Law
The Human Rights Implications of Predictive Policing Algorithms
Quick fact
A 2016 RAND study found that predictive policing systems can lead to disproportionate police stops in minority neighborhoods, even when designed to be race-neutral, creating a self-fulfilling cycle of over-policing.
Why this is interesting
What if a police officer says you're more likely to commit a crime—not because of anything you've done, but because of an algorithm's prediction? Is that even fair?
Read the full explanation
Understanding The Human Rights Implications of Predictive Policing Algorithms
Imagine a police department that wants to prevent crime before it happens. They feed years of crime reports, arrest records, and even social media data into a computer. The computer analyzes all that data to identify 'hot spots'—places where crime is likely to occur—and even predicts individuals who might commit crimes. The idea is to use resources efficiently and stop crime early. But here's the problem: the computer learns from the past, and the past includes human biases. If police have historically patrolled minority neighborhoods more heavily, they will have more arrests there. The algorithm sees that pattern and concludes that crime is more likely to happen in those neighborhoods, leading to even more police presence. That creates a feedback loop: more police means more arrests, more arrests mean more data, and the algorithm is reinforced. This is called 'algorithmic bias', and it can lead to discrimination against entire communities, violating their right to equal treatment under the law.
A deeper explanation
Predictive policing algorithms work by identifying patterns in historical data—crime records, location, time, and sometimes personal attributes. They use statistical models, like regression or machine learning, to predict future crime hotspots or individuals at risk. However, these models are not neutral; they encode the biases of the data they are trained on. For human rights, this matters because the right to non-discrimination (Article 2 of the UDHR) and the right to privacy (Article 12) are fundamental. When algorithms perpetuate racial profiling, they violate the right to equal treatment and the presumption of innocence. Moreover, because these systems are often opaque—'black boxes'—individuals have no way to know why they were targeted, undermining due process and the right to a fair hearing if they are accused. The use of predictive policing also has a chilling effect on freedom of assembly and association, as people may avoid certain areas or activities if they feel monitored. Human rights law demands that any limitation on these rights be proportionate, necessary, and subject to oversight. Yet predictive policing often operates without clear legal safeguards, transparency, or accountability. Therefore, while the intent may be public safety, the actual implementation can erode the very trust in the legal system that the police aim to protect.