UN talks in Geneva fall short of binding rules on autonomous weapons

Diplomats meeting at the United Nations in Geneva have agreed to start discussing options to regulate lethal autonomous weapons, often called killer robots.
After a week of negotiations that dragged on until the small hours of Saturday morning, a compromise emerged: countries would have two years to suggest regulations that could be legally binding, but could also be optional.
Bleary-eyed negotiators insisted the agreement was better than nothing and showed the UN disarmament process could still function. But for many humanitarian organisations and many countries, the deal falls far short of what they had hoped for.
Before the negotiations got underway, the International Committee of the Red Cross, the guardian of the Geneva Conventions, said it was urgent for legally binding rules to be drafted within a year. That call was backed by the UN secretary-general and by Pope Leo.
Their concerns stem from the kind of warfare now being seen. In Ukraine, Sudan and Gaza, civilians are being killed by semi-autonomous drones. According to the UN, this year in Sudan 80 percent of civilian deaths were caused by drones.
Humans remain, albeit very remotely, in control for now, but humanitarian organisations fear artificial intelligence-assisted weapons may soon make the decision to kill entirely on their own.
ICRC legal adviser Georgia Hinds said there is a fundamental ethical concern with having life-and-death decisions determined by programming, by an algorithm. In her words, that is not something that can be fixed with better technology; it is a moral red line that cannot be crossed.
No one went into the negotiations thinking they would be easy. Some of the world's most powerful countries, among them the United States, Russia and Israel, had already indicated their reluctance to agree to binding regulations.
On day one, Russia delayed proceedings by objecting to the meeting's agenda. Then the United States proposed a five-year window to discuss nonbinding, optional guidelines for autonomous weapons.
One frustrated delegate said they were just kicking the can down the road. As the weekend approached, diplomats scrambled to get some kind of result from negotiations that had been billed as a last chance to regulate these weapons.
Finally, at 4:30 a.m. on Saturday morning, they emerged from the negotiating room to announce their compromise: no to a deal within a year, maybe to legally binding rules, or maybe not.
The ICRC immediately said it regretted the move and renewed its appeal for a binding treaty. Other campaigners called the deal a real disappointment and a nothing burger. Another added that he was concerned people would die because of this.
Anna Hehir, an expert in military AI governance at the Future of Life Institute, said the systems are likely to proliferate. In her words, they are not like nuclear weapons in that they are inaccessible; there are autonomous weapons that are easy to purchase or even to create for all kinds of actors: individuals, nonstate actors and criminal groups.
Hinds also said there are big questions about how the existing rules of war, the Geneva Conventions, can be upheld if those doing the killing are not humans but machines. In her words, the commander remains responsible for making sure the use of force is lawful and remains accountable; a machine cannot be held accountable for a violation of the law.
Although most UN member states do agree there should be some controls over autonomous weapons, the Geneva compromise means that, for the next two years at least, weapons that can use artificial intelligence to make life-and-death decisions will remain unregulated.


