Scientists just trained AI to spot hidden seahorses, shark fins, and sea cucumbers in airport X-ray scans, and the 86% to 96% detection rates could give customs officers a new way to catch marine wildlife smugglers

Wildlife traffickers have long relied on one simple advantage: most people do not know what a dried seahorse, shark fin or sea cucumber looks like when hidden inside luggage.

But a group of Australian researchers now believes artificial intelligence can make it easier.

In what the researchers say may be a world-first, scientists have developed and tested an AI system for detecting trafficked marine wildlife in luggage and mail pathways using 3D CT X-ray scans. Used in conjunction with the 3D X-ray scanners already in place at some airports, the system successfully identified scans containing seahorses, shark fins, and sea cucumbers between 86 and 96 per cent of the time, according to the study published in Frontiers in Ocean Sustainability.


According to the study, the AI correctly identified 95% of shark fins, 95% of seahorses, and 85% of sea cucumbers in luggage CT/X-ray scan tests. The false-positive rate, wrongly identifying a package as containing wildlife, was around 13 per cent in the study.

A growing but often overlooked wildlife crime

A growing problem largely ignored, wildlife trafficking is a well-documented issue that often evokes imagery of elephant ivory or rhino horn, but researchers argue marine species are becoming more and more attractive to organised crime syndicates.

Seahorses are used widely in traditional Chinese medicine, shark fins are a delicacy, and sea cucumbers are highly prized in international markets. The trade in marine wildlife is, in part, a result of their commercial value. Last year, law enforcement authorities confiscated tens of thousands of live animals and large volumes of protected wildlife products, including marine species, during a global anti-trafficking operation coordinated by Interpol and the World Customs Organization.

These figures outstripped the combined seizures of trafficking in birds, primates and reptiles, revealing the magnitude of the issue. According to scientists involved in the research, the illegal trade in marine animals remains drastically underreported compared to trade in terrestrial animals.

This disparity, says Sarah Foster, a fisheries researcher at Canada’s University of British Columbia, is down to a public perception problem, with people far less aware of the trafficking in marine life than that of iconic land mammals.

"Our biggest challenge in ocean conservation is getting people to recognise fish as wildlife," she told Mongabay.

Training AI to think like a smuggler

AI trained on smuggling methods. To train its system, the researchers sourced 68 samples of real trafficked marine species, including shark fins, sea cucumbers and seahorses - much of which had been collected by law enforcement and placed in museum collections.

They then spent months re-creating thousands of realistic smuggling scenarios for scanning. Smugglers tried stuffing their contraband in every conceivable place, including wrapped in foil, hidden in socks, embedded within soft toys and disguised within more ordinary food products. With thousands of simulated bags loaded with nothing but innocent items, and an equal number of fake suitcases stuffed with seahorses, sea cucumbers and shark fins, they exposed them to scanning technology used in many airports, a 3D computerised tomography (CT) X-ray machine.

The AI was trained to learn the distinctive patterns of these marine animals.

"We spent six months thinking how traffickers think," she told Mongabay.

Not replacing humans, but helping them

An aid, not a replacement. While the new technology shows promise in the fight against wildlife trafficking, the authors are careful to stress it’s not a silver bullet and will not replace border officials, biosecurity officers or sniffer dogs.

According to Dr Pirotta, this technology should be seen as an assistant for human detection and as a first layer of defence.

"We can only mock up real-world trafficking scenarios based on what has been detected before. AI is not a silver bullet for detection, nor a replacement for human and sniffer dog detection," she said. The United Nations Office on Drugs and Crime agrees. In a statement to Mongabay, a spokesperson said that detecting suspicious baggage is the first step; however, success in securing convictions often relies on evidence collection, investigative work, and prosecutorial proceedings.”

Challenges remain

Despite the successes, the researchers identified a few limitations in their system.

The current AI can only identify the dried product, not live animals. The technology also works only with suitcases and other small pieces of luggage, as scanning large shipping containers for trafficked wildlife requires different techniques. Furthermore, not all airports have 3D CT scanners, and the AI was not able to identify sea cucumbers with as much success as it did sharks or seahorses, as these creatures vary in shape more widely. Nevertheless, the technology represents a positive development in a field struggling to keep up with criminals.

The United Nations Office on Drugs and Crime (UNODC) said AI-assisted scanners may play an increasing role in wildlife enforcement, calling the development an important innovation.

For wildlife conservationists, the research also marks a rare occasion in which AI has been used for more than simply optimising travel or increasing workplace efficiency, to directly protect wildlife.