Experts Split on Coding Morality Into AI Drones as 100-Plus Startups Build Autonomous Weapons
Updated
Updated · The Guardian · Jun 3
Experts Split on Coding Morality Into AI Drones as 100-Plus Startups Build Autonomous Weapons
2 articles · Updated · The Guardian · Jun 3
Summary
David Omand and other advocates say future militaries will need drones that can make some decisions on their own, with rules of engagement and legal constraints embedded as autonomy expands.
Zee Talat, Andrew Rogoyski and Jessica Dorsey counter that large language models are probabilistic tools, not moral agents, and cannot reliably encode contested human judgments such as civilian-combatant distinction.
Article 57 of the Geneva Conventions sharpens that concern by requiring feasible verification of military targets, raising fears that flawed AI judgments could be repeated at scale and at machine speed.
More than 100 US and European startups are already building drone systems, but they diverge on how far autonomy should go—from human-guided 'last mile' navigation to kill-box missions where humans approve an area, not each target.
The debate is intensifying as drones proliferate in Ukraine and AI assists bombing in the Iran conflict, while global legal and moral consensus on autonomous weapons still lags the technology.
Can AI's 'moral code' prevent war crimes, or will it just automate killing on an industrial scale?
As AI accelerates both military targeting and cyberattacks, is a new, uncontrollable form of warfare now inevitable?
When an autonomous drone mistakenly kills civilians, who is legally and morally responsible for the algorithm's decision?
Who Decides When AI Kills? The Urgent Global Debate Over Moral Codes, Human Control, and Accountability in Autonomous Weapons
Overview
As AI-powered weapons rapidly advance, former GCHQ chief David Omand has called for a moral code to guide their use, highlighting the urgent need for ethical and legal frameworks. This call has intensified debate, with critics arguing that AI cannot make the complex, life-or-death judgments required in warfare, as it only processes data and cannot distinguish between civilians and combatants. Public support for autonomous lethal decisions remains low, underscoring widespread concern. The report explores these challenges, emphasizing the importance of meaningful human control, accountability, and international cooperation to ensure AI in warfare aligns with human values.