But establishing these reserves is just one step. Enforcing their protection is another matter. And for many marine reserves—especially those in the Global South—there is no real way to do that, says Ted Schmitt, senior director of conservation at the nonprofit Allen Institute for AI (AI2). Thousands of square kilometers of open ocean is a lot to monitor. Even with satellites scanning the marine areas, the reality until recently was that you had to know what you were looking for: “When you have the vastness of the ocean, you can have analysts who are very well trained, looking for vessels,” he says. Even then, there is little chance of finding wrongdoing without intelligence from the ground.
In 2017, Microsoft cofounder Paul Allen began developing a tool called Skylight to provide analysts with more of that intelligence, using AI to help analyze satellite and ship-tracking data to detect suspicious behavior. The project moved to AI2 after Allen’s death in 2018, and the technology has since been adopted by more than 200 organizations in more than 70 countries. “We’re basically monitoring the entire ocean 24-7-365, and surfacing all these vessels,” says Schmitt.
To see how coast guards use the system, Schmitt points to a series of arrests in Panama in early 2025. That January, satellites found 16 boats about 200 kilometers off the coast inside the Coiba Ridge marine reserve, which serves as a migratory highway for sharks, rays, and large fish like yellowfin tuna. Skylight’s AI algorithms, trained to recognize the signature movements of various types of fishing, detected long-line fishing and requested higher-resolution images of the site from a commercial satellite flying overhead. The images and Skylight’s analysis were used by Panama’s environmental agency and military, which deployed ships and aircraft to the area, ultimately seizing six vessels and thousands of kilograms of illegally taken fish.
Skylight AI detects around 300,000 vessels per week, according to the company’s platform analytics. Stories like Coiba Ridge make clear that AI can benefit partners who are working tirelessly on the ground, says Schmitt: “The Panama case really was one of those ‘wow’ moments, not because the technology finally proved itself, but because the agencies that needed to operationalize it, and actually take it to a legal finish, did it.”
COURTESY OF WILDTECHDNA
Rapid DNA tests
When the conservation scientist Natalie Schmitt was researching snow leopards in remote areas of Nepal, she worked with people who could point out signs of these elusive big cats—often a pile of droppings. But the results weren’t reliable: Leopard scat can easily be confused with the poop of wolves and foxes, which share the same habitat and prey, she explains. What Schmitt wanted was a tool that could identify the animal involved, right on the spot—ideally by finding a way to sequence the DNA in the scat.
While some laboratories can take DNA samples of such material and identify species of interest, they are few and far between in rich countries and usually nonexistent in poorer countries, meaning that this process can be weeks long and involve shipping samples cross-country or across borders. This is a problem not just for field research but for wildlife trafficking enforcement. Imagine a border agent who has just opened a box of shark-like fins or a shipment of live parrots and needs to know whether the particular species is one that can legally be captured and transported. People in this situation don’t have weeks to spare.