Top News

The debate over introducing AI-based surveillance in prisons, especially in contexts such as Karnataka’s ongoing prison modernization efforts, highlights the tension between increased security and ethical risks.

AI systems can improve detection of smuggling of contraband (such as mobile phones or drugs thrown over walls), prisoner violence, threats, or gang formation. Advanced video analytics, including facial recognition and anomaly detection, allow real-time alerts in facilities where human guards cannot constantly monitor every area. For example, Karnataka’s Parappana Agrahara Central Jail has issued a tender for an AI-powered audio-video system to flag restricted access, unauthorized phones, or violations, addressing real problems like illegal activities among understaffed people.

However, the limitations of AI raise concerns. Algorithms detect patterns but lack the contextual understanding of human emotions – discomfort due to grief may be falsely flagged as suspicious, leading to inappropriate intervention. Bias has been documented in training data in US systems, where tools such as risk assessment or speech recognition disproportionately misidentify or penalize racial minorities (e.g. higher error rates for African American dialects in phone monitoring). While no specific British case matches claims of AI suicide prevention failing before multiple deaths, widespread issues remain in prison mental health support.

India’s prisons are severely overcrowded, with the national average occupancy **131%** (as per India Justice Report 2025), many facilities overcrowded at over 200%, and a large number of undertrials living in deplorable conditions without basic amenities or mental health care. Funds for expensive AI infrastructure could alternatively be used to support counselors, vocational training, or rehabilitation – investments that address the root causes such as delayed lawsuits that fuel the need for contraband.

Normalizing intense AI surveillance in prisons, where rights are limited, creates the risk of “surveillance expansion” into schools or public spaces, as seen in global trends as AI tools spread from corrections to education. Prisons should prioritize humane rehabilitation rather than permanent surveillance to promote reintegration.

Ultimately, AI provides the tools for safer prisons but it requires rigorous oversight, bias audits, and balanced priorities focused on dignity and reform.