Harness AI before criminals outpace you, criminologist tells cops

PETALING JAYA: Criminologist P Sundramoorthy has urged Malaysian police to adopt artificial intelligence (AI) to keep pace with evolving crime tactics. Concerned criminals are now utilizing digital platforms, encrypted communication, and tools like deepfakes, facilitated by broader AI accessibility. Sundramoorthy, from Universiti Sains Malaysia, warned criminals could soon surpass law enforcement if they don't adapt.

While this presents a significant challenge, AI also offers immense potential for the police. It can help identify patterns, connect data, predict threats, investigate intricate cases, and allocate resources more intelligently. To leverage these benefits, Sundramoorthy proposed establishing a national AI command centre in the police force.

This center would integrate AI-assisted crime intelligence, crime mapping, cybercrime monitoring, and organized crime network analysis to support real-time decision-making. The existing police information, including report files, investigation databases, CCTV records, traffic data, and financial crime investigations, could be pooled and linked via AI to reveal crime patterns, geographical concentrations, and connections between seemingly unrelated incidents.

However, Sundramoorthy emphasized that AI should enhance, not replace, human judgement. It can help investigators sort, categorize, and correlate vast amounts of data, allowing police more time for crucial tasks like interviewing witnesses, assessing credibility, establishing intent, corroborating evidence, and building cases. Nonetheless, AI cannot substitute for police experience, investigative skills, community cooperation, and human judgment.

The goal should be human-centered policing strengthened by AI, not AI-controlled policing. The future police force will be one that can transform information into intelligence, intelligence into prevention, and prevention into public safety. Officers must be trained in AI usage for policing, and protocols for AI-generated information should be developed to ensure it serves as leads, not direct evidence.