The ocean is already teeming with sensors, and has been for over half-a-billion years. Those sensors are cost-efficient, fully autonomous, and rapidly process the information around them into actionable intelligence. Harnessing this existing realm of sensors means converting the organs that let sea creatures live into an early warning system for things like submarine intrusion. It’s an audacious goal, and a task the Defense Advanced Research Projects Agency has set to solve: Can the creatures of the sea become an sensor platform?

Under the friendly acronym PALS, for Persistent Aquatic Living Sensors, DARPA wants to “tap into marine organisms’ innate abilities to sense and respond to perturbations in their environments” and then use those sense to detecting, characterizing, and reporting on human-made objects traveling underwater, from uncrewed submersibles to nuclear submarines.

Fortunately for the undersea PALS enlisted into submarine detection, it looks like the program envisions a largely passive use of animal sensing capabilities. That is, rather than some gothic horror of fish bound together in a human-made cage feeding signals through wires into a control unit, the animals will be left free to roam. Instead, the human-controlled sensor will simply observe animals as they are, and note the differences in animal behavior between when a submarine is nearby, and when one isn’t.

On April 11, DARPA selected Northrop Grumman to prototype those sensing capabilities. To that end, Northrop Grumman says it will “develop biological sensing hardware that has increased sensitivity for certain sensor modalities, achieving greater range.” Then, using artificial intelligence, this hardware will observe patterns in the marine environment. With luck and iteration, it might be able to tell the difference between a school of fish fleeing a seal and a school of fish fleeing an underwater robot.

Ultimately, any sensor system built from PALS will have to distinguish between sea creature reaction to human-created vehicles and a range of false positives, from debris or other sea creatures. Should it work, it would open up and undermine whole fields of biomimicry. If fish really do scatter differently for robots and manta rays, that could erode any stealth gained from robots shaped like manta rays. Conversely, if there’s a way for submarine movements to be disguised as, say, the movement patterns of sharks or the drifting of debris, then the new PALS sensors could be cleverly fooled.

Should any of this work, the immediate advantage for the first military that could tap into the natural sensing going on throughout the ocean is almost unfathomable.