers experience a regulatory environment that has certainty and consistency in identifying a responsible party for compensation and recovery purposes.

An expansive approach to current schemes is reasonable in the short-to-medium term, given the low presence of highly automated AVs in Australia, but it is likely a more drastic overlay to liability will be appropriate in the long term.

*Michael Majstorovic is the University of Technology Sydney winner of an AILA student prize for achieving top marks in insurance law subjects. This article is an edited version of the second part of a research essay on autonomous vehicles. The first part, which examined AVs’ impacts on motor accident injury insurance schemes, product liability insurance and cyber insurance, was published in the March 2023 issue of Resolve.