Amid a rising tide of low-cost weaponized adversary drones menacing American troops abroad, the US military is pulling out all the stops to protect its forces from the ever-present threat of death from above. But between expensive munitions, futuristic but complicated directed energy weapons, and its own growing drone arsenal, the Pentagon is increasingly eyeing an elegantly simple solution to its growing drone problem: reinventing the gun.

At the Technology Readiness Experimentation (T-REX) event in August, the US Defense Department tested an artificial intelligence-enabled autonomous robotic gun system developed by fledgling defense contractor Allen Control Systems dubbed the “Bullfrog.”

Consisting of a 7.62-mm M240 machine gun mounted on a specially designed rotating turret outfitted with an electro-optical sensor, proprietary AI, and computer vision software, the Bullfrog was designed to deliver small arms fire on drone targets with far more precision than the average US service member can achieve with a standard-issue weapon like the M4 carbine or next-generation XM7 rifle. Indeed, footage of the Bullfrog in action published by ACS shows the truck-mounted system locking onto small drones and knocking them out of the sky with just a few shots.

The Bullfrog appears effective enough against drone targets to impress DOD officials: According to Defense Daily, Alex Lovett, the deputy assistant secretary of defense for prototyping and experimentation within the Pentagon’s Research and Engineering office, told reporters at a demonstration event in August that the testing of the “low-cost” Bullfrog solution had “gone really well.” Should the Pentagon adopt the system, it would represent the first publicly known lethal autonomous weapon in the US military’s arsenal, according to the Congressional Research Service. (The Office of the Secretary of Defense did not yet respond to WIRED’s request for comment.)

Shooting down small, fast-moving drones with conventional firearms is a significant challenge to even the most talented marksman, and the US military has been pursuing various ways to make its small arms more effective against unmanned airborne threats. Those efforts include the procurement of small- to medium-caliber munitions and “buckshot-like” ammo that can replicate the effects of the shotguns that have proven effective counter-drone measures amid Russia’s invasion of Ukraine; rifle-mounted radio frequency and GPS jammers to disorient incoming drones so troops don’t have to carry separate, bulky counter-drone weapons like the Dronebuster or NightFighter; and “smart” optics from companies like SmartShooter and ZeroMark that purportedly only allow a weapon to fire when it locks on target. The Army has even started integrating counter-drone exercises into its basic training regimen, part of a broader effort to make such schooling as “routine” as conventional marksmanship training.

For ACS cofounder and CEO Steve Simoni, a former Navy nuclear engineer, the best way to optimize a firearm for drone threats isn’t through novel accessories or enhanced training, but a combination of advanced robotics and a sophisticated AI that can take the guesswork out of target acquisition and tracking.

“During the Russian invasion of Ukraine, we [Simoni and his ACS cofounder, fellow Navy veteran Luke Allen] saw the proliferation of drones on both sides of the conflict, and we read in various news outlets the Ukrainians were firing AK-47s in the air at them,” Simoni tells WIRED. “We thought, ‘That’s a good robotics problem.’ It’s hard to hit something flying so fast, but a robot can do that with modern-day computer vision and AI control algorithms.”

While smart optics like SmartShooter’s SMASH fire control system also use AI and computer vision to track potential targets, Simoni believes that physically keeping a rifle zeroed in on a fast-moving drone is a more significant challenge than even a highly trained human soldier with an advanced weapon-mounted scope is equipped to address. The solution, Simoni says, is to remove humans from the equation altogether and allow an extremely precise, custom-built motion control system to maintain a steady bead on an incoming drone amid the chaos of a battlefield.

“We are electrical engineers, and we decided that in order to solve this problem of hitting a fast drone that’s accelerating at five Gs at a couple hundred yards, you would need an incredibly high-end current that goes through a motor and encoders that know the position of your gun at all times,” Simoni says. “To put that form factor in the hands of someone with an M4 seemed like a very tough problem.”

“A DJI Mini [drone] is a little bit bigger than my hand, and our system can down one at 200 yards with two shots,” he adds. “No human could make that shot.”

Gun turrets are nothing new for the US military and its allies, from the manually operated Common Remotely Operated Weapon Station (CROWS) that adorns a growing number of American ground combat vehicles to the beloved semi-autonomous Phalanx Close-In Weapon System (CIWS) that’s essential to defending Navy warships from incoming missiles. More recently, the Army has experimented with several other gun-based counter-drone solutions—including a Ghost Robotics robot dog with its own AI-enabled AR/M4-pattern rifle turret.

At less than 400 pounds, the Bullfrog is smaller and simpler than the hulking 15-foot-tall, 12,000-plus-pound Phalanx CIWS and, in turn, better suited to provide flexible protection for individual platforms like, say, a Joint Light Tactical Vehicle on the move. And where the Phalanx CIWS sprays what essentially amounts to “a cloud of bullets,” Simoni claims the Bullfrog is precise enough to not only down airborne targets with minimal ammo expended, but deliver crossing shots at targets from the side, allowing the turret to accurately engage threats to units other than itself—and, with multiple systems, create a layered air defense network.

“Internally, we used to call the Bullfrog the ‘mini-CIWS,’” Simoni says. “But the CIWS control system isn’t as accurate.”

Then there’s the matter of the Bullfrog’s autonomy, a prickly subject for military planners. In accordance with the Pentagon’s current policy governing lethal autonomous weapons, the Bullfrog is designed to keep a human “in the loop” in order to avoid a potential “unauthorized engagement.” In other words, the gun points at and follows targets, but does not fire until commanded to by a human operator. However, ACS officials claim that the system can operate totally autonomously should the US military require it to in the future, with sentry guns taking the entire kill chain out of the hands of service members and freeing them up to focus on other, more mission-essential tasks.

“Our system is fully autonomous-capable, we’re just waiting for the government to determine its needs,” Brice Cooper, ACS’s chief strategy officer and a Green Beret veteran who previously served as head of US Special Operations Command’s counter-drone program, tells WIRED. “Legacy systems in our category are just not there yet.”

The Bullfrog’s debut at T-REX couldn’t have come at a more fortuitous time. The Pentagon recently announced the second increment of its Replicator initiative that was launched last year to catalyze a surge in US military drone and counter-drone capabilities ahead of potential future conflicts with “great power” threats like China or Russia. Where the first Replicator increment focused on rapidly procuring and fielding low-cost, attritable drones for US forces around the world, the second will tackle “the threat posed by small uncrewed aerial systems (C-sUAS) to our most critical installations and force concentrations,” as Defense Secretary Lloyd Austin put it in a late September letter announcing the new stage of the initiative.

American military commanders in the Middle East have for years maintained that cheap, easily weaponizable commercial drones represent the greatest threat to their deployed forces since the advent of the improvised explosive device during the Global War on Terror, a concern driven home by attacks on Navy warships and commercial vessels in the Red Sea and US troop outposts in Iraq and Syria in the year following the outbreak of the current Israeli-Hamas conflict. In January, three US service members were killed in a drone attack on a military outpost in Jordan near the Syrian border, an incident that the Pentagon later determined had occurred because American forces there lacked the appropriate capabilities to defend against incoming drone threats.

“The threat landscape has become hyper-accelerated,” Cooper says. “A year ago, nobody was really concerned about the small drones that are destroying all kinds of armored vehicles in Ukraine. In reality, we’re in the very nascent stages of what the unmanned threat landscape is going to look like, and it’s going to involve a serious investment from the Defense Department in perpetuity.”

It’s unclear at this time what specific types of counter-drone capabilities Replicator 2 might yield. At the moment, it appears that all options are on the table. They include remotely operated weapons turrets like the XM914 30-mm chain gun loaded up with explosive proximity rounds; missile systems like MSI’s Electronic Advanced Ground Launcher System (EAGLS) laser-guided rocket pod, L3Harris’s Vehicle-Agnostic Modular Palletized ISR Rocket Equipment (VAMPIRE) system, and Raytheon’s Coyote interceptor; “autonomous air vehicles” like the Air Force’s Paladin flying gun and the Roadrunner from defense upstart Anduril; vehicle-mounted sensor- and signal-jamming electronic warfare suites like the Marine Corps’ Light Marine Air-Defense Integrated System (L-MADIS) that’s been knocking drones out of the sky since 2019 and the Army’s Mobile Low, Slow Unmanned Aircraft Integrated Defense System (M-LIDS) that was most recently deployed to protect the service’s temporary pier in Gaza, both of which also feature “kinetic effectors” like machine guns and missiles; exotic directed energy weapons like high-energy lasers and high-powered microwaves to disrupt and disable drone electronics mid-flight; and a growing arsenal of infantry weapons from counter-drone rifles to the tried-and-true FIM-92 Stinger man-portable air-defense system. These weapons are all intended to turn every service member in American combat formations into a “drone defender.”

One major factor driving the Pentagon’s counter-drone push, however, is cost. Many kill systems currently in use are still disproportionately expensive defense against comparatively cheap drones. And while directed-energy weapons like lasers and microwaves, at something like $10 a shot, offer a far more cost-effective solution than traditional munitions, they are also still relatively experimental (and incredibly complicated) technologies that, despite their ongoing testing and even active use in the Middle East, are not yet seen as reliable enough to field at scale.

“If something like [the Bullfrog] works where you can modify an existing system across the fleet, that’s ideal,” Mike Clementi, a former congressional defense appropriator who previously worked on the US military’s counter-drone efforts, tells WIRED. “And if you can use it effectively for the cost of a handful of 7.62-mm rounds, you’re getting ahead of the game.”

“The real question is: How do you eliminate massive amounts of cheap targets?” he adds. “If the Bullfrog pans out, it would be the cheapest solution out there.”

Simoni believes AI-powered autonomous gun systems like the Bullfrog can provide enough precision to deliver a cost-per-kill ratio similar to that associated with laser and microwave systems without the intensive maintenance and logistics upkeep. And while the Bullfrog isn’t the only counter-drone gun system on the Pentagon’s radar, Simoni believes that the algorithmic accuracy enabled by the turret’s unique machine-learning software and motion control system remains superior to that of other offerings.

“Tracking and shooting is only the tip of the iceberg: Eventually, the system will consist of different guns, reaching out to longer ranges, engaging drones that are moving in different acceleration patterns—that’s all based on updates to our AI model,” Simoni says. “We envision a convoy of vehicles outfitted with turrets employing coordinated fires that are effective even if they’re driving down a bumpy road. That’s what we’re building over the next 12 to 18 months.”

Regardless of Simoni’s ambitious vision and the Pentagon’s recent push into the realm of autonomous weapons, questions still persist in defense circles about the potential dangers of taking a human operator out of the kill chain.

“Anything with robotics requires software to make the determination of friend or foe, and that’s a concern with anything that’s automated,” Clementi says. The use of fully automated systems would be “uncharted territory,” he says. “There’s always been a person in the loop before.”

Should the Pentagon end up embracing an autonomous robotic gun system as its counter-drone defense of choice, the next big conflict involving American troops could end up looking like a joke from an episode of The Simpsons: “most of the actual fighting will be done by small robots.”

“The future battlespace is incredible autonomous robots like ours shooting each other,” Simoni said. “I don’t think there’s much room for people with guns.”