On a Sunday in April, Argentine officials stopped an unusual shipment at an airport near Buenos Aires. Inside, they found so many dead and dying fish, octopuses, and crabs that a rescue center had to install ten emergency tanks just to keep the survivors alive.
It was the third illegal sea life seizure at the same airport in a single year. Marine wildlife trafficking is a growing global problem, and for a long time, the tools to fight it have not kept up.
Now, a team of researchers believes artificial intelligence could help change that, and the early results are genuinely promising.
Hiding in Plain Sight
Most people picture elephant ivory or rhino horn when they think about illegal wildlife trade. Marine animals rarely get the same attention, even though the numbers tell a very different story.
In 2025, Interpol seized 91,000 pieces of trafficked marine life, nearly double the combined total of reptiles, birds, and primates. The trade is fueled by demand for ornamental fish, luxury seafood, and traditional medicines, and most of it moves quietly through airport luggage and airmail packages.
Sarah Foster, a fisheries researcher at the University of British Columbia, put it simply. “Our biggest challenge in ocean conservation is getting people to recognize fish as wildlife, the way they care about elephant ivory or rhino horn,” she said.
Marine species have long lived in the shadow of their land-based counterparts when it comes to trafficking awareness, enforcement attention, and conservation funding. That gap has allowed a massive illegal trade to grow largely unnoticed.
Beyond biodiversity loss, illegal wildlife trafficking spreads infectious diseases, introduces invasive species, and connects to other forms of organized crime and labor abuse. The scale of the marine trade makes closing that recognition gap more urgent than ever.
Think Like a Smuggler
Researchers at Macquarie University in Australia, led by marine biologist Vanessa Pirotta, set out to build the first AI algorithm dedicated specifically to detecting trafficked marine wildlife. They chose three target species: shark fins, seahorses, and sea cucumbers, all commonly smuggled and all protected under international trade agreements.
Collecting samples was an experience in itself. Most came from the Australian Museum, originally seized in real trafficking busts. One shark fin was collected fresh from a beached bull shark.
Another, of unknown species, was simply bought from a grocery store in Sydney’s Chinatown alongside some dried sea cucumber samples. “I was just like every other tourist going into Chinatown, it was a non-event,” Pirotta recalled. Dried fins and sea cucumbers were easy to find, which says a lot about how openly parts of this trade operate.
The team then spent six months thinking carefully about how smugglers actually hide their cargo. They created nearly 6,000 simulated bags, 3,500 containing hidden animal samples buried among toys, clothing, and tin foil, and another 2,400 bags with no animal parts at all.
All of these were scanned using 3D X-ray machines, and the resulting images were used to train the algorithm to recognize what trafficked wildlife looks like when it is deliberately concealed.
What the AI Found
The results were encouraging. The algorithm detected shark fins and seahorses with 95 to 96 percent accuracy. Sea cucumbers proved slightly harder, with an 86 percent detection rate, largely because they vary more in shape and size than fins or seahorses do.
False alarm rates were low, just one to two percent for shark fins and sea cucumbers, and nine percent for seahorses.
The algorithm performed best with species that have distinctive, recognizable shapes. Fins look like fins. Seahorses look like seahorses. Sea cucumbers, on the other hand, come in many forms, which made the AI’s job harder.
Pirotta noted that distinguishing between legal and illegal samples within the same species could be an even greater challenge as the technology develops further.
There are other honest limitations worth acknowledging. The algorithm has so far only been tested on dead, mostly dried samples. It requires 3D X-ray machines to work, which are not yet available at every airport around the world.
And it was designed around small carry-on luggage, meaning a large portion of sea freight moving through shipping containers is still beyond its reach. “We know we’re missing a lot,” said Michelle Anagnostou, a researcher at the University of Oxford who is currently working on AI tools for maritime shipments.
Not a Magic Fix
Everyone involved in this research is careful to frame the AI as one part of a larger solution, not a complete answer on its own. “AI could be used to complement existing detection. It’s not a silver bullet, but an assistant and a tool,” Pirotta said.
Right now, front-line detection of smuggled wildlife relies almost entirely on human inspection and biosecurity dogs. AI could work alongside those methods, flagging suspicious bags faster and more consistently than any single officer could manage alone.
The United Nations’ Office on Drugs and Crime made the point clearly: “Technology can flag a bag. People, forensics and prosecutors turn a flagged bag into a sentence.” Detection is only the first step in a much longer chain.
Anagnostou also stressed the importance of thinking beyond seizures entirely. “We’ve been arresting people for decades and it hasn’t gotten us very far,” she said, pointing to the need for better-resourced enforcement agencies, public education, and efforts to tackle corruption and poverty in the countries where trafficking originates.
Pirotta’s next step is to share the algorithm’s framework so researchers in other regions can adapt it to detect additional species.
For her, the project has been a personal journey as much as a scientific one. “When I first started this work I never thought AI would be such an instrumental part of what I do as a scientist,” she said. “I’m optimistic, knowing it’s not going to be the answer to everything, but genuinely excited about what it can contribute.”
Sources:
Leave a Reply