Challenging a ‘Hurt First, Fix Later’ Algorithmic System: Is the Tort of Negligence a Regulatory Solution?

Abstract

The use of algorithmic systems to identify suspected welfare fraud in the Netherlands, India, Australia, and the UK has led to mounting concerns that governments are taking a ‘hurt first, fix later’ approach to the adoption of algorithmic systems that will impact many of the poorest and most vulnerable members of society. The question this article addresses is one of effective regulation. While various strategies have been explored to tackle challenges arising from the adoption of algorithmic systems in welfare fraud investigations in recent years, this article follows the approach adopted by the applicants in the Robodebt class action in Australia in proposing the tort of negligence as a source of common law regulation in the era of algorithmic systems. First, it explains why a duty of care offers a principled and practical answer to the challenges posed by the misuse of algorithmic tools in welfare systems. Second, it argues that, drawing on the reasoning in two well-known English cases of the 1970s, Home Office v Dorset Yacht Co Ltd and Dutton v Bognor Regis, the common law of negligence can provide a strong foundation for recognising such a duty. Finally, it considers two significant challenges to the proposal and argues that, despite these challenges, the tort of negligence offers a valuable opportunity to enhance fairness, legitimacy, and equity in both system design and regulatory practice, while also mitigating litigation risks.