Autonomous Weapons
Autonomous weapons are broadly defined as weapons systems that can — through algorithms rather than direct human orders — select and engage targets without meaningful human intervention. They range from semi-autonomous systems, where a human approves each strike, to fully autonomous systems where the machine identifies, tracks, and acts alone — though the distinction matters less than it appears when the human has seconds to react and almost never overrides. The line between the two is — as ethicists, military commanders, and diplomats have all discovered — the subject of intense international debate with no resolution in sight.
Drone Warfare
The most widespread autonomous weapons are military [[drone-warfare|drones]], which have evolved from remote-controlled surveillance aircraft to [[artificial-intelligence|AI]]-piloted combat systems in less than two decades. [[ukraine|Ukraine]] produces over four million first-person-view drones per year — small, disposable suicide drones guided by operators wearing VR headsets that cost around $400 and can destroy tanks worth millions, a cost asymmetry that military strategists compare to the [[machine-gun|machine gun]] in [[world-war-i|World War I]]. [[iran|Iran]] launched over 300 drones and missiles at [[israel|Israel]] in a single night in April 2024, demonstrating that cheap, expendable swarm tactics can overwhelm even the most advanced air defenses with sheer numbers.
AI Targeting
[[artificial-intelligence|AI]]-powered targeting systems generate potential target lists faster than any human intelligence team, and several are already deployed in active combat zones. [[israel|Israel]]'s [[lavender-ai|Lavender]] system generated a list of 37,000 suspected militants in [[gaza|Gaza]], with human operators given approximately twenty seconds to approve each strike. The system used surveillance, phone metadata, and behavioral data to flag individual targets — not on evidence or testimony, but on algorithmic pattern-matching — and officers described a protocol of tracking each target to their home and striking the building, accepting civilian casualties as a trade-off for operational efficiency. [[gospel-ai|The Gospel]], another Israeli AI system, generates bombing targets for buildings and critical infrastructure at what officers described as unprecedented rates, essentially replacing teams of analysts who previously spent days building a single target package.
The [[pentagon|Pentagon]]'s [[project-maven|Project Maven]], launched in 2017, uses AI to analyze drone surveillance footage across multiple theaters simultaneously, flagging potential targets based on pattern-of-life analysis. When [[google|Google]] employees protested the company's involvement in 2018, the contract was not publicly renewed — but the work continued at [[palantir|Palantir]], [[anduril|Anduril]], and other defense contractors, proving that the ethical objection did not end the program but merely moved it to companies without the objectors. The broader lesson was unmistakable: once military-grade AI targeting capabilities exist, they do not simply disappear because one company's employees object.
The Kargu-2 Incident
A [[united-nations|United Nations]] report documented the first known fully autonomous drone attack in [[libya|Libya]] in March 2020. A Turkish-made [[kargu-2|Kargu-2]] drone, a small rotary-wing weapon designed to operate independently in GPS-denied environments, reportedly tracked and engaged a retreating soldier without any human command — the first documented case of a machine autonomously deciding to use lethal force against a human being. [[turkey|Turkey]] disputed the UN's characterization, but the report was not retracted. The incident received remarkably little mainstream media attention given its historic significance — a threshold crossing that military ethicists had warned about for decades had apparently occurred in a conflict most of the world was barely following.
The Arms Race
The [[united-states|United States]], [[china|China]], [[russia|Russia]], [[israel|Israel]], [[south-korea|South Korea]], [[turkey|Turkey]], and at least a dozen other nations are developing lethal autonomous weapons systems, and there is no international treaty banning them. Negotiations at the [[united-nations|United Nations]] Convention on Certain Conventional Weapons have stalled for over a decade, with major military powers — particularly the US, Russia, and Israel — blocking binding agreements. The central argument against a ban is that autonomy makes weapons more precise and reduces civilian casualties. The equally forceful argument for a ban is that delegating the irreversible decision to kill to a machine violates fundamental principles of human dignity and the [[laws-of-war|laws of war]] — the [[geneva-conventions|Geneva Conventions]] require distinction, proportionality, and accountability, but an autonomous weapon that kills a civilian raises the unanswerable question of who bears responsibility. Meanwhile, and perhaps most alarmingly, commercially available drones are being rapidly retrofitted for lethal combat use by non-state actors and governments alike around the world, lowering the barrier to autonomous violence to the budget of a hobbyist.