ELECTRICITY theft, which continues to worsen the energy crisis and deepen the country’s circular debt, has now crossed unprecedented levels. Pakistan’s installed power generation capacity is over 46,000MW, while the actual con-sumption ranges from 8,000MW in winter to 30,000MW during peak summer.

This discrepancy is not due to lack of capacity, but because of the massive losses in the distribution system, particularly non-technical losses caused by theft and meter tampering.

Traditional enforcement methods have proven ineffective. The time has come to embrace artificial intelligence (AI) to detect and curb theft more efficiently. India and Brazil are among the countries that have successfully adopted such models.

According to a World Bank report, smart meters and AI-based consumption ana-lytics in Delhi cut losses from 53 per cent to under 8pc. In Brazil, machine learning and geospatial analysis have been able to detect electricity theft with over 90pc accuracy. South Africa has also been testing AI-based solutions to flag meter bypassing in informal settlements.

Smart meters, AI-powered anomaly detection, and monitoring based on geographic information system (GIS) are some of the tools that identify theft in real time, reduce the need for manual inspections, and recover billions in lost revenue.

The relevant authorities in Pakistan should prioritise AI integration in the distribution network as a long-term strategy to reform Pakistan’s power sector.

Muhammad Faraz Qureshi
Hyderabad

Published in Dawn, September 3rd, 2025