Facial recognition technology has significant implications for legislators and policymakers, criminal justice officials, and society in general, and further developments should limit the potential risks in a way that maximizes the benefits of such technology. This report examines current and near-future uses of facial recognition by law enforcement and identifies the concerns and related policy domains that could be used to guide public policy.
- What differences are there across uses and policy needs, and how should those needs vary by use?
- What is the accuracy of the technology across uses, and how should accuracy shape use?
- What are the complex organizational and procedural issues, such as the ownership of FR systems and data?
Technological advances in society often have implications for policing. Much as fingerprinting and DNA evidence were groundbreaking advances that have been used to solve serious crimes, remote biometric identification, such as facial recognition (FR), is rapidly developing and being used in policing. This has significant implications for legislators and policymakers, criminal justice officials, and society in general that should be understood, and further developments should limit the potential risks while maximizing the benefits of such technology. This report examines current and near-future uses of FR by law enforcement and identifies the various concerns and related policy domains that could be used to guide public policy.
The authors examine the various dimensions along which use of FR for law enforcement purposes can be considered, summarize the variety of policies and views about FR acceptability, and review a wide range of relevant considerations when making these decisions about FR use and policy options that fall between the extremes of permissive use and banning law enforcement use of the technology altogether. To inform this effort, the authors conducted a literature and document review, interviews with subject-matter experts, and a nationally representative survey of public perceptions of FR use by law enforcement. The authors also present a road map to help policymakers navigate the concerns about FR. The authors explain the different components of a policy and management structure for FR and how requirements for transparency and oversight that are responsive to different levels of trust in police agencies may be addressed.
- Federal legislation and many state legislatures still lag behind FR use in setting regulations for law enforcement.
- Combining FR with pervasive surveillance is possible with current technology and represents the most concerning application of this technology because it is overly broad and pervasive.
- Ways to regulate FR use by law enforcement include narrowing the potential set of uses of FR to maximize benefits and minimize risks, along with careful review to ensure that regulations were upheld. This narrowing could consist of limiting FR to investigating serious crimes, increasing the level of authorization required, reducing the number of people being searched or doing the searching, restricting the types of images or the quality of images that can be used, setting a threshold for search results, and deleting data.
- Having a requirement to seek warrants from an outside source is a policy option some states are using that acts as both oversight and a practical check on the power of law enforcement to collect and use information.
- Transparency about when FR is used is also a critical check. This includes public awareness of preapproved-use cases and FR use in individual cases (e.g., notification to defendants). If the system is flawed or used inappropriately, full disclosure can potentially identify issues and reduce or prevent the high cost of an innocent person being arrested and convicted.
- Regulations and policies need to be specified clearly and in relation to actual use so that concerns are adequately addressed.
- Misuse of FR technology by law enforcement or system failures (e.g., convicting an innocent person) is highly detrimental to public trust of law enforcement, and potential risks should be minimized as much as possible.
- In-domain or field-accuracy testing and human-factors testing needs to be done.
- To weigh concerns and potential benefits, descriptive information about FR use by law enforcement is needed, which might examine the whole chain of use, starting from image acquisition.
- Then, implementation and outcome evaluations are needed to support or reject the use of FR as an improvement over and above standard practice.
- A more fully developed set of standards is needed to avoid unintended variability in how law enforcement agencies use FR.
Table of Contents
Introduction: The Use of Facial Recognition by Law Enforcement
Developing a Spectrum of Use Cases as a Step Toward a Broadly Practical Approach for Regulation
Understanding the Landscape of Concerns
Developing a Policy Road Map to Help Navigate the Landscape of Concerns About Facial Recognition