Race and Privacy in Broadcast Police Communications (2407.01817v1)

Abstract: Radios are essential for the operations of modern police departments, and they function as both a collaborative communication technology and a sociotechnical system. However, little prior research has examined their usage or their connections to individual privacy and the role of race in policing, two growing topics of concern in the US. As a case study, we examine the Chicago Police Department's (CPD's) use of broadcast police communications (BPC) to coordinate the activity of law enforcement officers (LEOs) in the city. From a recently assembled archive of 80,775 hours of BPC associated with CPD operations, we analyze text transcripts of radio transmissions broadcast 9:00 AM to 5:00 PM on August 10th, 2018 in one majority Black, one majority white, and one majority Hispanic area of the city (24 hours of audio) to explore three research questions: (1) Do BPC reflect reported racial disparities in policing? (2) How and when is gender, race/ethnicity, and age mentioned in BPC? (3) To what extent do BPC include sensitive information, and who is put at most risk by this practice? (4) To what extent can LLMs heighten this risk? We explore the vocabulary and speech acts used by police in BPC, comparing mentions of personal characteristics to local demographics, the personal information shared over BPC, and the privacy concerns that it poses. Analysis indicates (a) policing professionals in the city of Chicago exhibit disproportionate attention to Black members of the public regardless of context, (b) sociodemographic characteristics like gender, race/ethnicity, and age are primarily mentioned in BPC about event information, and (c) disproportionate attention introduces disproportionate privacy risks for Black members of the public.

• The paper finds that Black males are disproportionately mentioned in Chicago police broadcasts, exposing systemic racial bias.
• The analysis reveals that 20% of communications contain PHI, with nearly 60% of sensitive disclosures involving Black individuals.
• The paper demonstrates that large language models like GPT-3.5 identify sensitive information with 66.5% accuracy, highlighting emerging AI privacy risks.

Race and Privacy in Broadcast Police Communications

The paper "Race and Privacy in Broadcast Police Communications," authored by Pranav Narayanan Venkit et al., explores the interplay of race and privacy within the framework of police radio communications. The paper leverages an unprecedented dataset consisting of 80,775 hours of broadcast police communication (BPC) from the Chicago Police Department (CPD) to analyze the nature and impact of these communications, specifically focusing on racial disparities and privacy vulnerabilities.

Objectives and Research Questions

The paper addresses four primary research questions:

1. Do BPC reflect reported racial disparities in policing?
2. How and when are gender, race/ethnicity, and age mentioned in BPC?
3. To what extent does BPC include sensitive information, and who is most at risk due to this practice?
4. Can LLMs exacerbate these privacy risks?

The aim is to provide a comprehensive understanding of the implications of BPC on racial attention disparities and privacy vulnerabilities, and to evaluate how emerging AI technologies can influence these dynamics.

Methodology

The paper employs a mixed-methods approach, combining lexical analysis, thematic qualitative coding, and empirical testing with LLMs. The researchers selected three zones within Chicago: Zone 4 (majority White), Zone 8 (majority Black), and Zone 13 (majority Hispanic). They analyzed BPC from these zones on August 10th, 2018, from 9:00 AM to 5:00 PM, focusing on verbatim transcripts of radio transmissions.

Lexical Analysis

The paper focuses on bigram frequency to explore the vocabulary used in BPC, given the dense and coded nature of police radio utterances. For example, the term "male black" frequently appeared in all zones, and male-gendered terms dominated the communications, highlighting a gender disparity. Furthermore, it was found that Black males were mentioned disproportionately, even in predominantly White areas, indicating a racial bias in police attention.

Thematic Qualitative Coding

The researchers classified the transcriptions into six categories based on their purpose:

• Event Information Transmissions
• Procedural Transmissions
• Liminal Transmissions
• Miscellaneous Policing Transmissions
• Casual Transmissions
• Unclear Intent Transmissions

This categorization provided insights into how different types of information are communicated and the contexts in which sensitive information is disclosed. Approximately 20% of the utterances contained PHI, with Event Information Transmissions being the most frequent type of communication involving PHI.

Empirical Testing with LLMs

The paper also explored the risk of privacy breaches by employing GPT-3.5, an off-the-shelf LLM, to identify PHI in BPC. The model achieved an accuracy of 66.5% in identifying PHI, showcasing its potential to automate the extraction of sensitive information, raising significant privacy concerns.

Key Findings

1. Racial Disparities in Attention: The analysis revealed a disproportionate attention toward Black males in police communications, consistent across different demographic zones. This finding underscores systemic racial biases inherent in law enforcement practices.
2. Inclusion of Sensitive Information: A significant portion (20%) of BPC contained PHI, increasing the risk of privacy breaches. Black and African American individuals were most vulnerable to privacy risks, with about 60% of PHI-related utterances involving them.
3. Impact of Emerging Technologies: The use of LLMs, such as GPT-3.5, demonstrated a high capability to extract sensitive information from BPC without extensive user training, highlighting the increased risk of privacy vulnerabilities with the advancement of AI technologies.

Implications and Future Work

The implications of this research are multifaceted. Practically, it emphasizes the need for stricter policies and practices to mitigate racial biases and protect sensitive information in police communications. Theoretically, it contributes to our understanding of how collaborative communication technologies can exacerbate existing social disparities.

Future research could extend the analysis to additional geographic contexts and longer temporal frames to generalize findings. Moreover, the development of AI safety mechanisms to filter and protect sensitive data in real-time communications is crucial. Addressing these concerns requires a concerted effort between technology developers, policymakers, and law enforcement agencies.

Conclusion

By exploring the intersection of race, privacy, and emerging technologies in the context of BPC, this paper provides critical insights into the systemic issues embedded in police communications. It calls for enhanced ethical standards and practices to safeguard civil liberties and address societal inequalities exacerbated by technological advancements in law enforcement.