1

Closed-circuit television (CCTV) cameras equipped with artificial intelligence (AI) are tracing the movement paths of suspect vehicles, bringing shifts to police tracking investigations. The system filters through massive amounts of video footage to pick out targets matching specific license plate numbers and physical appearances, relaying location coordinates to officers on the scene. As local governments nationwide broaden the adoption of AI control systems, expanding operational foundations, close coordination between police and control centers to translate analytical findings into swift arrests is becoming increasingly critical.

According to the Ministry of the Interior and Safety alongside local governments on the 13th, the number of CCTV cameras monitored across 217 local governments nationwide stands at approximately 650,000, with AI technology deployed across 37.5% of them.

In Dongducheon, Gyeonggi Province, AI video analytics contributed to the apprehension of a vehicle theft suspect. On April 11 last year, after receiving a report of a stolen car in Saengyeon-dong, the Dongducheon Police Station requested cooperation from the city's integrated CCTV control center. The control center inputted the vehicle's license plate and physical details into an AI analysis system to search for matching vehicles and trace the getaway route. Monitoring personnel confirmed search results and relayed the car's location and direction of travel to field officers. Police spotted the suspect vehicle near Songnae-dong and arrested the suspect following a pursuit, coming roughly 30 minutes after the initial report. This case illustrates how AI video analytics led to an arrest through verification by control personnel and tracking by the police.

CCTV, once utilized merely to secure crime scene footage, has evolved into a tool for actively tracking fleeing suspects. Incorporating AI searches into the process—where officers previously had to comb through footage manually—allowed authorities to pinpoint where the vehicle headed. The role of surveillance is broadening beyond finding evidence in recorded video to utilizing analyzed location intelligence for active field tracking.

The Seoul Metropolitan Government is also bolstering its video analysis infrastructure to support police searches and tracking operations. The city plans to invest 27.1 billion won this year to upgrade 8,536 CCTV units through new installations, replacements of aging equipment, and transitions to intelligent systems. It also intends to pilot generative AI surveillance that moves beyond identifying people and objects to analyzing the contextual circumstances surrounding hazardous situations.

Translating these systems into public safety results, however, requires seamless coordination between police and control centers. CCTV operations are primarily managed by local governments, while on-the-ground searches and dispatches are handled by police. When a control center confirms and relays the location and direction of a suspect vehicle, police can launch pursuits based on that intelligence. Alongside equipment deployment, standards and protocols for relaying analytical data to officers in the field must also be refined.

Ultimately, human verification of AI analysis remains essential. Because systems may misidentify similar vehicles or mistake routine actions for dangerous situations, a process of reviewing video footage alongside on-site conditions is indispensable.

Lee Yoon-ho, professor of police administration at Dongguk University, remarked, "AI offers the advantage of analyzing vast amounts of information faster than humans and providing data necessary for decision-making." He added, "However, AI must not become the primary decision-maker. AI should strictly serve as an auxiliary tool to assist police judgment, and the final assessment and decision must be made by the officer in charge."

                                                                                                            Seol So-young