In June 2020, federal officials explored using a millimeter-wave heat weapon and a long-range acoustic device to disperse protesters outside the White House. They were advised the National Guard didn’t have either system on hand, so neither was deployed. In 2018, the Department of Homeland Security considered using the same heat weapon at the U.S.-Mexico border; Secretary Kirstjen Nielsen rejected the idea and forbade it being discussed again. In January 2026, police in Minneapolis deployed a directional LRAD against demonstrators protesting federal immigration enforcement operations, following the killings of two people by federal agents. In March 2025, opposition officials and rights groups in Serbia alleged that a military-grade sonic weapon was used against peaceful anti-corruption protesters; the government denied it.
The technology exists. It’s been tested on over 10,000 volunteers. It’s been deployed to Afghanistan and withdrawn without use. It’s been considered for protesters, prisoners, and migrants. It fills what the Pentagon describes as “the gap between shouting and shooting”—a phrase that manages to be both precisely accurate and deeply unsettling, depending on whether you’re the one doing the shouting or the one being shot at with a beam of concentrated pain.
What these systems actually are
The term “non-lethal weapons” covers a wide category, from rubber bullets and tear gas to technologies that sound like they were invented by a defense contractor who read too much science fiction. The three most technologically advanced systems—and the ones raising the most urgent ethical questions—are directed-energy heat weapons, long-range acoustic devices, and high-powered microwave systems designed for electronic disruption that can be repurposed for personnel effects.
The Active Denial System is the flagship. Developed by the Air Force Research Laboratory and the Joint Non-Lethal Weapons Directorate over roughly $40 million and two decades, it projects a focused beam of 95-gigahertz millimeter-wave electromagnetic energy at a target up to 500 meters away. The beam penetrates the skin to a depth of about 1/64 of an inch—enough to heat the water molecules in the outer skin layer and trigger an intense burning sensation. The pain is immediate, reflexive, and by design intolerable. Subjects move out of the beam involuntarily. In approximately 10,000 test exposures on volunteers, the injury rate was less than 0.1 percent—six pea-sized blisters across all tests. One airman in 2007 received second-degree burns on both legs after an overdose and was hospitalized for two days. One lab accident in 1999 produced a small second-degree burn. The Pentagon’s Human Effects Advisory Panel concluded the system has “a high probability of effectiveness with a low probability of injury.”
The Long Range Acoustic Device—manufactured by Genasys and used by every branch of the U.S. military, the Navy on every ship, and police departments in multiple countries—creates a focused 30-degree beam of sound capable of reaching 137 to 154 decibels, depending on the model. For reference, the human pain threshold for sound is approximately 120 decibels, and OSHA requires hearing protection above 90. The LRAD has two modes: a “voice” mode that functions as an extremely directional loudspeaker capable of projecting intelligible speech at distances up to 1,500 yards (one military officer described it as “the voice of God”), and an “alert” mode that emits loud chirping or beeping sounds at the top of the device’s decibel range. The voice mode is a communications tool. The alert mode is a weapon. Protesters exposed to LRAD alert tones in New York City reported migraines, sinus pain, dizziness, facial pressure, and persistent ear ringing. The city settled the resulting lawsuit in 2021 and agreed to ban the alert feature.
Beyond these two, the broader category includes stun grenades, pepper spray and tear gas, water cannons, rubber bullets (which have caused permanent injury and death despite being classified as “less lethal”), and Tasers (which have been implicated in hundreds of deaths, predominantly among people with cardiac conditions).
The doctrine problem
The Pentagon’s official definition of non-lethal weapons specifies that they are “explicitly designed and primarily employed so as to incapacitate personnel or material, while minimizing fatalities, permanent injury to personnel, and undesired damage to property and the environment.” The key word is “minimizing.” Not eliminating. Every non-lethal weapon carries a probability of causing serious injury or death. The question isn’t whether they can kill—rubber bullets can and have—but whether the probability is low enough to justify their use in situations where lethal force would be disproportionate.
The Active Denial System was designed for military applications: checkpoint defense, perimeter security, area denial in counterinsurgency operations where civilian casualties undermine the mission. Its proponents describe it as a way to give forces “decision time”—time to assess whether a threat is real before resorting to lethal force. A former Principal Deputy Assistant Secretary of Defense called its recall from Afghanistan an “opportunity missed,” arguing that its non-lethality could have been critical in operations where avoiding civilian casualties was essential.
But the technology doesn’t stay in the military domain. Raytheon marketed a reduced-range version for law enforcement. The Los Angeles Sheriff’s Department announced plans to use it in a detention facility to break up prisoner fights. Federal officials explored its use against White House protesters. The migration from military weapon to crowd control tool is not a slippery slope argument—it’s a documented trajectory.
The ethics of pain as policy
The philosophical problem with non-lethal directed-energy weapons is not that they kill. It’s that they work. A weapon that inflicts unbearable pain without leaving marks, without requiring physical contact, without producing the visual spectacle of tear gas clouds or water cannon impacts, and without (usually) causing lasting physical injury is a weapon that is very easy to use and very difficult to regulate.
The Active Denial System causes no visible injury in 99.9 percent of exposures. There’s nothing to photograph for the evening news. There’s no residue to test, no canister to trace, no wound to document in a hospital. A government using tear gas on protesters produces images that circulate globally and generate diplomatic consequences. A government using a millimeter-wave beam on protesters produces people who feel like their skin is on fire and then, once they’ve moved, feel completely normal—with nothing to show for it.
Physicians for Human Rights has raised concerns about the system’s short- and long-term medical impacts, noting that the testing conducted so far—exclusively on volunteers who consented and could leave the beam at will—cannot replicate real-world deployment conditions where people may be trapped, restrained, or unable to escape. The organization stated it is “hard to conceptualize a test that would fulfill federal ethics guidelines for research on human subjects” while adequately studying the weapon’s effects on non-consenting populations in uncontrolled conditions. The 2007 burn injury occurred because the subject received an overdose—too much energy for too long. In a crowd scenario, where the beam operator can’t monitor individual exposure times for hundreds of people, the probability of overdose is not zero.
The LRAD presents a different version of the same problem. As a communications device, it’s unambiguously useful—a way to clearly deliver instructions at distances where megaphones fail. As a weapon, it causes hearing damage to people who may not be able to leave the area, may not understand the instructions being broadcast, or may be exercising their legal right to be present. The New York City settlement acknowledged this by banning the alert feature while preserving the voice feature—a legal distinction between “talking to people loudly” and “hurting people with sound.”
Countermeasures against these systems are straightforward enough to raise their own questions about tactical utility. The Active Denial System’s beam is absorbed by water—rain, fog, and sea spray degrade its effectiveness. Heavy clothing reduces skin exposure. A metallic sheet or even a trash can lid can reflect or block the beam. The LRAD’s effectiveness drops with distance and atmospheric conditions. These limitations suggest that the weapons are most effective against lightly clothed, unprotected populations in clear weather—which describes protesters in summer cities more accurately than it describes adversaries in combat zones.
What the market says
The global non-lethal weapons market was valued at roughly $3.8 billion in 2017 and is projected to reach $6.6 billion by 2026. The growth is driven by what market reports describe with remarkable directness: “demand for crowd control weapons to tackle protests and riots.” The demand signal isn’t coming primarily from military applications. It’s coming from governments that want more sophisticated tools for managing domestic unrest—tools effective enough to disperse crowds and clean enough to avoid the political costs of visible violence.
The gap between shouting and shooting is real, and weapons that fill it will save lives in some scenarios. A soldier at a checkpoint who can compel a vehicle to stop without firing is a soldier who doesn’t accidentally kill a family that didn’t understand the hand signals. The question isn’t whether non-lethal weapons should exist. It’s whether a technology designed to inflict pain without evidence—scalable, deniable, and deployable against any population a government designates as a target—can be governed by frameworks designed for weapons that leave marks.
We cover non-lethal weapons alongside directed energy, autonomous systems, and the full landscape of emerging military technology across our Battlefields of the Future course—including why the weapons most likely to be used on civilians are the ones specifically designed not to kill them.