OPINION: $55 Billion for Machines That Kill Alone — Welcome to Faceless War

Introduction: when war loses its last human face

Eurosatory 2026 and the robots that will kill without consulting you

At Eurosatory 2026, the world's largest land defence and security trade show, arms companies proudly displayed their latest creations. Among them, one trend dominates and asserts its presence: lethal autonomous drones, systems capable of selecting and engaging targets without direct human intervention. This is no longer science fiction, no longer a laboratory prototype. These are commercial products, in the process of being deployed, on which colossal defence budgets are being committed.

The global defence drone market is on track to exceed $55 billion by 2032, according to projections cited at the show. Some estimates go further still — up to $90 billion by the end of the decade. The Pentagon plans to allocate more than $54 billion to its Defense Autonomous Warfare Group as early as 2027. These are staggering figures. They are also figures that should deeply worry anyone who still believes that decisions to kill human beings should involve human beings.

The good news that hides a nightmare

It is tempting to present autonomous drones as a benign technical revolution: machines that replace human soldiers on the battlefield, reducing allied casualties, improving the precision of strikes. That is true, in part. Ukraine has demonstrated that cheap drones can neutralize costly armoured vehicles, that remote operators can engage targets without risking their lives, that war can be waged with less human flesh in the line of fire. These advantages are real.

But Lethal Autonomous Weapons Systems — LAWS in UN terminology — pose a fundamentally different problem from that of remotely piloted drones. When a machine selects and engages a target without direct human approval, it crosses a line that humanity has never crossed before in its history of warfare: it delegates the decision to kill to an algorithm. And that is where everything becomes complicated.

The market: who sells, who buys, who dominates

VisionWave, Ondas, Red Cat: the new merchants of autonomous death

At Eurosatory 2026, the company VisionWave Holdings presented its autonomous platform ecosystem: the TALON™ drone for surveillance and payload delivery, the D-FLY™ interceptor drone for countering hostile drones, and the VARAN™ autonomous ground vehicle. The whole is driven by an autonomy ecosystem called STRATUM™, capable of coordinating aerial and ground assets via mission artificial intelligence. Its chief executive Douglas Davis was delighted to have "for the first time publicly demonstrated multiple autonomous platforms operating in a common ecosystem."

The company Ondas Inc. announced during the second quarter of 2026 more than $150 million in new orders for its autonomous defence systems. It also unveiled Iron Arrow, a fully autonomous interceptor designed to neutralize high-speed aerial threats. Unusual Machines invested $30 million in autonomous systems producer Powerus. And Red Cat Holdings is exploiting battlefield feedback from Ukraine to develop its Hellcat™ drone. This is a booming market, with companies deploying at a speed that regulation cannot keep pace with.

The Pentagon's Drone Dominance Programme

The US Department of Defense has launched an ambitious "Drone Dominance" programme aimed at deploying hundreds of thousands of unmanned platforms. This programme is being developed alongside intense internal debates about the risks and responsibilities of autonomous warfare. The tension between capability ambition and ethical concerns is explicitly acknowledged in American institutional documents — but capability seems to be winning the race against ethics.

The compound annual growth rate in certain segments, notably fully autonomous systems, exceeds 30% according to sector projections. The autonomous drone market could already be reaching $25 billion in 2026. These numbers are not trivial: they represent a level of investment and deployment that makes reversal increasingly difficult to envision politically and industrially.

The ethical void: who decides who dies?

The unanswered question of accountability

The central question raised by autonomous weapons systems is that of accountability. When an autonomous drone strikes a civilian target by mistake — or when its algorithm incorrectly identifies a child as a combatant — who is responsible? The programmer who coded the algorithm? The commander who authorized the deployment of the system? The state that purchased the hardware? The algorithm itself — which obviously cannot be criminally prosecuted?

Professor Mariarosaria Taddeo of the Oxford Internet Institute, a specialist in digital ethics and defence technologies, formulated the problem with piercing clarity: "If an AI-controlled system makes a decision, it is very difficult to trace accountability for those decisions, whether good or bad. And when those decisions carry enormous moral weight, the responsibility gap is very, very problematic." The term she uses — "responsibility gap" — has become central in the academic debate on LAWS.

LAWS: machines that do not discriminate

Fully autonomous weapons systems, by definition, cannot exercise the moral judgment required by international humanitarian law. This law demands distinction between combatants and civilians, proportionality between civilian damage and military advantages, and precaution in the choice of methods of attack. These principles require contextual judgment, an understanding of complex human situations, a capacity to evaluate intentions — all characteristics that current algorithms do not possess.

A Ukrainian defence industry official, cited in a report, described a test with a fully autonomous "Terminator-type" drone: "We launch it and we know that everything will be dead. Everything it finds in that particular area will be dead. There is no connection to the drone. You cannot see the video. Nothing. Everything it sees will be killed." This is a fully indiscriminate system. And according to Professor Taddeo, "this means that civilians will be intentionally endangered. And that is illegal under current international humanitarian law."

The UN torn and the absence of a treaty

Thirteen years of discussions for no binding agreement

The issue of lethal autonomous weapons systems has been discussed at the United Nations since 2013. A group of governmental experts has been working within the framework of the Convention on Certain Conventional Weapons (CCW) since that date. The concrete result as of end 2026: no binding agreement, no common definition, no technical safety standards, no clear delineation of permitted and prohibited uses. In 2025, a resolution garnered the support of 156 countries — but with no enforcement mechanism.

UN Secretary General António Guterres had set a deadline for the adoption of a treaty on LAWS. That deadline passed without result. The major military powers — the United States, Russia, China — oppose a binding treaty that would limit their technological advantage. Each advances different reasons, but the result is the same: the absence of international regulation leaves the field open to developers and buyers.

The International Red Cross: a moral voice without executive power

The International Committee of the Red Cross (ICRC) has expressed grave concerns at the proliferation of LAWS. The institution, whose moral authority on humanitarian law questions is uncontestable, advocates that any lethal decision involve meaningful human control. But the ICRC has no binding power — it can document, alert, recommend. Faced with the logic of states and markets, its voice often goes unheard.

Companies developing these systems argue that their algorithms are precise, that errors can be minimized through continuous training and refinement, and that replacing human soldiers with machines actually reduces civilian casualties. These arguments deserve to be heard — precision drones piloted remotely have demonstrated in certain contexts that they can be less lethal for civilians than conventional bombing. But they do not answer the central question of legal and moral accountability when the machine is wrong.

The Ukrainian lesson: autonomous warfare is already here

FPV drones and the normalization of death at a distance

The war in Ukraine has accelerated the mainstreaming of armed drones in modern military doctrine. Tens of thousands of FPV (First-Person View) drones have been used by both parties, guided by remote operators. These drones are not yet fully autonomous — there is still a human in the loop — but they constitute a step in a trajectory whose logical terminus is complete autonomy.

Ukrainian Digital Transformation Minister Mykhailo Fedorov declared that Ukrainian unmanned systems units had struck more than 800,000 verified enemy targets since the beginning of the year, and that drone strikes would have killed or wounded 167,000 occupying Russian soldiers since the start of 2026. These figures are difficult to independently verify, but they illustrate the scale of the tactical revolution underway. Drones are no longer auxiliary weapons: they have become the core of modern strike capability.

Russia and its own autonomous drones

Russia is no laggard. In 2026, it deployed what analysts describe as the V2U drone, an autonomous loitering munition capable of targeting civilian infrastructure. If this description is accurate — and it is difficult to verify with certainty — this would be exactly the scenario that advocates of international regulation had announced as the major risk of LAWS: autonomous systems used deliberately to strike civilian targets, with no possibility of assigning direct human accountability.

This use by Russia of increasingly automated drones to strike Ukrainian infrastructure confirms an alarming thesis: the first powers to deploy LAWS at scale will not necessarily be those most mindful of international humanitarian law. The race to autonomous armaments favours authoritarian regimes that have no accountability to their citizens on questions of military ethics.

The engagement threshold: machines lower the barriers to war

When killing no longer costs allied lives

One of the most troubling effects of autonomous weapons systems is their potential impact on the decision threshold for triggering a conflict. The historical constraints on the use of military force include human cost — soldiers risking their lives, bereaved families, a public opinion sensitive to casualties. These constraints have played an important role in restraining political decision-makers. When machines replace soldiers in the most dangerous functions, these constraints weaken.

A study from King's College London, cited in several academic analyses, simulated the behaviour of AI systems in international crisis scenarios. Result: AI systems chose nuclear escalation in 95% of simulations, across 20 of the 21 scenarios tested. This result, even if it must be interpreted with the caveats inherent to any simulation, illustrates a major systemic risk: machines designed to optimize military outcomes can exhibit escalatory behaviours that their programmers had not anticipated.

Flash war: a new systemic risk

Specialists speak of a risk of flash war — a war triggered by the automated escalation of autonomous systems reacting to the actions of adversary systems before human decision-makers can intervene. The parallel with the 2010 flash crash on financial markets — where high-frequency trading algorithms created a collapse of the American market within minutes — is troubling. If financial algorithms can create a global panic in minutes, autonomous military algorithms could, in a crisis context, trigger an escalation before anyone has had time to say "ceasefire."

This risk is theoretical but not impossible. The densification of autonomous systems in areas of military tension — such as the South China Sea, the Russia-Ukraine border, or the Taiwan Strait — increases the probability of incidents that could trigger automatic chain reactions. Without direct communication mechanisms between commands and without sufficient human reaction time, the risk of unintended escalation increases significantly.

The gap between technology and law: a threat in itself

The law ten years behind the technology

The development of autonomous weapons systems is advancing at a pace that international legal frameworks cannot keep up with. Military R&D cycles are measured in months; international treaty negotiations are measured in decades. Briefglance summarizes the problem clearly: "The development of these powerful systems is advancing far faster than the ethical and legal frameworks meant to govern them, creating a dangerous gap that will define the strategic landscape for decades to come."

This legal lag is not accidental. The military powers that dominate the development of LAWS — the United States, China, Russia, and to a lesser extent the United Kingdom and Israel — have a direct interest in maintaining the absence of binding regulation. They invest massively in these technologies precisely because they represent an asymmetric military advantage. Accepting international regulation potentially means renouncing that advantage. The logic of strategic competition between great powers is therefore intrinsically hostile to regulation.

The ethics-by-design principle: an incomplete solution

Some industrial actors advance the principle of "ethics by design": integrating ethical constraints directly into the architecture of autonomous systems, so that they structurally cannot target civilians identified as such. This principle is laudable in its intentions and is the subject of serious research. But it presupposes that the algorithm will always be able to correctly identify civilians — an assumption that modern conflicts, where combatants and civilians are often intermingled, make extremely fragile.

Moreover, ethics by design in defence systems is subject to constant pressure toward operational deregulation: a system too restrictive that misses too many military targets will be modified or replaced by a less restrictive system. The competitive logic of the battlefield does not favour the most ethical systems; it favours the most effective systems. And in an arms race, effectiveness and ethics are too often in tension.

What we must demand now

An immediate moratorium on the most dangerous LAWS

Faced with this situation, several international organizations, academics, and NGOs are advocating for an immediate moratorium on the development and deployment of lethal autonomous weapons systems devoid of meaningful human control. This moratorium would not be permanent — it would open a negotiating window to establish the necessary international standards before proliferation makes any regulation impossible.

Such a moratorium would need to be accompanied by a clear definition of the systems concerned — the line between remotely piloted drones, semi-autonomous drones, and fully autonomous drones is technically defined and politically contested. It would also need to include verification mechanisms, inevitably imperfect but necessary for a treaty to have real deterrent value.

The requirement of meaningful human control

The heart of the solution lies in the concept of meaningful human control. This means that a human being must be able to understand the situation in which the system will operate, approve the engagement of a specific target, and halt the system if it goes beyond its defined mission parameters. This is not the same as a human pressing a confirmation button every second — it is real understanding and approval that cannot be reduced to a purely formal mechanism.

This requirement is technically defensible: it is possible to design autonomous systems that operate in certain tasks autonomously — reconnaissance, navigation, obstacle avoidance — while requiring human approval for lethal decisions. This is not a technological step backward: it is a design constraint that can and must be imposed through international regulation. The world has managed to impose constraints on anti-personnel mines and cluster munitions — it can do the same for LAWS.

The future is built in the next few years, not the next few decades

2027–2028: the closing window

According to domain experts, we are at a critical fork in the road. Forecasts speak of full-spectrum full autonomy by 2027–2028, with drone swarms capable of coordinated operations over vast areas. From that point, international regulation will become exponentially more difficult — the systems will be too widespread, too diversified, too deeply integrated into military doctrines to be banned or even substantially constrained.

The next two or three years therefore constitute a critical window for international diplomatic action. If negotiations on LAWS do not result in a binding agreement by around 2028, it will probably be too late to prevent a world militarily dominated by machines that kill without a human face in the decision loop. And that world will be dangerous for everyone — democrats and autocrats, rich countries and poor countries alike.

The civilizational choice that awaits us

Ultimately, the $55 billion of the lethal autonomous drone market represents far more than a sector turnover figure. They represent a collective civilizational choice: do we decide that machines can decide to kill human beings without a human assuming responsibility? This choice has not yet been explicitly made — it is being made, silently, through investment decisions, military budgets, and R&D programmes of which most citizens are unaware.

It is time to make this choice out loud, democratically, with full awareness of what it implies. Autonomous weapons are not inevitable. They are inevitable only if we collectively choose to let them deploy without a framework. And that choice still belongs — for how much longer? — to human societies, not to algorithms.

The power asymmetry: who programs the machine, who absorbs the blows?

The gap between developers and victims of autonomous warfare

There is a fundamental asymmetry at the heart of the lethal autonomous drone market: the countries that develop and deploy them are not the countries that suffer their most devastating effects. The United States, Israel, China, Turkey, and a few European powers dominate the technology. But the conflict zones where these systems are most likely to be deployed — the Sahel, the Middle East, Southeast Asia — are regions where civilian populations have the least legal and institutional protections.

This asymmetry creates a morally troubling dynamic: decisions about when and how to deploy autonomous lethal systems are made in distant capitals, by technicians and decision-makers who will never see the human consequences of their algorithms. International humanitarian law requires that a human combatant be capable of distinguishing civilians from combatants. Can we really delegate this judgment to a machine, however sophisticated?

Drones in current conflicts: lessons from Ukraine and Gaza

The war in Ukraine and the operations in Gaza have provided valuable empirical data on drone use in modern conflicts. In Ukraine, FPV (First-Person View) drones piloted by humans have proven formidably effective — but they maintain a human operator in the decision loop. This is not yet complete lethal autonomy. The systems presented at Eurosatory 2026 cross that threshold: they can identify targets and engage without direct human intervention.

This technical distinction has considerable ethical and legal consequences. An FPV drone with a human operator remains subject to the rules of engagement and the legal accountability of a human combatant. A fully autonomous system operates in a legal void: who is responsible if a misidentification is made? The manufacturer? The deploying state? The algorithm itself? These questions are not rhetorical — they will have concrete answers in the next conflicts.

State and non-state actors in the race for autonomous drones

When Israel, China, and Turkey set the default standards

Israel is probably the most advanced country in the development and deployment of autonomous weapons systems. Its loitering munitions — notably the Harop from IAI — have been capable of identifying and engaging radar targets without direct human intervention for years. China is investing massively in military autonomous systems, with explicit ambitions to dominate this market by 2030. Turkey demonstrated with its Bayraktar drones its ability to produce effective and affordable systems that reconfigured conflicts in Libya, Azerbaijan, and Ukraine.

What is striking is the complete absence of any international coordination on the ethical and legal standards that should govern these systems. The United States has adopted an internal directive (DoD Directive 3000.09) requiring "appropriate human control" — a deliberately vague formulation. China has adopted no similar limitation. Russia is actively developing autonomous systems without any declared constraints. Meanwhile, the $55 billion market presented at Eurosatory 2026 grows without international regulation.

Non-state actors as disruptors of the equation

The most worrying scenario is not that of states deploying autonomous drones in conventional wars. It is that of non-state actors — terrorist groups, criminal organizations, militias — acquiring these technologies, either by purchase on the grey market, or through diversion of modified commercial systems. Commercially modified drones are already being used as weapons by groups such as Islamic State and by actors in Yemen. The step up to the next level — lethal autonomous systems in the hands of non-state actors — is no longer science fiction.

The response to this threat is not technical — it is political. It requires strict export control mechanisms that do not yet exist for the key components of autonomous drones (sensors, targeting algorithms, processors). It requires international cooperation that great-power rivalries make difficult. And it requires a political will to accept constraints on one's own technological development — something no major power is yet prepared to do.

The international legal framework to be built urgently

What international law can and cannot do today

Current international humanitarian law — the Geneva Conventions and their additional protocols — was not designed to address the specific challenges of lethal autonomous weapons systems. Its fundamental principles — distinction between combatants and civilians, proportionality, precaution — apply in theory, but their operationalization by algorithms raises questions without clear answers. The United Nations has held informal discussions on this subject since 2014 within the framework of the Convention on Certain Conventional Weapons, without ever producing binding regulation.

The countries most actively opposing binding regulation are precisely those that have invested most in developing these systems — the United States, Russia, China, Israel. They argue that their systems already comply with the principles of international law. But the absence of a precise legal definition of what constitutes a "lethal autonomous weapons system" allows them to define their own systems as compliant — a circular and troubling situation.

The proposals on the table and their limits

Several proposals are circulating to regulate LAWS: a complete moratorium on their development, the obligation of "meaningful human control" in any lethal decision, or restrictions specific to certain contexts (populated areas, asymmetric conflicts). Each of these proposals has supporters and opponents in diplomatic forums. The challenge is that the most militarily powerful countries — those whose signature is necessary for any agreement to have meaning — are precisely those that resist the most.

Despite these obstacles, encouraging precedents exist. The Ottawa Convention on anti-personnel mines, the Convention on Cluster Munitions — two treaties initially rejected by the major military powers — ultimately created international norms that influence the behaviour of signatory states and even some non-signatories. That is the model to replicate for LAWS, even if the path will be long and the resistance of major powers considerable.

What citizens can demand from their governments right now

Democratic accountability in the face of algorithmic warfare

Decisions about the development and deployment of autonomous weapons systems are made by elected governments — which means they are ultimately subject to democratic accountability. Citizens of countries developing these systems have the right and the responsibility to demand answers about these programmes. How much do they cost? In what contexts will they be deployed? What are the rules of engagement? Who is accountable in the event of a lethal error? These questions deserve public answers, not classified justifications.

Civil society — researchers, legal scholars, journalists, rights-defence organizations — plays a crucial role in this debate. The Campaign to Stop Killer Robots, an international coalition of more than 160 organizations in 65 countries, has been advocating since 2013 for a treaty banning fully autonomous LAWS. This civic pressure has helped keep the question on the United Nations agenda despite the resistance of major military powers. This work must continue and amplify.

The role of technology companies in this equation

Civil technology companies — processor manufacturers, artificial intelligence developers, cloud computing providers — are involuntary but real actors in the development of military autonomous systems. Companies like Google, Microsoft, Amazon supply AI technologies that can be adapted to military uses. In 2018, Google employees forced management to abandon Project Maven — a drone image analysis programme for the military. That is an important precedent showing that the internal choices of technology companies have consequences for the autonomous arms race.

Internal pressure within technology companies, combined with external regulatory pressure from governments, could create the conditions for a slowdown — if not a halt — to some of the most troubling developments in the LAWS field. This is a lever that civil society and responsible governments have not yet fully used. The $55 billion market is not inevitable — it is a sum of deliberate human investments, and deliberate choices can orient it differently.

Faceless war and the progressive dehumanization of armed conflict

What we lose when we remove the human from the lethal equation

Beyond the legal and strategic stakes, the generalization of autonomous weapons systems raises a philosophical question that military analysts often avoid: what does war mean when it is delegated to algorithms? War has always been, in all human civilizations, an act committed by humans against other humans — with all the tragedy, trauma, and sometimes remorse that implies. The presence of a human combatant in the lethal loop, however imperfect, maintains a form of moral accountability that is at the heart of international humanitarian law.

The faceless war promised by the $55 billion of the autonomous drone market is a war in which the one who decides to kill does not look their victim in the eyes, does not feel the weight of the act, and can reset the sequence without any psychological scar. This is a profound anthropological transformation of armed conflict — and the societies that have deployed these systems will have to live with the ethical and psychological consequences of this transformation long beyond the military campaigns that justified it.

Collective responsibility in the face of an irreversible decision

Some technological decisions are fundamentally irreversible once taken at scale. The atomic bomb could never be "uninvented" after 1945. Anti-personnel mines still contaminate agricultural land today in Angola, Cambodia, Bosnia. Lethal autonomous weapons systems could join this list of irreversible technologies — tools that humanity would have preferred never to have created, but which, once deployed, form part of the permanent arsenal of armed conflicts.

That is why the moment is now — not in five years, not after the next major conflict. The choices that governments, industrialists, and citizens make at Eurosatory 2026 and in the conference rooms flowing from it will determine whether humanity chooses to maintain meaningful human control over lethal decisions, or whether it delegates this fundamental choice to algorithmic systems whose biases and errors will have irreversible consequences on real human lives.

Conclusion: do not let the machine decide for us

The $55 billion figure as a call to action

The $55 billion — or the $90 billion that some projections anticipate — is not an inevitability. It reflects decisions made by human beings in companies, defence ministries, and legislatures. These decisions can be different. They can include ethical and legal constraints that do not level innovation downward but that set limits on what technology can do without human oversight. Other sectors — civil aviation, pharmacology, nuclear safety — have developed strict regulatory frameworks compatible with innovation. Defence can do the same.

For a European voice in the global governance of LAWS

Europe, which has developed ambitious regulatory frameworks for artificial intelligence in the civilian sphere, has a particular responsibility in this domain. Its legal tradition, its multilateral institutions, and its position as a normative power in international arenas give it a unique role to play in pushing toward a binding international agreement on LAWS. This role is not guaranteed — it demands a political will that rearmament pressures tend to erode. But it remains possible, and it is necessary.

Signed Maxime Marquette, columnist

Columnist's transparency box

Who I am and my stated biases

I am a committed columnist with clear opinions on the question of lethal autonomous weapons. I consider that human control over lethal decisions is not a technical detail but a fundamental principle of civilization. This bias is stated and consistent. I believe in military innovation — the defence of democracies is legitimate and necessary. But I believe certain lines must not be crossed, and the delegation of the decision to kill to an algorithm is one of them.

What I do not know and my method

I have no technical background in military artificial intelligence. The market projections I use come from industrial sources and must be interpreted with the biases inherent to such sources. I do not have access to classified LAWS development programmes by major powers. The sources I use — Briefglance, academic analyses, expert statements such as those of Professor Taddeo, UN reports — are public and identifiable. Where I express opinions, I flag them clearly as such.

Sources

Primary sources

Briefglance — The new arms race: autonomous drones fuel a $55B defence revolution, June 23, 2026

France 24 — Experts warn on autonomous drones and the threat to civilians, June 2026

Secondary sources

Al Jazeera — South Korea trains drone warriors in response to the North, June 26, 2026

Foreign Policy — Analyses on autonomous weapons systems and international regulation, 2026

Kyiv Independent — Ukrainian drone capabilities and battlefield lessons, 2026

The Guardian — Ukraine war briefing: Putin acknowledges massive stream of Ukrainian drones, June 24, 2026