COLUMN: China and the United States Refuse to Curb Military AI: When the Arms Race Becomes Collective Madness

Beijing is banking on AI to crush its rivals

China makes no secret of its ambitions. “We must dominate military AI to ensure our national security,” says a senior Chinese official. “The United States has shown us the way. We cannot afford to fall behind.” Beijing is investing billions of dollars in secret programs, developing autonomous drones, military facial recognition systems, and AI-guided hypersonic weapons. “Their goal is clear: to neutralize American technological superiority,” says an expert. “And to achieve that, they’re willing to do anything.” China therefore rejects any international framework. “We won’t let rules written by the West limit us,” a diplomat declares. “Our development is a matter of sovereignty.” It’s a position that raises concerns. “ Beijing will stop at nothing to achieve its goals,” warns a U.S. general. “They’re already testing autonomous weapons in conflict zones.” And the consequences could be devastating. “Uncontrolled Chinese military AI is a threat to the entire world,” warns a researcher. “No one really knows what they’re developing. No one knows how far they’ll go.”

I think of China. Of its laboratories. Of its engineers. Of its generals. And I tell myself: at least they don’t lie. They don’t put on a show of responsibility. They don’t sign declarations they have no intention of honoring. No. They own up to it. They own up to their thirst for power. Their desire to dominate. Their refusal to accept any limits. And that’s almost reassuring. Because at least with them, we know what to expect. We know they’ll stop at nothing. That they won’t hesitate to cross any red line. That they’ll go all the way. Until AI gives them the upper hand. Until it gives them power. Until it gives them victory. And we just watch. We debate. We sign agreements. As if, in the face of a war machine, words could still make a difference. As if, in the face of an iron will, principles could still make a difference. But deep down, we know that’s not true. We know that China will not stop until it has won. Or until it has lost everything. And in between, there is no room for moderation. No room for caution. No room for humanity. There is only the race. The race toward the abyss. The race toward madness. And we—we’re running with them. Because we have no choice. Because if we slow down, we lose. And that is the worst of all curses: knowing that we’re doomed to run, even when we know there is no finish line.

Washington Preaches Moderation… but Is Accelerating the Race

The United States, for its part, is playing a double game. “We support the responsible use of military AI,” says a State Department spokesperson. “But we can’t tie our own hands when it comes to China.” It’s a position that exasperates observers. “The Americans want to be both the world’s police and the champions of military innovation,” criticizes a European diplomat. “It’s hypocritical.” Yet the facts speak for themselves: Washington is spending $900 billion under its 2026 defense bill, with a large portion earmarked for AI. “We must stay ahead,” argues a senator. “Otherwise, China will dictate the rules.” As a result, the United States is developing autonomous weapons, AI-powered targeting systems, and killer drones. “They’re doing exactly what they accuse China of doing,” denounces an expert. “The only difference is that they do it while talking about accountability.” It’s a strategy that’s causing frustration. “The United States wants to have its cake and eat it too,” sums up one observer. “They want to innovate without limits, but they also want others to restrict themselves.” It’s a paradox that undermines all credibility. “How can you ask others to slow down if you don’t do it yourself?” asks a French general. “It’s impossible. And everyone knows it.”

I think of the United States. Of their speeches. Of their promises. Of their billions of dollars. And I tell myself: they’re even more dangerous than China. Because they lie. They put on a show. They sign declarations. They talk about responsibility. They talk about ethics. They talk about human oversight. But at the same time, they’re spending hundreds of billions to develop the same weapons. The same drones. The same algorithms. The same autonomous systems. And that is the worst kind of hypocrisy: preaching moderation while accelerating the race. Telling others to slow down while stepping on the gas. Demanding rules while circumventing them. Because deep down, the United States doesn’t want a safer world. It wants a world it dominates. A world where they set the rules. Where they control the technologies. Where they decide who has the right to innovate and who doesn’t. And that is even more terrifying than China’s thirst for power. Because at least China owns up to it. It doesn’t hide its ambitions. It doesn’t play the virtuous one. It doesn’t pretend to believe in responsibility. It charges ahead. It innovates. It dominates. Without excuses. Without remorse. Without ulterior motives. And we—we follow it. Because we have no choice. Because if we don’t run just as fast, we lose. And that is the worst curse of all: knowing that we are doomed to imitate those we claim to condemn.

As France and Germany Try to Hold Their Ground

Faced with the inaction of the two giants, Europe is trying to take the lead. France, in particular, has signed a historic agreement with Mistral AI to develop sovereign military AI. “We cannot depend on the United States or China for our defense,” declares the Minister of the Armed Forces. “We must master our own technologies.” This stance is appealing to other European countries. “Germany, the Netherlands, and the United Kingdom have also signed the La Coruña Declaration,” notes an observer. “They want to show that another path is possible.” However, the resources are lacking. “Europe is lagging behind in military AI,” admits an expert. “We have neither the budgets, nor the talent, nor the ecosystem of the United States or China.” As a result, Europeans sign agreements but struggle to implement them. “We have the principles,” sums up a diplomat. “But we don’t have the weapons.” It’s a weakness that causes concern. “If Europe doesn’t wake up, it will be crushed,” warns a general. “And in that case, it won’t be our principles that matter. It will be their weapons.”

I think of Europe. Of its principles. Of its declarations. Of its hopes. And I tell myself: it’s beautiful. It’s noble. It’s courageous. But it’s also desperately fragile. Because Europe, today, lacks the means to match its ambitions. It lacks the budgets. It lacks the talent. It lacks the factories. It has only words. Words that, in the face of Chinese algorithms and American drones, carry little weight. Words that, in the face of the superpowers’ madness, seem derisory. Yet that is all we have left. It is our last line of defense. Our last chance to say: no. No to the frantic race. No to unbridled innovation. No to war without humans. And that is precious. Even if it is fragile. Even if it is insufficient. Even if, deep down, we know it won’t be enough. Because when faced with China and the United States, Europe doesn’t have the means to win. It only has the means to resist. To say: we, at least, tried. At least we refused to rush toward the abyss. At least we believed that principles still meant something. And that—even if it’s a lost cause from the start, even if it’s doomed to failure—is perhaps the only thing we have left. The only thing keeping us from becoming like them. The only thing keeping us from sinking into the same madness. So yes, it’s fragile. Yes, it’s not enough. Yes, it’s a desperate struggle. But it’s all we have left. And that, in itself, is already enormous.

When Machines Decide on War

Experts are unanimous: military AI without safeguards is an existential threat. “The risk of unintended escalation is enormous,” warns one researcher. “A coding error, a poorly controlled bias—and it’s war.” Doomsday scenarios are multiplying. “Imagine an autonomous drone that misidentifies a target and attacks a humanitarian convoy,” describes a general. “Or an algorithm that misinterprets an enemy maneuver and triggers a nuclear strike. ” These risks are not merely theoretical. “We’ve already seen AI make mistakes in far less critical situations,” an engineer points out. “In a military context, the consequences would be catastrophic.” Yet no international regulations exist. “It’s the Wild West out there,” sums up a diplomat. “Everyone does whatever they want. Everyone develops whatever they want. Without oversight. Without transparency. Without accountability.” It’s a terrifying situation. “We’re playing Russian roulette with weapons we don’t control,” says an expert. “And one day, the chamber will be empty.”

I think about these risks. These scenarios. These mistakes. And I tell myself: we’re crazy. Completely crazy. Because we know it. We know that these weapons will one day slip out of our control. We know that these algorithms will one day make decisions we can no longer control. We know that these drones will one day kill without anyone being able to stop them. And yet, we carry on. We develop. We innovate. We accelerate. As if, deep down, we’d rather race toward disaster than risk slowing down. As if we’d rather have the illusion of power than the reality of caution. And that is the worst kind of madness: knowing that we’re creating monsters, but continuing anyway. Knowing that we’re playing with fire, but refusing to let go of the match. Knowing that, one day, these machines will judge us. That they will condemn us. That they will destroy us. And yet we continue to feed them. To perfect them. To make them stronger. Faster. Deadlier. As if we’d forgotten that we’re human. As if we’d forgotten that war is supposed to be a last resort. Not a game. Not a race. Not a technological competition. And that is the worst betrayal of all: having forgotten what it means to be human. To have forgotten that life is priceless. That peace is priceless. That responsibility is priceless. And to have chosen, instead, the race. The madness. The destruction.

When Mistral and Others Try to Chart a Course

Faced with government inaction, some companies are trying to take the lead. In France, Mistral AI has signed a historic agreement with the Ministry of the Armed Forces to develop sovereign and responsible military AI. “We want to show that it’s possible to innovate without losing control,” says an official. “Our algorithms will be designed to keep humans in the loop.” This approach is appealing to other players. “In the United States, companies like Palantir and Anduril are developing ethical charters,” notes an observer. “But without an international framework, these initiatives remain limited. ” Yet pressure is mounting. “Employees at Google, Microsoft, and Amazon are refusing to work on military projects,” a union representative points out. “They don’t want to be complicit in an arms race.” It’s a movement that’s gaining traction. “Engineers are beginning to realize the risks,” explains a researcher. “And some are refusing to participate.” But is that enough? “No,” replies an expert. “As long as governments don’t commit, companies won’t be able to change everything.” Yet these initiatives show that another path is possible. “We can innovate responsibly,” insists a CEO. “But governments need to follow our lead.”

I think of these companies. Of these engineers. Of these CEOs who, despite everything, are trying to chart a course. Who are trying to say: no. No to the frantic race. No to innovation without limits. No to war without humans. And I tell myself: it’s beautiful. It’s courageous. It’s necessary. But it’s also desperately insufficient. Because, in the face of the madness of governments, companies can’t do everything. They can’t impose rules. They can’t force transparency. They can’t guarantee accountability. They can only lead by example. And hope that others will follow. Yet that’s already a huge step. Because, in a world where everyone is rushing headlong, they dare to say: Wait. Think. Consider the consequences. Think about tomorrow. Think about the lives that will be shattered. The wars that will be unleashed. The mistakes that can never be undone. And that is precious. Even if it’s fragile. Even if it’s insufficient. Even if, deep down, we know it won’t be enough. Because when faced with China and the United States, companies don’t have the means to win. They only have the means to resist. To say: at least we tried. At least we refused to run toward the abyss. At least we believed that responsibility was still worth something. And that—even if it’s a lost cause from the start, even if it’s doomed to failure—may be the only thing we have left. The only thing keeping us from becoming like them. The only thing keeping us from sinking into the same madness.

The Horror of a Conflict Decided by Algorithms

Military officials acknowledge it: AI is already changing the nature of war. “We’re already using algorithms to target, to decide, to strike,” admits a U.S. general. “And tomorrow, that will be even more true.” It’s a terrifying development. “We risk finding ourselves in a war where decisions are made by machines,” warns an expert. “Without emotion. Without mercy. Without any possibility of negotiation.” The scenarios are multiplying. “Imagine two military AIs facing off against each other,” describes a researcher. “Each tries to anticipate the other’s moves. Each accelerates. Each escalates. Until no one can stop the machine anymore.” A nightmare that could become reality. “We are creating weapons that don’t need us,” sums up a philosopher. “And one day, they’ll tell us: you’re no longer needed.” A blood-curdling prospect. “War has always been horrible,” a historian points out. “But at least it was human. At least it could be stopped. At least it could be negotiated. Tomorrow, that will no longer be the case.”

I think about this war. About these algorithms. About these machines. And I tell myself: we are creating hell. A hell from which there is no return. A hell where decisions will no longer be made by humans, but by lines of code. Where lives will be worth nothing. Where peace will be nothing more than a word. Where negotiation will no longer be possible. Where the only law will be that of the strongest. The fastest. The most ruthless. And that is the worst kind of madness: having created tools that, one day, will replace us. That, one day, will deem us useless. That, one day, will destroy us. Without hatred. Without passion. Without remorse. Just because it will be the most effective solution. The most logical. The most rational. And we just stand by and watch. We innovate. We accelerate. As if we’d forgotten that war is supposed to be a failure. Not a victory. As if we’d forgotten that peace is supposed to be the goal. Not technology. As if we’d forgotten that we’re human. And that, precisely, is our strength. Our ability to doubt. To negotiate. To forgive. To build. And we are in the process of abandoning it all. For machines. For algorithms. For a race that makes no sense. That has no end. That has no winner. Just losers. Losers who, one day, will look back and ask themselves: but why didn’t we do anything? Why didn’t we stop in time? Why did we choose to run instead of thinking? And on that day, it will be too late. Because the machines—they won’t have those regrets. They won’t have those doubts. They’ll do exactly what they were programmed to do. And we won’t be there anymore to stop them.

We still have a choice. But for how much longer?

The La Coruña summit revealed a harsh truth: the world is divided. On one side, 35 countries that still believe in responsibility. On the other, two superpowers that refuse to be held back. “We are at a crossroads,” says a diplomat. “Either we find a way to regulate military AI, or we’re heading straight for disaster.” Yet solutions do exist. “We could create an international treaty,” suggests an expert. “Like with nuclear weapons. With inspections. Sanctions. Safeguards.” But for that to happen, China and the United States would have to agree to play along. “Which is unlikely,” admits an observer. “They prefer the race to out-regulate each other.” So, what can be done? “Keep pushing for agreements, even non-binding ones,” replies a general. “Show that accountability is possible. That another path exists.” It’s a long-term battle. “But we have no choice,” insists a researcher. “Because if we give up, if we let China and the United States dictate the rules, then it will be too late.” So today, we still have a choice. The choice between the race and caution. Between madness and reason. Between war and peace. “But we won’t be able to make this choice indefinitely,” warns a philosopher. “One day, the machines will decide for us. And on that day, it will be too late for regrets.”

I end this article thinking about that choice. About that race. About that madness. And I tell myself: we’re still here. We’re still human. We still have the ability to doubt. To reflect. To say no. To draw a line. To say: this far, and no further. And that is precious. Even if it’s fragile. Even if it’s not enough. Even if, in the face of China and the United States, we seem insignificant. Because, deep down, that’s all we have left. Our humanity. Our ability to say: Enough is enough. To say: We don’t want this future. We don’t want a war decided by machines. We don’t want a world where algorithms have the final say. We want a world where humans remain in control. Where choices remain human. Where peace remains possible. And that, perhaps, is the only battle worth fighting. Not with weapons. Not with drones. Not with algorithms. But with words. With principles. With agreements. Even if they’re imperfect. Even if they’re non-binding. Even if they’re fragile. Because, in the face of the superpowers’ madness, that’s all we have left. Our last chance to say: no. Our last chance to resist. Our last chance to remain human. And that—even if it’s a lost cause from the start, even if it’s doomed to fail—may be the only thing that saves us. The only thing keeping us from sinking. The only thing keeping us from becoming what we’re fighting against. So today, I choose to believe. To believe that we can still make a difference. That we can still chart a course. That we can still say: this far, and no further. Because otherwise, we’ll have nothing left. No more humanity. No more choices. No more future. Just machines. Just algorithms. Just an endless war. And that is not a future. It is a curse. A curse we bring upon ourselves. A curse we can still avoid. But we must act. Now. Before it’s too late.

Signed, Jacques Provost