An academic team has built a database system that uses large language models to track illegal fishing, seafood fraud, and forced labor across global supply chains.
The project, called IUU+DB, ingests unstructured documents — enforcement reports, news articles, trade records — and classifies whether each one describes a relevant incident. From there, it extracts structured fields: actors, vessel names, species, locations, violations, and enforcement outcomes. The researchers coined the term "IUU+" to stretch beyond the standard definition of illegal, unreported, and unregulated fishing and capture the wider web of supply chain crimes tied to the industry. Case studies showed the system can surface geographic hotspots and behavioral patterns that scattered evidence alone would not reveal.
Fisheries crime is genuinely hard to quantify. Incidents are documented across dozens of jurisdictions, languages, and agencies, and no single database captures the full picture. A system that can hoover up heterogeneous documents and produce structured, deduplicated records could give regulators, NGOs, and industry risk teams a shared view of a problem they currently see only in fragments.
The tool is framed for researchers and policymakers, but the same extraction pipeline — classify, extract, deduplicate — is the playbook behind most compliance and due-diligence software already sold to seafood importers. Whether IUU+DB stays academic or finds a commercial home will depend on who funds it next.