My dissertation evaluated the suitability of the current liability regime in respect of autonomous vehicles (AVs), and considered whether software developers could be held liable for accidents arising from their use. The liability regime has been relatively untouched for many decades and has always been designed with the view that the operator would be in charge. With autonomous vehicles set to be introduced, potentially within the next three years, it has been necessary to assess whether the current scheme is suitable for if, and when, accidents arise.
My dissertation began by examining what autonomous vehicles are, their potential benefits and how the current uncertain state of liability may undermine developments.Autonomous vehicles – those which are able to control the lateral andlongitudinal driving task, in a way similar to human capabilities – are set to bring many benefits. These include a reduction in accidents, increased fuel efficiency, and increased mobility for the elderly and the disabled, as well as the associated monetary benefits. However uncertainty as to who will be liable should accidents arise will prevent these benefits from coming to fruition due to the inability of developers to internalise potential risks.
Having examined the applicability of the common law of negligence, and the Consumer Protection Act 1987, I concluded that the legal regime as it stands is unable to determine whether, or when, developers will be liable. Although some difficulty arises in establishing whether a duty of care would be held to exist, the more significant problems with negligence occur in finding where the standard of care lies and in proving causation. Whilst these problems are not unique to the application of negligence to AVs, it does indicate the problem of uncertainty.
The application of the product liability regime under the CPA is even less certain. In the first instance, it has long been debated whether software systems can even be considered a product for the purposes of the Act. Additionally, the damage arising as a result of some accidents is unlikely to fall within the regime, with damage to commercial property and the vehicle itself excluded. The most problematic part, however, would be in establishing whether there has been a defect for the purposes of s3(1) which requires “the safety of the product is not such as persons generally are entitled to expect …” The dissertation first evaluated issues that may be common to any claim, before taking a closer examination of three scenarios specific to autonomous vehicles. Common issues relate to the marketing of the product and the applicability of the developmental risk defence under s 7(e), whilst in specific scenarios there is the repeated question as to when consumers can no longer expect potential but costly and unrealistic risk-minimising measures to be put in place.
The latter half of my dissertation then looked at whether and how the uncertainty of liability is being resolved. Actions are currently being undertaken by the government, for example the Automatic and Electronic Vehicles Act 2018 (AVEA) and the proposed regulatory changes. The Act only covers vehicles that are roughly equivalent to Levels 4 and 5 on the NHTSA scale, and appears to exclude Level 3 vehicles that have high levels of automation. This is problematic as this is the Level of vehicle most likely to give rise to the complicated questions of whether machine or man ought to be held responsible. This is a clear failure of the Act. There are also issues around insurance regimes before the scheme is brought into effect.
Having analysed the viability and potential effect of the proposed regulatory regime, I concluded that the rolling programme has the potential to be beneficial prima facie, but must be created in partnership with developers. There are other methods that may be beneficial in managing the liability of developers. Proposed solutions include a new liability scheme which seeks to address the failures of both negligence and the CPA, or a joint-fund insurance scheme, modelled after the Price-Anderson Act. It is also possible that establishing legal personality, thus removing the developer from the equation, could provide better redress.