Law Enforcement Use of Artificial Intelligence and Directives in the 2023 Executive Order

The use of artificial intelligence (AI) has expanded in a variety of arenas, including by law enforcement. AI has been broadly conceptualized as computerized systems operating in ways often thought to require human intelligence. It is defined in the U.S. Code (15 U.S.C. §9401(3)) as

a machine-based system that can, for a given set of human-defined objectives, make predictions, recommendations or decisions influencing real or virtual environments. Artificial intelligence systems use machine and human-based inputs to-

(A) perceive real and virtual environments;

(B) abstract such perceptions into models through analysis in an automated manner; and

(C) use model inference to formulate options for information or action.

AI involves a host of technologies and applications. In the law enforcement realm, researchers note that while the use of AI is not yet widespread, existing tools may be enhanced with AI to expand law enforcement capabilities and increase their efficiency. Examples include the following:

• Automated license plate readers can be leveraged to employ machine, or computer, vision for capabilities such as automating the issuance of red-light violation tickets.
• Security cameras outfitted with certain AI-embedded hardware can be used for real-time facial recognition of potential suspects.
• Facial recognition technology and text analysis tools can be enhanced with AI to scan online advertisements to help identify potential crimes such as human trafficking.
• In addition to gunshot detection technology that can detect shots fired, security cameras can be outfitted with AI-enhanced software to detect weapons and alert police before shots are fired.
• AI redaction capabilities can be used to reduce possible bias in officers' narratives by removing certain identifying characteristics of suspects and victims—such as race—that could influence charges brought by prosecutors.
• Body-worn cameras can use AI software to redact or blur faces or sensitive footage before it is released to the public.
• Automated speech recognition software can use AI to help properly identify speakers' voices in audiovisual materials such as witness testimonies or interrogations.
• First responders use computer aided dispatch (CAD) systems to capture data that inform decisions on resource deployment; AI-enhanced CAD systems can improve resource allocation by using historical data, making predictions, and automating workflow.
• AI can be used along with predictive policing models to help identify individuals or places most at risk of being involved in crime.
• Law enforcement agencies can employ AI to enhance its communications with the public; they can use chatbots to respond to questions and push out emergency information.

A number of concerns have been raised about law enforcement use of AI, including whether its use perpetuates biases; one criticism is that the data on which the software are trained contain bias, thus training bias into the AI systems. Another concern is whether reliance on AI technology may lead police to ignore contradictory evidence. Policymakers may consider increased oversight over police use of AI systems to help evaluate and alleviate some of the shortcomings.

On October 30, 2023, President Biden issued Executive Order (E.O.) 14110 on the Safe, Secure, and Trustworthy Development and Use of Artificial Intelligence. This E.O. advances a government-wide approach to "governing the development and use of AI safely and responsibly" and directs efforts in AI policy areas involving safety and security, innovation and competition, worker support, equity and civil rights, individual protections, privacy protections, federal AI use, and international leadership.

E.O. 14110 acknowledges the risk of AI exacerbating discrimination and directs federal law enforcement in various ways. (In doing so, it references accountability focused directives for federal law enforcement previously outlined in the May 25, 2022, E.O. 14074 on Advancing Effective, Accountable Policing and Criminal Justice Practices to Enhance Public Trust and Public Safety.) Directives in E.O. 14110 include the following:

• The Attorney General (AG) shall coordinate and support enforcement of federal laws addressing discrimination and violations of civil rights and civil liberties related to AI. The Department of Justice's Civil Rights Division shall also coordinate with other federal civil rights offices to assess how their offices can prevent and address discrimination in automated systems—including algorithmic discrimination.
• The AG, with the Homeland Security Secretary and Office of Science and Technology Policy Director, shall submit a report to the President on the use of AI in the criminal justice system, including how AI can enhance law enforcement efficiency and accuracy, consistent with privacy, civil rights, and civil liberties protections. The report should also recommend best practices for law enforcement, including guidance on AI use, to address concerns outlined in E.O. 14074 with respect to law enforcement use of "facial recognition technology, other technologies using biometric information, and predictive algorithms, as well as data storage and access regarding such technologies."
• The interagency working group established by E.O. 14074 shall share best practices for recruiting law enforcement professionals with AI expertise and training them on responsible AI use. The AG, along with the Homeland Security Secretary, may review these and recommend best practices for state, local, tribal, and territorial law enforcement.
• The AG shall review the Justice Department's capacity to "investigate law enforcement deprivation of rights under color of law resulting from the use of AI," including through increasing or improving training for federal law enforcement officers and prosecutors.

Policymakers conducting oversight of executive branch activities to ensure that AI is used in a fair and equitable manner may examine not only these elements of E.O. 14110 that specifically relate to federal law enforcement but also other elements—such as the development of industry standards on AI—that may in turn affect law enforcement use of AI. They may also explore whether there should be specific standards for AI use in the criminal justice sector or AI-specific requirements for criminal justice entities receiving federal grants. Additionally, policymakers may continue to debate law enforcement use of specific AI technologies in its toolbox such as facial recognition technology.