Earlier weapons debates are the reference point. Some states eventually agreed to ban landmines after concluding that they caused unnecessary suffering and superfluous injury. A protocol on blinding lasers was treated as a model when talks on autonomous weapons began, and groups that had pushed for a ban from the start hoped to repeat that path.
New York Times reporting suggests that the first fully autonomous drone to take a human life has already been launched.
Envoys from 128 nations met in Geneva on Monday to decide whether to open negotiations on legally binding rules for weapons that can select and kill targets without human involvement.
The session tested whether governments will move past ten years of argument over legal technicalities and into formal talks on how far states may rely on such systems and how tightly they must constrain their use.
In an autonomous weapon, the system itself identifies, selects, prioritizes and engages a target through an algorithm, rather than leaving each of those steps to a person.
Depending on the definition, the technology is not new. States have developed autonomous weapons for defense for decades, including cases in which there is too little time to react to an incoming missile and automated processes identify and engage the threat.
What has changed is that a debate long treated as futuristic is now tangible. There are growing cases in which a weapon is used in autonomous mode to finish an attack.
The war in Ukraine has sharpened that shift. Autonomy is being combined with drones and missiles for offensive last-mile targeting in areas where remote control is not feasible because communications are denied.
Russia and Ukraine are both believed to be testing attack drones that go beyond locking onto targets chosen by human operators and instead choose the targets themselves. There have also been reports that such systems have been used against people.
The core concern is who decides on the use of force. States agree that humans, and commanders in particular, remain responsible under international law. The unresolved question is how autonomous features affect their ability to judge that a use will be lawful.
The main legal test is whether these weapons can be used in compliance with existing international humanitarian law.
Historically, major limits have followed documented cases in which a technology caused unnecessary harm to civilians. States have, in a few cases, later agreed that a weapon they had fielded was problematic on legal or ethical grounds.