Are we there yet? Criminal liability in the era of driverless cars
Since the Motor Car Act 1903, the safety of our roads has been regulated largely through the criminal law. Drivers will incur criminal liability if they drive in an unsafe way, or if they evade the scheme of regulation and compensation (i.e. by driving without a licence or insurance).
What will happen, then, when drivers are taken out of the picture? As the technology advances, who should be responsible if a driverless car speeds? Or drives in a way that, had it been the result of a human driver, would be considered careless or dangerous? Or if a driverless car kills someone?
The advent of automated vehicles has been progressing at a pace that allows lawmakers to address these questions in anticipation rather than reaction. Yet as last year saw the first death involving a driverless car in Arizona, the time for reflection is running out.
The Automated and Electric Vehicles Act 2018, which has not yet been brought into force, envisages how civil liability will operate for automated vehicles. It includes a provision that, where an accident occurs as a direct result of software alterations prohibited by an insurance policy, or a failure to install software updates, the insured person is liable for any damage caused.
One might expect criminal liability to follow in a similar vein. The Law Commission’s on-going consultation (closing on 8 February 2019) envisages that, where a vehicle is driving itself, the human user will not be liable for any offences arising out of the automated driving. Instead, the entity responsible for the automated system will be subject to a series of regulatory sanctions. However, if the owner of the vehicle has failed to update software then the fault may lie with them. It is also suggested that, at least in the immediate term, every vehicle will still need a “user-in-charge” who is capable of taking over driving when necessary, bears the various obligations not related to the driving itself, such as reporting accidents, and is subject to the same rules on licencing and sobriety that would apply if they were driving the car.
The assignment of fault between the company that designed the system, the registered keeper of the car and the user-in-charge raises three issues of which legislators and practitioners will need to be mindful.
The first consideration concerns the practicalities of prosecution. If there is a question mark over the cause of a minor traffic offence – such as speeding or going through a red traffic light – then what ought to be a straightforward and relatively efficient prosecution may become increasingly complex and technical. It may need to be determined a) whether the car was in autonomous mode; b) whether the software was responsible for the speeding; and c) whether the owner ought to have updated the software and whether that update would have prevented the speeding. There may even be other issues raised by potential defendants, such as whether the software has been hacked, whether the roads or signs have been tampered with in a manner that tricks the software, or whether the user-in-charge ought to have realised the system was not functioning correctly and taken over. Aside from the changes to legislation, the courts and its users will all need to be ready for what could be an incredibly complex development. The need for proper case management in driving cases, as was highlighted in 2016 by Senior District Judge Riddle in R v Cipriani, could be tested to the extreme.
Moreover, for more serious offences, the Law Commission has raised the question of whether new corporate offences ought to be introduced where faults in an automated system result in death or personal injury. There is some concern that the current law on corporate manslaughter would not be fit for purpose. Moreover, will a new or any existing offence accommodate for those situations where, faced with a choice between two evils, an algorithm has been written to choose one injury or fatality over another? Can a jury be expected to decide if the choice to write the software in a certain way constituted manslaughter?
The second issue is effective deterrence. A human driver is heavily dissuaded from driving offences by the sanction of the points system, disqualification and further punishment. A financial penalty for a company whose software was faulty may not carry the same weight, and in practice the courts may not be all too willing to withdraw authorisation for that software to be used, if that would mean that consumers could no longer use an automated system on which they rely. There may need to be some thought on how best to compel companies to keep their software up-to-date and compliant with the ever-changing regulations governing all the country’s roads. The National Transport Commission of Australia has tentatively raised the possibility of imprisonment terms being available for ‘duty holders’ within the entities responsible for an automated system (see paragraph 8.2.2 of their policy paper), but that may not be realistic.
Developers may think twice about investing in this technology if their liability and legal expenses are going to be extortionate.
Finally, there is the need to encourage progress. If driverless vehicles will be, on the whole, safer than human-driven cars, then parliament will not want to introduce laws that prevent use and development of the technology. So, in ensuring that companies and users are sufficiently deterred from letting automated vehicles drive dangerously, Parliament will also need to strike a balance with the need to avoid excessive regulation likely to stifle progress. If a user-in-charge may obtain a criminal record by sitting in a driverless car whose software has not been updated, or if they are expected to pay close attention and take over if the software starts malfunctioning at all, they may conclude that it would be better to stick with human driving. Or similarly, developers may think twice about investing in this technology, or making it affordable, if their liability and legal expenses are going to be extortionate.
As the law tries to develop in step with the technology, it is an unenviable task to walk the line between effective protection of road-users in the short-term and the encouragement of a potentially life-saving innovation in the long-term. Whatever becomes of the Law Commission’s current project, there are likely to be many challenges ahead for criminal practitioners in road traffic law.