Last summer the Department of Defense loaded three beehives onto a truck and drove them to a field deep in the Texas foothills. Each hive contained roughly 10,000 bees, which were counted electronically as they flew in and out. Bees forage by smell and would normally investigate a field like this in long, lazy loops. Yet almost immediately the drones began clustering—not around flowers, but around petri dishes. As it happened, the field contained four dishes, three of which had been dusted with trace amounts of dinitrotoluene, an explosive similar to TNT. When researchers analyzed the tapes afterward, they confirmed the success of the test. The bees had gone straight for the explosive—just as they had been trained to. 

The U.S. military isn’t typically assigned insect recruits, but that could change if Alan Rudolph gets his way. A zoologist by training, Rudolph joined the Defense Department seven years ago to head the Controlled Biological and Biomimetic Systems (CBS) program, a six-year, $60 million pilot set up by the Advanced Research Projects Agency, or Darpa, in 1996. It was Rudolph’s idea to enlist animals, and especially insects, in the war on terror. “They constitute the largest biomass on the planet and occupy the widest range of environments,” he explains. “Some of them can also be trained.” Rudolph has acquired the defense industry knack for being conversational without saying too much, and his explanations sometimes come off like a grant application. He came to Darpa, the DOD’s futuristic idea factory also responsible for the Internet, after spending a decade with the Office of Naval Research, where he worked on the problem of freeze-drying blood by studying a dehydration-proof worm. The ONR had no category for a zoologist, Rudolph recalls, “so they called me a chemist.” 

Insects are nature’s no-tech nanobots; virtually ineradicable, they’re DIY biosystems that adaptation and speciation have equipped with extraordinary tools for survival. Indeed, evolution, in Rudolph’s view, has already solved many of the problems military designers find most vexing, such as identifying airborne chemicals at a distance or maneuvering rapidly over rough terrain. Great sphinx moths, for instance, have become models for next-generation sensors: A highly acute sense of smell enables males to track females from miles away and through bidirectional winds. Rudolph, though, champions the bee, both for its speed—it can cover a mile in just four minutes—and for its intelligence. 

“Bees can be trained to be attracted by many smells,” says University of Montana biologist Colin Henderson, a Darpa grantee, “anything from decomposing bodies to biological weapons.” At Henderson’s boot camp, the insects are taught to associate the taste of sugar with a particular scent: explosives, say, or the byproducts of an underground weapons factory. In addition to sniffing out mines in an open field, the bees could someday be deployed as roving reconnaissance agents, zooming through the streets of Baghdad in search of fissionable material. With this in mind, Henderson and his colleague Jerry Bromenshenk have prototyped a small, portable hive that can be used to airdrop bees into hostile territory. 

Rudolph’s battle bugs are somewhat unique, but the idea of harnessing animals for military use is not a new one. During the Vietnam War, the Navy is rumored to have deployed a top-secret team of knife-wielding “warrior porpoises” to attack Vietcong divers, and there has even been talk of nuclear whales: Orcas trained to tow warheads to enemy shores. Though reports have not been verified, defense contractor Sandia may have developed an experimental weapons delivery system consisting of a mule controlled by a sun compass and brain electrodes. During the Cold War, the Soviet Union’s program in animal warfare caused American generals to worry about the possibility of a growing “dolphin gap”—and a possible combat-dolphin escalation between the superpowers. 

That China might begin fretting over a bee gap seems less likely, given the many practical difficulties associated with airdropped hives. This doesn’t bother Rudolph, who sees plenty of opportunities for bug deployment closer to home. Among other things, bees have been trained to work as scouts for environmental cleanup sites and to sniff out drugs (part of a theoretical plan to equip every drive-through customs station with its own hive). In Montana, tests have already proven that trained bees will swarm a truck carrying a bomb and ignore decoys. “It’s like having a thousand flying dogs,” Rudolph says. And because their fur has a slight positive charge, bees can also double as passive reconnaissance agents. Motes and spores, such as pollen, stick to the bees’ pipe-cleaner bristles, which can be analyzed to reveal faint traces of TNT, anthrax, methamphetamines, or other contraband once the drones return to the hive. Alternatively, bees coated with reactive agents could become flying chem labs, treated to fluoresce in the presence of sarin or smallpox. 

As malleable as they are, bees have their limitations. They won’t go out in the dark or the rain, and even under ideal circumstances they can be hard to track. One Darpa contractor proposed equipping the bees with miniature transceivers, but so far even the lightest microdevices upset the insects’ equilibrium, and in any case have a range of less than two miles. The alternative, radar, would only work on large swarms, or on bees equipped with tiny titanium antennae. 

To circumvent these problems, Rudolph has begun toying with the idea of building a more reliable bee, one with a motor that can be controlled remotely. This is the assignment of CBS' biomimetics division, where researchers carefully record every motion of spiders and flies, even mapping wing turbulence in specially constructed miniature wind tunnels. In the interest of understanding flight, one University of Washington biologist recently suggested wiring a moth with superlight accelerometers attached to its wings with superglue and baking soda. 

It’s a tempting idea. Why design a mechanical system from scratch when you can crib a field-tested version from nature? And why gamble on a fragile live animal when you can build an indestructible one? In practice, of course, things are a bit more complicated. Since Darpa coined the term biomimetics a decade ago, the idea has blossomed into a major industry. Companies as diverse as Monsanto and Xerox have seized on it to create “organizational ecosystems” based on the growth patterns of redwood forests. More literally, the Canadian Air Force has begun painting planes with false cockpits—ones designed to fool enemy pilots, like the fake eyespots on a beetle’s behind. 

If it all sounds a little far-fetched, that’s the problem. For every useful insight, biomimicry has produced a bin full of crackpot washouts—robot flies that “die” without batteries, mechanical geckos that fall off the wall—none of which are proving especially good at surveillance. And while it’s true that a robomoth won’t bang its equipment into a porch light, we’re still nowhere near building replicas that function as well as their evolutionary counterparts. 

As a result, one of Darpa’s newer strategies has been to combine the best of both worlds, rigging live animals with microelectronic controls. Last January, viewers of ABC’s Evening News got to see one of CBS' weirder successes: a live rat that could be maneuvered remotely by a researcher using a joystick. The rat, which had electrodes implanted in its brain and a tiny camera strapped to its back, steered forward, back, left, and right, and was touted as a tool for searching through rubble. Put the same system in a bee or a cockroach and you have a fine tool for espionage—though admittedly, one vulnerable to birds, flyswatters, and heavy boots, not to mention Raid. (One imagines agriculture collapsing as terrified governments begin defensive fumigation.) 

Under the circumstances, it’s hard not to wonder whether Rudolph’s multimillion-dollar bee budget might be better spent. While it’s conceivable that bees will eventually surpass police dogs as the border patrol species of choice, it seems more likely that bugs will revert to their status as, well, bugs—prototypes to be studied, then brushed aside as nanotechnology flies off the drawing board and into Circuit City. In that case, a $4 million gnat would turn out to be as useful as the $600 hammers and $1,000 toilet seats in the Pentagon’s supply closet. But even if Rudolph’s program succeeds unexpectedly, its victory could be Pyrrhic. After all, who wants to live in a world where every bug might be bugged, where nature is suspect, and wilderness a breeding ground for covert surveillance? This is not Alan Rudolph’s intent, of course. He’d like to see his insects keep the peace, locating land mines so they can be safely defused. 

In the meantime, keep an eye on the screen door.