When Criminals Use AI, Can Police Training Keep Up?

Burhan Rasool’s Technology in Policing offers a practitioner’s account of digital reform. As Pakistan connects its criminal justice institutions and confronts AI-enabled crime, the book makes a strong case for training that evolves as quickly as the threats.

A convincing voice message from a senior official, a video that never happened, or a synthetic identity assembled from stolen information can now become the starting point of a criminal investigation. Generative artificial intelligence has made deception cheaper and easier to scale. Police officers must increasingly establish not only what a digital record shows, but whether it is genuine.

That makes Technology in Policing: Innovations, Challenges, and the Road Ahead, by Burhan Rasool and published by the National Book Foundation in 2024, worth revisiting. Across eight modules, the book introduces police administration, crime prevention, investigation, prosecution, organisational change and emerging technologies. Its distinctive material is not the catalogue of tools, but the account of what happened when a government tried to make technology work inside a resistant institution.

What the police-station registers concealed

Rasool’s account of Punjab’s police computerisation programme begins with a revealing detour. Before deploying new software, he interviewed officers at different levels and identified 23 deficiencies involving infrastructure, operational logistics and investigative capacity. The implication was uncomfortable for technologists: software could not compensate for missing operational basics.

The programme initially digitised 25 registers in 85 Lahore police stations. A reproduced case study describes the discovery of approximately 41,000 discrepancies in 152,000 personnel records checked against NADRA data. Another passage reports that, in some stations, a single investigator carried at least 80 percent of the investigative workload. These are historical findings reported in the book, not evidence by themselves that policing outcomes improved.

The author is most persuasive when he acknowledges resistance, abandoned initiatives, redesigns prompted by poor user experience and technical debt accumulated under pressure to inaugurate systems quickly. His account treats reform as an organisational negotiation, not a ribbon-cutting exercise. But it is also a participant’s account. A fuller assessment would need independent evidence of lasting changes in investigation quality, public confidence and accountability.

Why integrated justice makes the book timely

Punjab’s Integrated Criminal Justice System (ICJS) provides a contemporary test of these lessons. Linking information held by police, prosecution, courts and prisons promises to reduce repeated requests and missing records. Yet connectivity cannot turn an incomplete investigation into a sound one, or make an inaccurate entry reliable merely because several institutions can now retrieve it.

The book’s treatment of digital evidence, chain of custody and data stewardship speaks directly to this problem. A police officer preparing a case for prosecution must be able to explain where evidence came from, who handled it and whether it changed. An integrated platform increases the importance of those safeguards because mistakes can travel further and faster.

A warning against automating the wrong thing

One of the sharpest passages appears in the final module. Through a fictional property dispute involving a forged authority letter and repeated court adjournments, Rasool illustrates how electronic filing could reproduce an existing vulnerability. He writes: “If we proceed with automating existing processes without fixing lacunae like these, we will create new problems instead of solving the old ones.”

That warning is relevant beyond criminal justice. The Pakistan Digital Authority, established under the Digital Nation Pakistan Act 2025, has outlined work on national data governance and secure exchange between government systems. Its emerging agenda raises similar questions of identity, institutional ownership, data quality and citizens’ rights. The book predates the Authority, and should not be mistaken for a blueprint for it. Rather, it offers a practical caution about scaling integration before fixing underlying processes.

The AI chapter now needs a sequel

Module 8 surveys artificial intelligence, predictive analytics, machine vision, robotics, digital authentication and digital forensics. It acknowledges human-rights, privacy and ethical concerns. But developments since publication make a stronger distinction essential: using AI to investigate crime is not the same as investigating crime committed with AI.

Police officers now need to recognise voice-cloning fraud, synthetic identities, manipulated video, AI-assisted social engineering and increasingly automated cybercrime. They also need to know how to preserve original files, document provenance, obtain platform records lawfully and avoid treating an AI-generated summary as verified evidence. A plausible video is not proof of authenticity; equally, a genuine recording must not be dismissed as a deepfake without examination.

A revised edition should also be more sceptical about technologies that claim to infer suspicious intent or criminal involvement from faces and emotions. These applications are not equivalent to established forensic techniques. Their reliability, error rates, biases and legal admissibility require explicit discussion rather than inclusion in a broad inventory of possible tools.

Practical exercises would make the material more durable: an impersonation complaint involving cloned audio; a disputed video whose provenance must be reconstructed; an AI-assisted fraud with multiple synthetic accounts; and an investigation in which an automated tool produces a convincing but false lead. Each exercise should require officers to distinguish investigative intelligence from admissible evidence.

A useful foundation, not the final curriculum

At 176 pages, the book’s reach exceeds the depth it can give every subject. Its historical survey and technology catalogue make it accessible to trainees, while the Punjab case study supplies the lived detail that distinguishes it from a generic introduction. The strongest argument is that adoption depends on leadership, field realities and trustworthy information, not software alone.

The next edition could make that argument more valuable by updating the threat landscape and placing evidentiary safeguards alongside every proposed AI capability. Short, regularly revised case studies would serve officers better than a static list of technologies whose capabilities and risks change from year to year.

Pakistan has made progress in digitising police records and is working to connect the institutions that handle criminal cases. The harder question is whether investigators are prepared for evidence that can be fabricated at scale and analytical tools that can be confidently wrong. Rasool’s book helps explain how police institutions learned to adopt technology. Its next task is to help officers question technology, including the technology used against them.

BOOK DETAILS

Technology in Policing: Innovations, Challenges, and the Road Ahead | Burhan Rasool | National Book Foundation | 2024 | 176 pages