Autonomous Weapons & Algorithmic Warfare: Why Defence Experts Want an AI Geneva Convention

1. The Silent Sky: Why Electronic Warfare Forced the Autonomous Pivot

For twenty years, the public image of drone warfare was the MQ-9 Reaper: a multi-million-dollar aircraft flying at 25,000 feet, piloted via satellite link by an officer seated in an air-conditioned trailer in Nevada. The human was always in the loop. The human pulled the trigger.

That paradigm is obsolete. In modern high-intensity electronic warfare across Eastern Europe and the Middle East, the electromagnetic spectrum is a violent storm of jamming. High-power spoofers fry satellite GPS signals; broadband jammers sever the radio control links between pilots and their drones.

When a $1,000 FPV (First-Person View) loitering drone loses connection at 200 meters from a target, it crashes uselessly into the mud. The military countermeasure was inevitable: onboard Edge AI.

Militaries now equip $40 edge-compute processors (like Raspberry Pi or Jetson modules) running lightweight convolutional neural networks directly on the drone. When radio connection is severed, the autonomous software takes over: it tracks optical contrast, matches the vehicle shape against a neural model of enemy armor, and executes the terminal dive with zero human input.

2. The Speed Trap: Flash Wars at 20 Milliseconds

Military commanders do not adopt Lethal Autonomous Weapons Systems (LAWS) because they dislike humans; they adopt them because of the physics of modern combat:

This creates an unstable escalation dynamic known as hyperwar. If Nation A keeps a human in the loop, while Nation B removes the human to save 2,000 milliseconds, Nation B wins every tactical engagement. Every armed force is structurally incentivized to remove the human safety catch.

3. The Three Pillars of an AI Geneva Convention

Just as the 1925 Geneva Protocol banned chemical and biological weapons after the horrors of World War I, international legal and military ethicists are demanding a binding treaty on autonomous algorithms:

• Pillar 1: Prohibition on Anti-Personnel LAWS. An absolute ban on autonomous weapons designed to identify and execute individual human beings without positive, human optical sign-off.
• Pillar 2: Predictability & Explainability Standards. Forbidding black-box models whose lethal targeting behavior cannot be rigorously audited in post-incident military tribunals.
• Pillar 3: The Nuclear Air-Gap. Strict international treaty verification ensuring that strategic nuclear early-warning and launch authorization chains maintain physical, human air-gaps immune to algorithmic automation.

4. The Moral Invariant: Preserving Human Accountability

Warfare is the ultimate expression of human tragedy, but throughout history, it was governed by moral and legal accountability. When a soldier fires a weapon, they carry the legal and spiritual weight of that decision; they can be prosecuted for war crimes if they violate the Geneva Conventions.

Who do you put on trial when an unexplainable neural network misidentifies a school bus as an armored troop transport and launches a missile? The software engineer? The product manager? The GPU manufacturer? The commanding general?

Delegating the decision to kill to a mathematical weight matrix degrades the ultimate value of human life. Maintaining human agency in warfare is not a technical impediment; it is the last thin line protecting civilization from algorithmic barbarism.