By Oleksandr Korniyenko, Ukrainian politician and Member of the Verkhovna Rada. He has served as First Deputy Chairman of Parliament since 2021 and is a senior figure in the Servant of the People party. He was first elected to parliament in 2019, having previously worked as a political consultant and campaign manager.

On Invitation for LI’s Reimagined Newsletter – The Liberal Dispatch

Sign Up for The Liberal Dispatch

Artificial intelligence now touches almost every aspect of our lives – the labour market, education, healthcare, even life expectancy. In Ukraine, however, it has become central to something even more consequential: how we survive a war we did not choose. For more than five years, we have been resisting an aggressor with a larger territory, more weapons and, as experience has shown, far fewer moral constraints than we have. That is why we have harnessed AI for our defence and, in the process, have learned a great deal about both its potential and its limitations.

In March 2026, our Ministry of Defence launched the Defence AI Center “A1”, with the support of the UK Government, to accelerate the systematic integration of AI into the Armed Forces. At the same time, the Brave1 cluster has evolved into a genuine innovation ecosystem: more than 200 companies are developing AI-enabled drones, over 300 artificial intelligence and computer vision projects have been registered on the platform, and more than 70 of them are already being deployed on the battlefield. Within the Delta situational awareness system, the “Tower” module processes around 75,000 videos each month and identifies 70% of enemy equipment in just 2.2 seconds. AI-guided interceptor drones, capable of locking onto their targets during the final stage of flight even under enemy electronic warfare when communication with the operator has been lost, destroyed more than 33,000 enemy UAVs in March alone. One of the participants in Brave1 has automated 95% of the interception process for Shahed-type drones. The technology has already been tested in combat in the Kharkiv region, with confirmed successful interceptions.

Drone swarms deserve special mention—a technology with enormous potential to redefine the future of warfare, where individual systems are coordinated into intelligent autonomous groups. No country in the world has yet demonstrated the combat use of a drone swarm under conditions of intense electronic warfare and communications suppression. This is an “extra-credit challenge” that we are striving to solve first. To make this possible, every drone within the swarm must be capable of operating independently, even if communications are completely severed or neighbouring drones are physically destroyed. This has evolved into a national effort involving dozens of teams and some of the country’s finest engineers. It is precisely the combination of cutting-edge expertise and practical combat experience that gives us a unique opportunity to pioneer this technology.

Yet throughout this process, we have remained committed to one fundamental principle: a human remains in the target engagement cycle. AI may identify and track a target, but it is the operator who makes the final decision and presses the launch button. This principle is fundamental to our approach, and I want to explain why—because, in our circumstances, it would be easy to argue the opposite.

The Temptation We Do Not Succumb To

When the survival of a nation is at stake against an enemy unconstrained by conscience, the argument seems almost self-evident: give AI greater authority, raise the threshold of autonomous decision-making, allow machines to act at machine speed because humans may simply be too slow. We have considered this possibility seriously. Yet we remain convinced that it is the wrong path. Technology exists to serve humanity—it must never replace human judgement.

Why, in the middle of a full-scale war, are we concerned with such questions at all? One could argue that, in our circumstances—where national survival comes before everything else—it would be more logical to delegate as much authority as possible to AI rather than spend time debating ethical safeguards. Yet it is precisely because the stakes are so high that we believe these questions must be addressed now, rather than postponed until after the war.

AI governance in the military domain is essential regardless of whether a country is at war, and there are concrete reasons for this. The internal logic of language and classification models remains opaque even to their own developers. The well-known “black box” problem becomes unacceptable the moment it affects decisions involving life and death. These models are trained on known patterns, whereas war is defined by the constant emergence of new ones. It is often said that generals prepare to fight the previous war—with AI, this danger becomes even more acute. Military operations conducted at machine speed may outpace a human’s ability to intervene. And the very same technologies used to monitor the enemy can just as easily be turned inward to monitor a country’s own citizens. The only barrier between those two outcomes is ethics—not institutional safeguards. We therefore need robust institutional safeguards as well.

Why Regulation Is Complicated

Technology invariably outpaces regulation. New capabilities can emerge within a matter of months, while regulatory processes are slowed by bureaucracy and politics. At the same time, developers often adopt their own ethical frameworks, which do not always align with the expectations of the state. The debate surrounding Anthropic and the U.S. Department of Defense illustrates how sharply the perspectives of private companies and governments can diverge, and how underregulated this area remains even in the country that leads the field.

Can a Liberal Regulate AI While Remaining a Liberal?

This question is even more difficult because most of the technology is privately owned.

I believe the answer is yes.

Liberalism seeks to maximise individual freedom precisely for the sake of the common good. Our responsibility is to ensure that a technology of such extraordinary power becomes a force for good rather than harm—as so often happens under dictatorships, where those in power are accountable to no one.

European liberals, in particular, already possess the foundations for responsible regulation: the rule of law, an independent judiciary, a free press and the separation of powers. The diversity within the liberal family, and the absence of a single imposed line, is a strength rather than a weakness. It allows different approaches to be tested against one another instead of forcing everyone to follow a single doctrine handed down from above.

That is precisely why we want this conversation to take place, with Ukraine’s experience serving as one of the cases for discussion rather than as an exception that can simply be set aside with the argument: “But you are at war.”

In 1991, Manuel DeLanda published War in the Age of Intelligent Machines, a book that, in many ways, anticipated our present reality by nearly four decades. One idea from that book has stayed with me ever since: he argued that the introduction of strategic war games after the Second World War fostered a certain detachment among military personnel from human casualties—turning lives into numbers on a map rather than lived realities.

Today, however, we are not dealing with war games or abstract maps. We are dealing with something far more consequential—something that is already transforming the face of warfare in real time. This demands from us the highest level of responsibility the moment requires.

We must approach this challenge with the sense of responsibility it demands of us.