Shell shock, cheaply

For roughly $500, an army can combine a camera, a battery and an airframe into a basic first-person-view drone. Add a warhead and its operator has a guided munition that can pursue a moving person or vehicle. More range, payload or protection against jamming costs extra. Even so, the arithmetic is compelling. Ukraine's defence ministry says its drone units recorded more than 800,000 verified strikes against enemy targets in the first five and a half months of 2026. In May alone the tally topped 181,000. Cheap guided weapons are now used on an industrial scale.1

The danger is not confined to soldiers. In May the United Nations verified at least 64 civilians killed and 539 injured by short-range drones, a category broader than FPVs. That was the highest monthly toll from such weapons since the full-scale invasion began; they were also the leading cause of civilian casualties near the front.

The front itself has become a belt rather than a line. In May Ukraine's drone commander put the corridor exposed to regular, dense strikes at 25km or more on either side, depending on the sector. Fibre-optic FPVs help explain the spread. Their pictures and commands travel through a filament, leaving ordinary radio jammers with nothing to drown out, though the cable can snag or be cut. Movement through this belt is now a combat operation. A military doctor in Kharkiv told Reuters in February that evacuation took more than three days on average. The UN has verified 20 short-range-drone attacks on clearly marked humanitarian vehicles or facilities in six months, most during evacuations.2

Those figures capture only the most visible damage. They do not distinguish primary blast injury from wounds caused by fragments, impact or burns, still less show who later developed traumatic brain injury.

The injuries behind the total

“Shell shock” once bundled several conditions together. Modern medicine tries to separate them. A high-energy explosion may cause traumatic brain injury, including concussion; fragments and violent movement add other damage. Low-level overpressure, studied mainly in troops who fire heavy weapons, has been followed by headaches, dizziness, slower thinking and problems with memory or concentration. The Pentagon cautions that exposure alone is not a diagnosis and that its long-term effects remain uncertain.3

Mild TBI may leave a missing stretch of time. After a blast, a person may be unconscious; awake but dazed or disoriented; or walking and talking without forming a continuous memory. The last state is post-traumatic amnesia. Standard clinical imaging will usually be normal. A documented gap is clinically meaningful, yet it also deprives doctors of part of the history needed to explain it.4

Researchers often find what a routine scan misses only after death or across groups. A small post-mortem study in 2016 found unusual scarring at tissue boundaries in five people with chronic blast exposure and three with acute exposure. In April 2026 a VA study of 107 combat veterans without a diagnosed military TBI found that greater lifetime blast exposure went with differences in resting brain networks, though not in brain volume. One study found pathology after death; the other found an association across a group. Neither gives a clinician an individual test.5

Evidence travelled slowly through the armed forces. Beginning in 2023 Dave Philipps of the New York Times traced similar symptoms through artillery crews and mortar teams. A defence laboratory found a distinctive pattern associated with blast exposure in the brains of eight Navy SEALs who had died by suicide; the results had not reached SEAL leaders. The reporting did not establish that blast caused the deaths. It showed how poorly the services were connecting exposure, symptoms and pathology. In 2024 the Pentagon ordered exposure monitoring, cognitive baselines and medical follow-up across the armed forces.6

Clinical rules make this murky event look crisp. Less than half an hour unconscious, or less than a day confused or amnesic, may still be classified as mild TBI. Such thresholds make triage, return-to-duty decisions and compensation possible. They do not make the event directly observable. A hole in someone's memory may reflect unconsciousness, confusion or a failure to lay down memories. No scan can replay the missing interval; nor is there an agreed mechanism of consciousness that tells a clinician exactly which signal to seek. The rules classify the evidence. They do not explain the mind.

Electric fields make this problem less abstract. Neural currents do more than carry messages through synapses. They also create weak extracellular fields which feed back on nearby cells. In experiments, fields as weak as those produced by the brain have altered spike timing and synchronised even neighbouring circuits without synaptic connections. In May 2026 researchers imposed a travelling field across the cortex: it shifted spike timing along 224 sites in one macaque, and reversing its direction produced different results in small human attention and memory experiments. Applied weak fields can therefore organise neural activity across a cortical span and alter behaviour. The experiment did not show that endogenous fields do the same.

Some theories go further, making the brain's electromagnetic field part of consciousness itself. That remains unproved, and the argument here does not require it. The narrower point is enough: a real feedback mechanism can help organise neural timing yet leave no mark on a structural scan. If blast disturbed that feedback, current tools would struggle to prove it. The currents which generate extracellular fields also generate the voltage and magnetic signals picked up by EEG and MEG. One altered recording could therefore fit damaged cells, changed collective timing, lost field feedback or all three. Blast cohorts show group-level changes in network co-ordination, but no study has isolated an ephaptic contribution in a patient. The clinical conclusion is not that electric fields explain blast injury. It is that a normal test cannot decide among mechanisms it was never designed to distinguish.78

Bureaucracy requires a yes or no; biology need not oblige. TBI and PTSD can arise from the same blast. Memory, concentration, sleep and mood problems overlap, and the VA says even neuropsychological tests may not distinguish them. The Pentagon now has policies for exposure logs, cognitive baselines and staged returns to duty; that does not ensure the record will exist or follow a veteran. A later VA assessment may depend on service files, witnesses and a memory containing a hole. PTSD should be treated when it is present. It should not become the residual label for every failure a scanner cannot see. A normal image is not a safety clearance.9

When the drone chooses

Autonomous weapons present the problem of mind in reverse. In the clinic, officials must infer what happened to a person's awareness. On the battlefield, they must decide what it means for software to “choose” when software need not be aware at all. Jamming supplies the practical push. A fibre-optic cable keeps a person in the loop; onboard software lets the weapon continue after its radio link disappears. Both reduce dependence on radio. Only one can transfer a decision.

Autonomy comes in degrees. A drone may fly a route, recognise an object or home in after a person clicks on it. The sharpest line is crossed only when software chooses the particular target and applies force. The person launching such a weapon may define an area or a target profile, but not the exact object, time and place of the attack.10

A machine need not be conscious or possess free will to cross that line. It needs only to turn sensor readings and a target profile into an attack. That is the nearer moral problem: software can acquire causal power without becoming a moral agent. It cannot answer for a mistake or understand why surrender, doubt or mercy should change a decision. Designers, commanders and states can. The farther automation reaches into judgments about civilians and proportionality, the more engineers must turn disputed ideas about what ought to be done into rules a machine can execute.

Marketing often blurs those degrees. On July 7th a Ukrainian company presented LITAVR as the world's first fully autonomous interceptor. It can take off, search and shadow another drone by itself, but waits for a person to give the final command. A different interceptor, which Ukraine said had passed combat tests in June, automates 95% of the engagement; there too an operator selects the target and orders the attack. Both fly autonomously. Neither delegates the final decision to attack.11

A more consequential claim surfaced in June. A Ukrainian drone executive said that in one trial in 2024 ten quadcopters flew without a control link, searched an assigned area and attacked targets selected by their own software. Reconnaissance later found two dead Russian soldiers and a damaged truck, he said. The account is retrospective and unverified. It nevertheless describes the thing that the looser marketing claims do not: a machine choosing the individual target. Ukraine says its policy is to keep people in final lethal decisions. That is evidence about doctrine, not proof of what every unit or contractor has tested. Nor does it cover Russia: analysis of recovered V2U drones, which lack communications equipment but carry processors and sensors, suggests that some can navigate, search and select targets onboard. Doctrine is not telemetry.12

The strategic change is not self-flight by itself. It is that one person can unleash many machines which need not call home. That eases the shortage of pilots, removes the control signal an enemy would jam and lets weapons wait. In May a Ukrainian company used a weather balloon to lift a Hornet strike drone above 8km. It released automatically and glided 42km, landing with 95% of its battery left. This was a range test, not an aerial ambush. Add persistent sensing and autonomous target selection, however, and such a contraption starts to resemble a mine in the sky.13

The rear is already less safe. On March 1st a single Iranian Shahed struck an operations centre at Port Shuaiba in Kuwait. It killed six American soldiers from a logistics command and wounded more than 30. Army Central says one general suffered a traumatic brain injury; some survivors say screening and treatment were slow. Across Iran's wider missile-and-drone campaign, satellite analysis by the Washington Post found damage to 217 structures and 11 pieces of equipment at 15 American sites, including aircraft, fuel systems, radar and communications gear.14

The strategic target is often not the fighter but the chain that makes it useful: the tanker, the radar, the runway, the satellite link and the people who run them. No American carrier is known to have been hit in the Iran war, and a moving carrier is much harder to find than a parked aircraft. Yet a carrier concentrates planes, fuel, sensors and people. Autonomy can make the terminal attack and saturation easier; it cannot by itself solve the ocean-wide search. In June America's government auditor said recent conflicts were pushing the navy to mix its large platforms with smaller, more numerous and more dispersed systems.15 Vulnerability does not make civilians lawful targets. The battlefield widens; the law does not.

The rules are moving more slowly. America's directive on autonomous weapons requires testing, senior review and “appropriate levels of human judgment”, without setting a universal minimum for human control. The UN secretary-general wants a binding agreement by the end of 2026. Government experts met in March and are due back in August, but their mandate does not prejudge whether any instrument will bind anyone.16

The price after purchase

Who controls the final approach is one question. Who pays for the following decades is another. The estimated $500 production cost of a basic Ukrainian FPV appears at once in a procurement budget. The cost of an injury arrives slowly, through treatment, rehabilitation, lost work, family care and disability support.

Nor is the cheap drone simply Chinese. Ukraine says that 95% of the drones it procures are made at home. Yet their price rests on a Chinese-centred civilian supply chain for controllers, motors, sensors, magnets and batteries. China makes almost 95% of permanent magnets and dominates many stages of battery production. Ukraine is replacing some inputs locally and in friendly countries, but usually at higher cost and smaller scale. The airframe may be Ukrainian; the industrial bargain beneath it is global and concentrated.17

A severe TBI can generate lifetime costs in the high six figures or millions in an American health system. That is the catastrophic end, not the typical mild concussion, but it shows how quickly the economics can invert. By November 2025 the Pentagon had recorded 533,519 first-time TBI diagnoses since 2000, four-fifths of them mild. Brown University's Costs of War project expects medical care and disability benefits for veterans of the post-9/11 wars and their dependants to reach $2.2trn–2.5trn by 2050. These are system totals, not a bill for drone strikes. They show how wartime injuries become public liabilities that last for decades.18

America's disability system can turn a qualifying condition into a continuing claim. VA compensation is monthly, tax-free and not means-tested. In 2025 the agency paid more than $195bn to 6.9m veterans and their families. When evidence is evenly balanced, American law gives the claimant the benefit of the doubt. Claims can still drag, especially when TBI is complex. But the design can turn uncertainty into a continuing public obligation.

Russia's federal wound-payment route is more front-loaded. A severe wound brings ₽3m; if it causes disability, the one-off total rises to ₽4m. Veterans can also claim pensions, housing help and rehabilitation: the system is not empty, but its chronic support sits in separate programmes. For wound compensation, a closed-head injury with concussion counts as “light” only if it prevents service for at least seven days and a doctor records it within three days. In the first half of 2026 military prosecutors took more than 2,800 enforcement measures to recover benefits and defend veterans' rights. Independent reporters documented at least 319 cases in which seriously wounded soldiers were returned, or threatened with return, before rehabilitation.

These are not like-for-like programmes, and the difference is not merely generosity. It is where the first line of risk falls when science cannot settle a claim. America's broad disability system can turn diagnostic uncertainty into a continuing public obligation. Russia's wound schedule demands prompt documentation and leaves the chronic dispute to separate pensions and services. A normal scan does not make that choice. Policy does.19

Ukraine's liability is nearer but scarcely easier to count. At the start of 2026 its register listed 1.42m combatants and 143,390 people disabled by war. The OECD says the country still lacks an integrated route from military care to civilian life. A separate international damage register now accepts claims for serious personal injury, but it is a compensation mechanism, not clinical surveillance. The war has a casualty tally and a claims ledger; it has no published population-level measure of brain injury caused by drones.20

The priorities are prosaic: record blast exposure and connect it with later symptoms; publish medically useful injury categories; protect medical evacuation; and state which decisions remain with an operator when guidance software takes over.

What is new is abundance. Armies already know how many drones they buy and roughly what each costs. They should know much more about the injuries those weapons leave behind—and exactly which decisions their software is allowed to make. The invoice for the drone arrives at once. The rest of the bill may take years.

■