Ethical review to support Responsible Artificial Intelligence (AI) in policing: A preliminary study of West Midlands Police's specialist data ethics review committee

The West Midlands Police and Crime Commissioner (WMOPCC) and the West Midlands Police (WMP) have for the past five years maintained a Data Ethics Committee to advise on the design, development and deployment of advanced data analytics and AI capabilities. This Committee comprises people drawn from backgrounds in academia, industry, public/third sector and policing. Since 2019, it has met at least on a quarterly basis, advising and making recommendations on each occasion on several projects and proposed tools, from in-principle analysis to tools ready for operational use. Its papers and minutes are published via WMOPCC. This interdisciplinary research used mixed-methods (including 26 interviews) to review the impact and influence of the Committee, and to recommend to national bodies, other forces and to WMP/WMOPCC factors that affect how best to go about using independent advisors in this context. Lessons from the Committee’s experience, together with a single structured framework could inform a coherent and consistent national approach. The Conclusions and Recommendations (for national strategy, police, Committee members, community representatives, academia and research funding bodies) fall into the following themes: - A Data Ethics Committee with diverse independent voices can contribute positively to the validity and responsibility of policing AI, thus supporting operational policing. It can develop understanding within the police of key ethical, scientific, legal and operational issues for planning and implementation. This will be successful only if the Committee has a clear function, is fully incorporated into the system of oversight and scrutiny, visibly championed by the Chief Constable & PCC, and suitably supported by a secretariat, robust process and communications; - This will be successful only if membership includes genuine representation from the community that the police serves, there is transparent engagement, and time taken to allow members to understand the technical and legal aspects of the work. - This will be successful only if the operational context is explained by operational police officers, and time taken to understand how AI outputs will be used, so as to enable potential benefits, risks/harms and proportionality to be assessed in the same conversation. Attention must be paid to police responsibilities for public safety (and how AI may support these responsibilities) as well as to risks related to privacy, fair trial and freedom of expression. - Police forces, PCCs and national bodies embarking on such an approach will need to be prepared for ambiguity. There are often no ‘black and white’ answers to ethical, legal or technical questions raised by policing AI, such as reconciling privacy and security priorities relevant to the assessment of the proportionality of using suspect data.