Short abstract

The paper concerns AI weapons detection, a surveillance tool that UK police are trialling. We examine how the tool fits into existing legal frameworks, influences police discretion, and impacts the public. We argue that such tools may reconfigure how suspicion is formed and justified.

Long abstract

This paper introduces artificial intelligence (AI) weapons detection, a probabilistic AI tool currently being trialled in UK policing. The proponents of the technology assert that it is a solution to the rising knife crime in England and Wales and that it will lessen reliance on traditional stop and search practices, thereby enhancing public trust in policing. We argue that the process is reversing ‘stop and search’ to ‘search and stop'. Further, without appropriate governance, we claim that such systems risk reinforcing existing inequalities while complicating the transparency essential for lawful policing.

Traditional stop and search practices in the UK, both historically and in the contemporary world, have been criticised for their disproportionate impact on minority populations. The weapons detection systems are reliant on pattern recognition algorithms, which are known to inadvertently reproduce discrimination through technical bias. We also place AI weapons detection within the existing legal framework in the UK, including the legal test of ‘reasonable suspicion’, which requires both subjective belief and objective justification based on specific, articulable facts. As AI tools increasingly intersect with frontline judgement, we need to understand how their outputs might influence, support, or obscure this reasoning process.

Following the socio-legal analysis, the paper also draws on empirical data from interviews with police officers, legal professionals, and civil society participants to explore how reasonable suspicion is constructed, applied, and scrutinised in practice. The findings highlight tensions between frontline pragmatism, accountability mechanisms, technological optimism, and concerns about transparency and fairness.