The Future of Policing

Project Trinetra in Maharashtra’s Akola district blends AI with ethics — marking India’s first serious, scientifically-grounded step toward predictive, preventive, and data-driven policing

- 

Beyond surveillance, Trinetra represents a shift in philosophy: using technology to empower human judgment, not replace it

Law enforcement in India faces a dual challenge: managing the rising complexity of crime while operating with limited resources.

Traditional reactive policing—responding only after a crime has occurred—cannot by itself ensure public safety, especially in environments dominated by repeat offenders, organised crime, and emerging urban vulnerabilities. Globally, jurisdictions from Los Angeles to London have experimented with predictive policing, using statistical models and data analytics to forecast where and by whom crime is likely to occur.

In India, however, predictive policing has remained largely an academic concept. Concerns about bias, transparency, and civil liberties have deterred its adoption. Against this backdrop, Project Trinetra in Akola district, Maharashtra, stands out as one of the first serious attempts to create a scientifically rigorous and ethically safeguarded framework for predictive policing in the country.

Trinetra—an acronym for Targeted Risk-based Insights for Next-crime Estimation & Tactical Resource Allocation—is a district-level innovation launched by the Akola Police under the leadership of Superintendent of Police Archit Chandak, with guidance from Kishan Panpalia, one of India’s foremost AI entrepreneurs.

At its core, Trinetra applies statistical modelling and machine learning to analyse repeat offenders. Instead of casting a wide net across entire populations, it narrows the focus to individuals with a history of serious or violent crime. Using a Repeat Offender Risk Scoring (RORS) model, each offender is assigned a dynamic probability score based on three weighted factors:

> Past Conviction & Crime Type (logistic weighting of severity, IPC sections, repeat 302s, etc.)