New arms race: India must help regulate autonomous weapons

A swarm of AI-powered drones or unmanned tanks could maneuver and strike in fractions of a second, outpacing any human commander's ability to respond. But the same speed that promises advantage on the battlefield also raises the spectre of uncontrollable escalation.

- POOJA ARORA

On a screen in a military command centre, an AI algorithm flags a target. Within seconds, a drone locks on and fires—no human pressed the trigger. This scenario is no longer science fiction. In recent conflicts, we have seen glimpses of these autonomous weapons systems (AWS) in action. Israel, for example, reportedly deployed AI tools codenamed "Lavender" and "Where's Daddy?" to identify suspected militants and track them to their homes for targeting. An algorithm called "The Gospel" sifted through surveillance data to generate lists of buildings for airstrikes. Meanwhile, a U.S. defence startup, Anduril Industries, is developing software to coordinate swarms of thousands of autonomous drones for the Pentagon. Major leaders in AI products such as Alphabet and OpenAI are also entering the realm of defence.

AWS differ fundamentally from previous advancement in weaponry. Nuclear, chemical, and biological arms magnify a human's capacity to kill, but a human is still in charge of when and where to unleash them. By contrast, AWS shift life-and-death decision-making from man to machine. In plain terms, these are weapons that, once activated, identify and attack targets on their own. An autonomous weapon might decide who lives or dies based on sensor inputs and code—not conscious human judgment. This handover of lethal decision authority represents a profound change in warfare. It's a change that the world, and India in particular, cannot afford to ignore. We urgently need international rules for this new class of weaponry, because allowing machines to make kill decisions without robust oversight is a recipe for disaster.