Overview
- A peer‑reviewed study led by Vanessa Pirotta and published on Sunday, June 7, 2026, reports an AI trained on 3D CT luggage scans that correctly flagged target marine products about 92 percent of the time and produced an overall false positive rate near 13 percent.
- The system performed differently by species, detecting shark fins about 95 percent of the time, seahorses about 96 percent, and sea cucumbers about 86 percent, with species‑specific false positive rates reported in the study.
- Researchers built and tested the model on 298 scans made from seized specimens and reference samples, simulated concealment scenarios such as wrapping and hiding items inside toys, and used Threat Image Projection to insert samples into real bag scans for training.
- The team says the tool is intended as a supplemental screening aid for frontline airport staff rather than a replacement for manual inspection or detector dogs because false positives will still need human follow‑up and not all airports have 3D CT scanners.
- The work targets a large but under‑detected marine wildlife trade worth an estimated $20 billion a year, highlights likely under‑recording of sea cucumber smuggling, and could improve seizures and data on smuggling routes if the technology is scaled and operational challenges are addressed.