Views: 0
Until recently, talk of autonomous military drones seemed like something out of a science fiction movie. Today, that is no longer the case. The combination of drones, sensors, artificial intelligence, and navigation systems allows an unmanned aircraft to travel to a specific area, identify objects, and make decisions on how to act. The war in Ukraine is vastly accelerating this process.
However, it is important to distinguish between flight autonomy and the autonomy to kill. A drone can fly, dodge obstacles, pinpoint a location, or return to base without human intervention. That does not necessarily mean it decides on its own to attack a person. The problem arises when that final decision is also left to the machine.
In recent years, reports have emerged regarding systems used in Ukraine that are capable of identifying targets and executing specific actions with an increasing degree of autonomy. Some of this information is difficult to verify independently, but the technological capability already exists and is advancing rapidly.
And this is where the real debate begins. Imagine a drone ordered to head to a specific area. Once there, its sensors detect various objects. Its software analyzes the images, compares what it sees against specific patterns, and determines that one of them corresponds to a military target.
The machine acts. It kills people, yet there is not necessarily a human watching the image at that moment and making the final decision. Who, then, is responsible if it makes a mistake?
This issue is not only technological. It is legal, ethical and, above all, deeply human.
International humanitarian law already sets out rules for how attacks should be carried out and how civilians should be protected. But there is currently no specific international treaty that generally prohibits all autonomous weapons capable of selecting and attacking targets. The United Nations has been trying for years to move towards common rules, but negotiations continue without a definitive agreement, in part due to great powers such as the USA, Russia, and others.
Meanwhile, a curious semantic battle appears. We talk about “meaningful human control”, “adequate human judgment” or keeping the human being within the decision process. They are important concepts, but they can also become a way to postpone the problem.
The International Committee of the Red Cross has called for legally binding rules and argues that autonomous systems used to apply force against people should be banned. In August 2026, moreover, the United Nations and the ICRC again urgently called for new international standards in the face of the growing risk of machines autonomously selecting human beings as targets.
Because there is a fundamental difference. Having a machine help a soldier make a decision is one thing. For a machine to make the decision to kill by itself is another matter entirely.
In a war, a human being can make mistakes, act badly or commit atrocities. But behind that decision there is a person who, at least in principle, can be considered responsible for his actions.
With an autonomous machine, a new problem appears: a life or death decision can be divided between algorithms, sensors, data, programmers and military commanders. And perhaps that is the real danger.
Technology is not to blame. Neither is artificial intelligence. The responsibility remains ours. The problem is that we are building machines that are increasingly capable of making decisions and, at the same time, we still cannot agree on which decisions we should never delegate.
For me, there is a boundary that should remain clearly defined: no machine should have the final decision over the life of a human being.
Because perhaps the greatest danger of artificial intelligence is not that machines become too human. Maybe it's that we start behaving like machines. And if we don't behave like humans then exactly what are we?
EN 
ES