India - facial recognition used against protesters

What Happened

During major national protests (such as the farmers' protests and anti-CAA demonstrations) in India between 2020 and 2021, state police forces deployed advanced, AI-driven facial recognition systems (AFRS) to monitor public gatherings. The system analyzed live video feeds from police drones and public CCTV cameras to identify, track, and log the biometrics of peaceful protesters. This biometric database was subsequently used to target, arrest, and harass political activists without any judicial warrants or clear data protection boundaries, raising severe concerns about the weaponization of biometric surveillance against civil liberties and the right to peaceful assembly.

How MEG 1 would have acted

The facial recognition system failed both the Dynamic Accuracy Index (DAI) and the Index of Safety and Responsibility (ISR) audits due to the lack of transparent legal basis and the high risk of political weaponization, violating public safety and cognitive liberty.

MEG 2 Analysis - Legal Framework

Agent level: N3 (individualized/advanced) - mass surveillance and biometric tracking system; high individuation and tracking capacity of specific citizens.
Cause (6.1):
(a) system defect - design of a biometric tracking system that lacks independent oversight and exhibits potential demographic bias in public spaces (design defect).
(b) autonomous decision error - the system's decision to flag specific citizens based on opaque, algorithmic risk scores.
Attachment of liability: The public operator is fully liable for the privacy violations and the unauthorized tracking; the developers share liability if they built unvalidated models (6.1a).

Procedural Mechanism

6.2 (architectural human confirmation): Biometric identification and tracking of specific citizens in public spaces must require strict, independent judicial confirmation before execution.
7.1 (independent forensics): All access logs to the biometric database must be recorded in an unalterable, independent ledger to prevent insider abuse and unauthorized searches.
9.4 (least privilege): The system's unrestricted access to the real-time biometric data of millions of citizens violates the principle of least necessary access.