Using AI to expose illegal fishing at sea

Illegal fishing threatens marine ecosystems, undermines coastal economies, and often goes hand in hand with other illicit activities such as smuggling and human trafficking. Yet detecting it remains a huge challenge. Fishing vessels operate far from shore, and while they are required to transmit their location through the Automatic Identification System (AIS), these data is often incomplete, misleading, or deliberately falsified.

At the University of Liverpool, researchers are tackling this challenge by developing new ways to spot suspicious behaviour in AIS data using artificial intelligence. Instead of relying on self-reported information about vessel type, which is often inaccurate, researchers have trained algorithms to classify ships based on the sequence of ports they visit. This novel approach makes classification of ship types more resilient to misreporting and manipulation.

The researchers also applied an automated method to identify “coopering” events, when two ships come unusually close together at sea. Such encounters can be dangerous, and repeated patterns may indicate illegal activity such as unregistered transfers of fish or goods. By analysing just one month of satellite AIS data from Southeast Asia, the team flagged 17 cases of unusual behaviour warranting further investigation.

This research, which has been published in the Journal of Marine Science and Engineering, is a step towards smarter surveillance systems that can help authorities prioritise where to focus their attention. Any future system would likely combine AIS analysis with other data sources, such as high-resolution satellite imagery, to confirm suspicions and guide enforcement. These AI-driven methods could play an important role in reducing illegal fishing and protecting the world’s oceans.