How algorithms now police our streets and seas

In places like Indonesia, authorities now monitor thousands of fishing vessels by combining satellite imagery with automated location signals. If a ship deviates from its permitted path, a software system automatically flags it for review. Similar technology is hitting the streets in America, where tech companies are selling police departments digital platforms that combine 911 logs, license plate scans, and camera footage into a single dashboard. These systems are designed to give officers a preview of a scene before they arrive, theoretically reducing guesswork and reaction times. The companies selling these tools argue that they are simply helping overstretched agencies handle a data deluge that has become impossible to manage by hand.

These systems function by continuously analyzing patterns. Rather than waiting for a human to notice a problem, the computer constantly compares live activity against a set of rules—like a map of protected fishing waters or a list of standard behavior. This process, often powered by an algorithm—a specific set of mathematical instructions used to perform a task—allows for constant, silent observation. In the past, the law only reached as far as the nearest patrol boat or police cruiser. Today, the enforcement cycle begins when an algorithm identifies an anomaly. The computer looks for tell-tale signs of non-compliance, such as a vessel disabling its location transmitter or a car matching a specific movement profile, and creates an alert. This changes the job from hunting for problems to reviewing the potential violations that the software has already prioritized for investigation.

This move toward automated surveillance is not just a technical change; it is a fundamental shift in how power operates. When an algorithm flags someone as a potential lawbreaker, that decision is often hidden inside what is known as a black box—a system so complex that its internal reasoning is unclear even to the people using it. While this can help authorities cover more ground, it creates a new type of vulnerability. If the data feeding the system is incomplete, or the software is biased, an automated report could trigger a confrontation based on a flawed calculation. As these digital systems become the new front line of governance, we have to ask ourselves: who is responsible when the software gets it wrong, and how do we hold a digital assistant accountable for a decision that affects a person’s real-world freedom?