Working paper: Enabling AI in law enforcement: Making a case for the adoption of a human rights principles-based approach

The working paper proposes adopting a high-level principles-based AI risk assessment premised on a human-rights-centric approach to govern the design, development, and deployment of AI tools by law enforcement agencies. The aim is to maintain the security of the nation and the safety of individuals while also ensuring responsible legal governance, balancing the need for national security with the protection of human rights.

While various approaches to national security can stem from differences in legal cultural norms or reflect a diversified strategy of approach, fundamental traditional human rights and their evolution in the age of AI will act as overarching principles, a metric of measuring the realisation of these principles when law enforcement agencies utilise AI systems. These principles are distilled from traditional and emerging human rights frameworks, prevailing AI ethical frameworks, and model AI law enactments.

As a backgrounder, the paper lays its foundations by setting out some fundamental understanding of the technology and most importantly, the different approaches to governing AI.

The paper proceeds to outline the various and prevailing uses of AI systems in law enforcement and policing, highlighting both the benefits and the risks, harms, and vulnerabilities (such as bias in AI algorithms when used in predictive policing, intrusions into individuals’ privacy, invasive use of emotional AI, and potential misuse of AI for surveillance) that erode individuals' fundamental human rights when designing and developing these systems and when deploying them.

Thirdly, the paper scopes various sources: the corpus of international traditional human rights frameworks and the emerging and evolving “new rights” found in established international AI ethical frameworks and AI legislation. This holistic approach distils a set of rights relevant to the design, development and deployment of AI tools in the context of national security and law enforcement, recommending these rights to be adopted as the “principles” underlying the proposed approach.

Against this background, the paper traces how risks of using AI tools engage with the abovementioned rights, mapping these risks to the principles and their potential erosion.
The paper posits that these high-level principles allow designers, developers, and deployers of these AI tools to build their capacity to ensure that safeguards and guardrails are in place at the various stages of the lifecycle of AI tools. The paper recommends that when operationalised and realised in a risk assessment and assurance tool, the principles-based approach serves as an effective Human Rights Impact Assessment (HRIA).