Conference Agenda

As online surveillance tools become increasingly integrated into public safety strategies, the role of police in digital spaces raises pressing questions about legitimacy, authority, and public trust. While legal frameworks for surveillance are often discussed, less is known about how citizens perceive and evaluate these practices—particularly when surveillance occurs in spaces they consider private, such as closed social media groups or chat-apps. This paper investigates the conditions under which citizens accept or reject police online monitoring, and explores the conceptual boundaries between public and private digital domains.

The study draws on a mixed-method design combining two surveys (N=261), 20 semi-structured interviews with civilians, and exploratory conversations with police professionals. In addition, citizen perspectives expressed on online forums and within political debates on privacy and surveillance are analysed. This multi-source approach allows for a layered understanding of public sentiment and the tensions between public safety and individual rights.

Preliminary findings show that citizen acceptance is shaped by transparency, perceived necessity, the identity of the surveilling authority, and the purpose of the monitoring (e.g., targeted prevention vs. broad oversight). Acceptance increases when monitoring is seen as essential for public safety, but concerns around misuse of data, loss of autonomy, and ambiguity about digital “publicness” remain strong. Notably, many respondents draw a clear distinction between open platforms and semi-private or private online spaces, where surveillance is more often perceived as intrusive, even when conducted for preventive purposes.

The paper also engages with the conceptual ambiguity surrounding the term “private” in the context of digital platforms. Citizens’ expectations of privacy in closed groups do not always align with legal or institutional definitions, creating friction between perceived and formal legitimacy.

These findings contribute to broader debates on policing and digital governance, highlighting the need for transparency, clear purposes, and citizen rights to be balanced in developing socially legitimate policies for online police surveillance. By situating citizen perceptions within the context of public order governance, this study offers practical and theoretical insights into how police and public institutions can navigate the blurred boundaries of digital authority in a democratic society.

Introduction

The current UK government is placing considerable emphasis on the use of Artificial Intelligence (AI) to improve productivity in the public sector, including policing, with an AI strategy (OAI, 2021). The National Audit Office (NAO, 2024) also stated some policy aims including:

1. The UK public sector will be world-leading in safe, responsible and transparent use of AI to improve public services and outcomes.

2. The public will benefit from services that have been transformed by AI.

3. Public and civil servants will have the tools, information and skills they need to use AI.

4. All public organisations will be more efficient and productive through AI adoption and have the foundations in place to innovate with the next wave of technologies

These policies have generated considerable interest in applications of Artificial Intelligence (AI) to policing processes as a potential source of productivity improvement. Some existing results have been mixed (Stanley, 2024). They point towards AI hallucinations and some loss of discipline in policing.

This paper reports on the findings of a case -based study that reviewed a prototype AI application to write witness reports. The application converts the audio of witness interviews into a series of required documents including the “MG11” witness statement.

Research Questions

This paper focuses on the risks and ethics raised during the study. The research questions we ask here are:

1. What are the risks, if any, of the use of AI in policing applications to replace some of the functions of police officers in tasks such as witness statement authorship?

2. Are there any ethical issues raised using AI to assist witnesses and officers to produce these reports?

Methodology

A mixed-methods approach was used in the study:

1. Semi-structured interviews were held with users, police stakeholders and developers of the AI to obtain an understanding of their perceptions of the utility of the application.

2. The quality of a sample of 26 reports (AI or human written) were assessed using established methods of understanding report readability and linguistic style (Friginal and Biber, 2016).

Findings

Key findings to be debated are:

• The risks of the output concerns subtle errors, omissions or hallucinations that may be difficult to detect

• There is a loss of the voice of the witness when using AI

• The impact on output quality, with a move to standardisation of work rather than high quality output

• The impact on police roles when using AI

References

Friginal, E. and Biber, D. (2016) Multi-dimensional analysis. Eds. In: Baker, P., Egbert, J. (Eds.), Triangulating Methodological Approaches in Corpus Linguistic Research. Routledge, Abingdon, 73–89.NAO (2024), Use of artificial intelligence in government, National Audit Office 12 March pp56.OAI (2021), National AI Strategy, UK Office for Artificial Intelligence, September 2021Stanley, J. (2024) AI Generated Police Reports Raise Concerns Around Transparency, Bias, ACLU Report, December, AI Generated Police Reports Raise Concerns Around Transparency, Bias | ACLU