Developing product liability networks for AI systems in the medical context

Product liability is again a matter of contemporary discussion due to the increased integration of technology into products. In particular, artificial intelligence (AI) has come to the fore, with machines, for example, using big data to make decisions faster and often with greater accuracy than humans. AI is being used at all stages of medicine. Such advances in technology have the potential to provide great benefits. However, there are potential risks. Many of these risks relate to data privacy but even in the field of safety uncertainty about the risks remain. This has led, in part, to reform of the European Union (EU) Products Liability Directive. The United Kingdom (UK) is, of course, no longer bound to follow, but it is likely it will also in time feel the need to do something in this space. In 2021 the Law Commission tentatively suggested ‘product liability and emerging technology’ for its programme of Law Reform, but concluded the time was not right. Subsequently, the Government has alluded to the need for modernising product liability law, but has only mentioned the possibility of extending the definition of product to include software and taking AI into account when assessing defectiveness. The context is a country hamstrung by Brexit, without a real vision for the future of UK consumer law nor for achieving the competitive advantages which Brexit promised to deliver. On the other hand, the EU has considered both a strict liability regime for high-risk AI products and a revised negligence regime for AI, as well as a revised strict product liability regime adapted to the AI context. This article will, first, consider how negligence might apply. Then, liability under the product liability regime, the EU’s amended regime and the proposed strict liability regime for high-risk AI will be considered. The advantages and disadvantages of negligence and strict liability will be discussed with a preference being for strict liability for all products. Finally, a plea to consider network liability or an even more ambitious insurance-based solution will be made.