Towards Understanding the Ethical and Operational Implications of Large Language Models in a Law Enforcement Environment

<p dir="ltr">Law enforcement agencies face a rapidly evolving digital landscape where vast and complex data sets require innovative analytical tools. In response, Artificial Intelligence (AI), particularly Large Language Models (LLMs), has emerged as a transformative solution, capable of analysing extensive information, identifying patterns, and informing decision-making. Despite offering substantial operational benefits—from drafting comprehensive reports to improving communication—LLMs present significant ethical, social, and legal challenges, including algorithmic bias and privacy risks.</p><p dir="ltr">This thesis examines the ethical and operational implications of deploying LLMs in a policing context. Through a mixed-methods approach—encompassing a broad literature review, a large-scale survey of law enforcement officers, and qualitative thematic analysis—it provides both empirical evidence and theoretical insights. Findings highlight that, although officers acknowledge potential efficiency gains and annual financial savings, they also express valid concerns regarding bias, data governance, security vulnerabilities, and community relations.</p><p dir="ltr">Strong training programs, robust regulatory frameworks, cross-sector collaboration, and active public engagement emerge as key elements. Ultimately, the thesis argues for a balanced approach grounded in ethical guidelines, transparency, officer education, and community engagement, to harness the transformative capabilities of LLMs without compromising fundamental rights. In doing so, it contributes to the emerging conversation on trustworthy AI, offering a practical roadmap for innovation that upholds justice, fairness, and accountability.</p>