OPEN LETTER: To the Generals Who Still Thought Drone Swarms Were Science Fiction

Introduction: Salisbury Plain, June 2026 — Drones Fly in Packs, War Changes

An experiment on the English plain that redefines doctrine

To you, senior officers, strategic planners, defense procurement bureaucrats who spent years reading reports on autonomous drone swarms and telling yourselves it is ten years away. Take note: it is now. On Salisbury Plain, in England, Australian, American, and British soldiers just completed three weeks of intensive experimentation as part of the Army Warfighting Experiment 2026. The results are no longer theoretical. Autonomous drone swarms flew over English woodland, identified targets in real time, shared data between platforms from three different nations, and refined their artificial intelligence models mission after mission.

The exercise, conducted under the framework of AUKUS Pillar II — the advanced capabilities strand of the partnership between Australia, the United Kingdom, and the United States — was not a public relations exercise. It was operational experimentation under « demanding conditions » with « near-constant rain ». Soldiers, engineers, and scientists from the three nations worked together on a central problem: how to ensure that drones of different nationalities, running different software, communicate in near-real time and operate as a coherent swarm in a contested environment? The answer, after three weeks, is: it is possible, and it is improving rapidly.

From two to six drones, from demonstration to operation

What stands out in the exercise account published by Australian Defence on June 22, 2026 is the progression across the three weeks. The activity began with swarms of two drones and progressively scaled up to swarms of six aircraft. That is not a large number — discussions about mass swarms evoke hundreds or even thousands of units. But the progression illustrates the fundamental principle of this technology: it is a system whose sophistication scales with experience and iteration.

The drones operated over assigned areas of interest with predefined mission profiles. They used swarm formation software to divide the sector, sweep it, and send images to the end-user's device. The AI model was continuously refined through repeated identification and confirmation cycles — the more a drone recognized targets in the English woodland, the better it became. A shared command hub enabled the near-real-time exchange of mission data and sensor information between AUKUS partners. For the first time, distinct national platforms shared AI models under common interoperability standards.

What the AUKUS Exercise Really Proved

Interoperability: the real technical challenge solved

The greatest technical achievement of the Army Warfighting Experiment 2026 is not that drones flew in a swarm — military experimenters have been doing that for years. The real achievement is national interoperability: Australian, British, and American drones running different software, developed by different companies to different standards, successfully shared data in real time and operated as a unified swarm. That is the problem that had blocked every previous exercise.

Engineer Hing-Wah Kwok from the Australian Defence Science and Technology Group (DSTG) articulated the breakthrough precisely: « Being able to train and retrain new models in the field and share them rapidly across AUKUS partners under common standards is a true force multiplier. » That sentence deserves reflection. It means an AI model trained on a specific theater of operations — say, recognizing Russian military vehicles in a Ukrainian forest — can be shared and deployed on a partner nation's platforms within hours. That is a revolutionary capability.

AI-assisted recognition in woodland: a lesson for Ukraine

One of the most complex scenarios tested during the exercise was target recognition in dense woodland. Identifying military vehicles or infantry positions under tree cover is one of the classic problems of land warfare — artificial vision systems struggle to distinguish thermal and optical signatures in vegetation. The AUKUS exercise proved that swarming drones, combined with real-time-refined AI models, can progressively overcome this challenge. The improvement in the ability to « correctly identify targets in difficult wooded terrain » is explicitly cited as an exercise result.

For Ukraine, this lesson is directly applicable. The wooded areas of the Donbas and the Russian defensive lines concealed in vegetation are among the most challenging environments for current Ukrainian drones. Swarms of drones capable of recognizing and designating targets in those environments would represent a significant tactical advance. The fact that the exercise was conducted in English woodland in the rain — conditions comparable to Ukrainian operational zones — is probably not a coincidence.

The Transition from Science Fiction to Operation: What Changed?

Ukraine as an accelerating laboratory

The transition of drone swarms from operational concept to near-deployable system has accelerated spectacularly thanks to Ukraine. Ukrainian forces conducted, according to the Modern War Institute at West Point, what appears to be the first fully unmanned combined arms operation in military history, in December 2024 near Lyptsi, involving unmanned ground vehicles and FPV drones with no infantry participation. While that is not a « swarm » in the strict technical sense — many platforms were individually controlled — it is a demonstration that autonomous and semi-autonomous platforms can execute complex tactical tasks without soldiers on the ground.

In parallel, companies like Helsing AI are delivering AI drones to Ukraine, and American firms like Anduril tested their autonomous drones in Ukraine before commercializing them. Ukraine is the only operational theater in the world where autonomous systems are being tested under conditions of intensive real combat, against an adversary capable of sophisticated electronic countermeasures. The data gathered in that real-time laboratory accelerates the development of systems like those tested at Salisbury Plain in a way that no simulated exercise could match.

The Replicator program: the United States appropriates the Ukrainian lesson

The Pentagon's Replicator program is the institutional manifestation of that awakening. Launched in 2023, it aims to deploy thousands of autonomous platforms rapidly, explicitly drawing lessons from the Ukrainian experience. Its objective is to counter China's quantitative advantage in a hypothetical Pacific confrontation by saturating enemy defenses with swarms of low-cost systems. That is the direct application of swarm doctrine to strategic competition with Beijing — and that is precisely why AUKUS exercises take place in this multi-partner format.

Turkey also conducted in January 2026 the first real strike of an armed drone swarm, with 20 autonomous KARGU loitering munitions coordinated, striking targets simultaneously in a real-fire environment with live warheads. That is the first real-conditions validation — by a nation other than the United States or China — of an autonomous swarm doctrine with kinetic capability. The world is changing, fast.

The Challenges of Defensive Saturation

The command question: who decides when an AI recognizes a target?

One of the most complex problems raised by autonomous swarms is that of command and control. In the Salisbury exercise, human operators « assigned areas of interest and mission profiles ». Drones then executed their tasks autonomously. But at what point can a drone decide to engage a target without explicit human confirmation? The current rule in most AUKUS systems is « human-on-the-loop »: a human supervises and can interrupt, but is not necessarily in the decision loop for every shot.

That is a crucial doctrinal point. The larger the number of drones in the swarm, the less individual human control is practicable. At six drones, one operator can supervise. At 600, it is impossible. Armies that deploy large-scale swarms will have to solve the problem of distributed command — and that resolution will have significant ethical and legal implications for the distinction between combatants and non-combatants, for accountability over strikes, and for international laws of armed conflict.

Defensive saturation: the central strategic challenge

The concept of defensive saturation is at the heart of swarm doctrine. The idea is simple: send simultaneously enough platforms to exceed the adversary's interception capacity. Ukraine applied it against Russia with hundreds of drones over Moscow. Turkey demonstrated it with 20 KARGU. AUKUS is experimenting with six drones in English woodland. The central question is: what is the saturation threshold of an advanced air defense system? No one knows with certainty — and that is precisely why exercises like Salisbury are needed to build real operational doctrine.

Defense against swarms is also under development. Systems like the Ukrainian SkyFall P-1 SunLong — an AI-guided interceptor to engage incoming drones — represent the emerging countermeasure. The swarm war will also be an anti-swarm war. And the nations that advance fastest on both dimensions — offense and defense — will hold a significant strategic advantage in future conflicts.

Swarm Doctrine in Practice: What the AUKUS Exercise Really Showed

Salisbury Plain, June 2026: six drones in coordinated formation

On June 22, 2026, on Salisbury Plain in England, Australian, British, and American teams conducted the first concrete combat exercise testing drone swarm tactics under the AUKUS partnership. This was not a simulation — it was a real-terrain demonstration involving six autonomous drones coordinating their reconnaissance and simulated engagement actions. According to Colonel Penley, who oversaw the tests on behalf of the Australian Army, this exercise demonstrated command and control interoperability between the three nations' systems — a first that opens real operational possibilities.

What the exercise concretely proved: drones belonging to different national systems can operate under common coordination, share data in real time, and collectively adapt to an evolving ground situation. That is the functional definition of a swarm. Six drones is modest. But the logic that allows six to function together can in theory apply to sixty or six hundred — provided command systems and bandwidth keep up. The limit is no longer technological. It is doctrinal and logistical.

From exercise to operation: the conditions for scaling up

The transition from a six-drone exercise swarm to an operational swarm of several hundred in real combat is not automatic. Several conditions must be met. First, command and control systems must be able to handle the complexity: coordinating hundreds of autonomous entities in motion, with changing objectives and sometimes degraded communications, requires algorithms of considerable sophistication. Second, radio bandwidth must be sufficient in an electronic warfare environment where the adversary will attempt to jam communications. Third, the logistics chain — loading, deployment, recovery or replacement of drones — must be designed to operate at scale.

These challenges are known to military planners. They are not insurmountable. The war in Ukraine showed that far lower levels of coordination than those in the AUKUS exercise were already producing significant strategic results. The Salisbury Plain exercise therefore represents a genuine qualitative leap — an advance toward the operational sophistication that Western armies are seeking to achieve before their adversaries.

Real Precedents: KARGU in Libya and the Lessons of the First Lethal Swarm

January 2026 in Libya: the first documented use of an autonomous swarm

Before talking about the future, let us look at the recent past. In 2020-2021, during the conflict in Libya, KARGU drones — Turkish-made, Kargu-2 model — were used in what analysts consider one of the first documented uses of autonomous loitering munitions in real combat conditions. These munitions targeted fighters autonomously, without real-time communication with a human operator. The event was confirmed by a UN expert report. That precedent is fundamental for understanding what the AUKUS swarm represents in the evolution of autonomous weapons.

This sequence is not anecdotal. It marks the transition from a laboratory concept to a battlefield reality. KARGU drones are relatively accessible systems. If limited-resource actors can deploy them in semi-autonomous mode, the world's most advanced armies — China, Russia, the United States — have exponentially superior capabilities. The AUKUS exercise at Salisbury Plain is not a futuristic experiment: it responds to a reality already present on the world's battlefields.

The ethical and legal implications of autonomous swarms

The use of autonomous drones in combat raises ethical and legal questions that the AUKUS agreement's signatory governments can no longer ignore. Who is responsible when an autonomous swarm kills a civilian by mistake? The algorithm's programmer? The commander who launched the operation? The state that deployed the system? International humanitarian law requires that a human being retain « meaningful control » over lethal engagement decisions. But what does that mean in a swarm of 500 drones operating in milliseconds? Colonel Penley himself acknowledged that military AI ethics questions must be addressed in parallel with capability development.

This is not an academic debate. It is a regulatory emergency. The nations that develop these systems first have a responsibility to set clear rules of engagement before use outpaces doctrine. The AUKUS exercise is an opportunity to establish those rules at a multilateral level — an international norm on the autonomy of weapons systems, before less scrupulous actors normalize uncontrolled use.

The Challenges of Defensive Saturation and Possible Responses

How do you defend against a swarm?

The military question posed by drone swarms is as daunting for defense as for offense. Against a single ballistic missile, a THAAD or PAC-3 system can be effective. Against a swarm of 200 drones arriving simultaneously from different directions, with variable trajectories and minimal radar signatures, conventional defensive systems show their limits. The cost of interception by SAM missiles can be exponentially higher than the cost of the attacking drones. That is an economic asymmetry that NATO's adversaries deliberately exploit.

Defensive responses under development include: high-energy lasers (capable of engaging targets at the speed of light at very low unit cost), electromagnetic railguns, autonomous counter-drones (defensive swarms against offensive swarms), and advanced electronic jamming systems. But these technologies are still in the development phase for large-scale deployment. In the interim, the capability gap between what offensive swarms can do and what defenses can counter is real — and AUKUS exercises aim to close that gap as quickly as possible.

Defensive saturation as strategy: Ukraine as laboratory

Ukraine is the world's laboratory for defensive saturation. Its forces have launched massive waves of drones — 190 to 200 aircraft in a single night toward Moscow on June 18, 2026 — precisely to saturate Russian air defense layers and allow some drones to get through. Moscow intercepted 194 drones that evening, according to Mayor Sobyanin — but several still reached the Kapotnya refinery. That is exactly the saturation logic: even with a high interception rate, if you send enough units, some get through.

The AUKUS exercise fits into that thinking. The partners are developing offensive saturation capability — to be able to use it, but also to understand how to defend against it. The best way to design a defense against a swarm is to build a swarm. That dual reasoning — offensive and defensive — is at the heart of the strategic value of the AUKUS drone program.

What AUKUS Means for Deterrence in the Indo-Pacific Against China

China observes and accelerates its own drone programs

The Salisbury Plain exercise was not designed in a vacuum. It responds to a reality: China already has advanced drone swarm programs. The CH-901s, the systems developed by CASIC (China Aerospace Science and Industry Corporation), and the demonstrations of 1,000-drone swarms during official Chinese spectacles are not mere show. They demonstrate industrial and algorithmic mastery that AUKUS takes very seriously. The technological competition in the Indo-Pacific is also being fought on that terrain.

The AUKUS response is deliberately multinational: by combining Australian industrial capabilities, British operational experience, and American technological leadership, the three partners create a collective advantage that none could achieve alone. That logic of capability pooling is exactly what military decision-makers in Canberra, London, and Washington wanted to demonstrate in June 2026. And it is directly relevant to stability in the Indo-Pacific.

The implications for Taiwan and deterrence

The question of Taiwan is implicit in every AUKUS military exercise. Military planners thinking through a Taiwan invasion scenario know that China would massively use drones to saturate Taiwanese, American, and allied defenses during an initial phase. The ability to counter such a swarm — and to use one's own offensive swarms to strike Chinese military targets — is a central element of any credible deterrence strategy in the Taiwan Strait.

The Salisbury Plain exercise is therefore, indirectly, a deterrence statement addressed to Beijing: AUKUS is developing real, not theoretical, swarm capabilities. Whether that actually tempers Chinese military calculations is another question. But deterrence always begins by demonstrating a capability. The exercise of June 22, 2026 is precisely that demonstration.

Conclusion: Act Now, on Doctrine and Ethics

What democracies must do

The AUKUS exercise of June 2026 is a clear signal: autonomous drone swarms are leaving the laboratory to enter operational military doctrine. Western democracies — which currently hold the technological advantage in this domain — have a window of opportunity to shape both capabilities and rules of engagement before their adversaries deploy similar systems with zero ethical constraints. That window is narrow. Colonel Penley said it: autonomous systems are entering the core of land operations. Action on doctrine, training, interoperability, and rules of engagement must happen now.

China is watching, and taking notes

China has developed its own drone swarms and tested them at scale in exercises. The technology is proliferating. Every week that passes without clear doctrine on the employment of autonomous swarms in the law of armed conflict is a week during which the technological advantage of democracies erodes. Salisbury Plain is not a laboratory exercise — it is the prefiguration of a military revolution. And like all military revolutions, it advantages those who prepare for it earliest and most seriously.

Signed Maxime Marquette, columnist

Columnist's transparency box

My sources and my limits

I am Maxime Marquette. I draw on the official Australian Defence statement of June 22, 2026, on academic analyses from the Modern War Institute, and on open information about previous AUKUS exercises. I am not a military engineer and my analysis of the technical implications of AI interoperability remains at a conceptual level.

Acknowledged uncertainties

I do not know how these systems actually perform in intense electronic jamming environments, such as those Russia deploys in Ukraine. The conditions at Salisbury Plain, while meteorologically challenging, did not include active adversarial countermeasures. The transition to real combat conditions remains a challenge not fully resolved.

Sources

Primary sources

Australian Department of Defence — AUKUS Partners Test Drone Swarming Tactics — June 22, 2026

UK Defence Journal — AUKUS Swarm Drone Tactics Tested at Major Army Warfare Experiment — June 2026

Militarnyi — Drone warfare context Ukraine — June 2026

Secondary sources

Defense News — AUKUS trio conducts drone swarm tests in UK — June 22, 2026

Foreign Policy — Autonomous weapons regulation debates — June 2026

Modern War Institute — Battlefield Drones and the Autonomous Arms Race — January 2025