Public debate has largely focused on Flock cameras, gunshot detection systems and police drones, yet those technologies are merely the front edge of a much broader shift. Artificial intelligence is rapidly becoming embedded in nearly every stage of modern policing, from crime analysis and dispatching to video review, facial-recognition searches, real-time crime centers, and report generation. The fundamental issue is no longer whether police should use a specific surveillance tool. It is whether society is prepared to allow technology to play an increasing role in interpreting information and influencing decisions that have traditionally been made by people.
Cleveland is already moving in this direction. The city reports that nearly 2,000 public and privately owned cameras are connected to its Real-Time Crime Center, where civilian analysts and police personnel support officers with real-time information. Cleveland also uses artificial intelligence to monitor camera-system performance and identify equipment failures. As the technology expands, the discussion is no longer limited to cameras. The National Institute of Justice notes that real-time crime centers are designed to integrate information, technology, and analysis into a centralized operational hub that supports emergency response and criminal investigations. The question for policymakers and the public is not whether technology will play a role in policing, but how much authority and influence society is willing to grant to it.
The scale of digital evidence being created in Cleveland is already substantial. In 2023, Cleveland police uploaded approximately 594,000 body-camera videos totaling about 128,000 hours of footage. Although those recordings do not represent 594,000 separate public interactions—because several officers may record the same event—the volume illustrates why artificial intelligence will increasingly be considered for organizing, searching and reviewing police evidence that cannot realistically be examined manually. Signal Cleveland
The public-safety benefits can be substantial. Analytics can help investigators examine volumes of records and digital evidence that no individual analyst could review quickly. In departments facing staffing shortages, technology may also return officers to patrol and allow investigators to concentrate on work that requires experience, judgment and human interaction.
Artificial intelligence is now entering one of the most consequential parts of policing: the official police report. Axon’s Draft One uses body-camera audio and officer-provided information to generate a preliminary report narrative. According to Axon’s description of the system, the officer must review, edit and approve the narrative, and the drafting activity is recorded in an audit trail.
That may reduce hours of paperwork, but it also raises questions that cannot be dismissed as technical details. A police report may influence an arrest, charging decision, plea negotiation or trial.
Agencies will need to determine whether the original transcript and AI-generated draft must be preserved, how errors or omitted context will be identified, what information must be disclosed to prosecutors and defense attorneys, and whether officers may become too dependent on a polished narrative that they did not personally compose.
The same concerns apply to predictive analytics and facial recognition. Artificial intelligence can identify patterns and possible connections, but its conclusions depend on the quality and completeness of the underlying data. Police records reflect reported crime, enforcement activity, sensor placement and human decisions made over many years. An algorithm may process those records efficiently without understanding why the data looks the way it does. The National Institute of Standards and Technology has documented that facial-recognition error rates can vary by age, sex, race, image quality and algorithm—demonstrating why a computer-generated match must remain an investigative lead rather than proof of identity.
Public acceptance will probably depend on what the technology does and how closely it is connected to government authority. People may readily accept software that finds relevant video, identifies a malfunctioning camera or transcribes recorded evidence. They may be far less comfortable with software that predicts who may commit a crime, recommends whom officers should stop, continuously identifies people in public spaces or produces conclusions that cannot be independently examined.
There are consequences on both sides of the decision. Implementing artificial intelligence without adequate rules could magnify errors, reinforce existing disparities, expose sensitive information, create dependence on private vendors and damage public confidence. A 2024 Government Accountability Office review found that Department of Homeland Security law-enforcement agencies used more than 20 types of monitoring technologies, while their policies did not always address important privacy protections.
Inaction also carries consequences. Police agencies already possess more video, digital evidence, dispatch information and investigative data than their personnel can realistically review without technological assistance. Refusing to address artificial intelligence will not necessarily prevent its use; it may instead encourage informal or unauthorized use by individual employees on personal devices, without security protections, audit trails or established standards.
The responsible course is neither automatic acceptance nor blanket rejection. Police agencies should identify every AI-enabled system they use, establish permissible and prohibited applications, require human verification, preserve audit records, protect confidential information, test for accuracy and bias, disclose appropriate uses to the public and measure whether the technology actually improves public safety. Ohio’s statewide IT-17 artificial-intelligence policy, effective January 5, 2026, establishes planning, procurement, security, privacy and governance requirements for state agencies. Local law-enforcement policies will need to address the additional consequences that arise when AI influences investigations, official reports and decisions affecting individual liberty.
The International Association of Chiefs of Police and the National Policing Institute are now urging departments to develop AI policies, designate responsible officials, involve legal counsel and establish oversight before these systems become deeply embedded in police operations. That advice recognizes an important reality: the rules governing artificial intelligence should not be written only after a serious error, lawsuit or loss of public confidence.
Where are we going?
Policing is moving toward an integrated environment in which artificial intelligence will search video, connect information across databases, identify patterns, prepare reports and help determine what officers and investigators examine first. The critical change is not simply that police will use more technology, but that software will increasingly organize and interpret the information on which official decisions are based.
Law-enforcement agencies must establish the rules before these systems become indispensable. To maintain credibility, agencies should define permissible uses, independently test accuracy, preserve source material and audit records, require meaningful human verification, disclose appropriate uses to the public and measure whether each system produces demonstrable public-safety benefits. Artificial intelligence may improve policing, officer safety and criminal investigations, but accountability must remain with the agency and the individual officials who act on its conclusions. Technology may assist professional judgment; it must never substitute for human responsibility.
Leave a Reply