A plain-language definition
Artificial weapons is an informal term for weapons that use artificial intelligence or autonomous functions. It is not defined in international law. The formal terms are autonomous weapon systems (AWS) and lethal autonomous weapons systems (LAWS).
What these definitions share is one test: can the system select and attack a target without a human making that decision each time? If yes, it is autonomous in the sense that matters for law and policy. If a human chooses and fires on every target, it is not, however advanced its sensors or software are.
On this site, "artificial weapons" is used more broadly. It covers autonomous weapons plus AI-enabled systems such as targeting software, drone swarms and AI-assisted defense, and we label which is which.
How the ICRC, the US DoD and the UN define it
Any weapon system with autonomy in its critical functions. That is, a weapon system that can select (i.e. search for or detect, identify, track, select) and attack (i.e. use force against, neutralize, damage or destroy) targets without human intervention.
Source: ICRC, What you need to know about autonomous weapons.
A weapon system that, once activated, can select and engage targets without further intervention by a human operator.
The directive also requires that autonomous and semi-autonomous weapon systems be designed to allow commanders and operators to exercise "appropriate levels of human judgment over the use of force." Source: DoD Directive 3000.09.
The first General Assembly resolution on lethal autonomous weapons systems. It was adopted by a recorded vote of 152 in favour to 4 against (Belarus, India, Mali, Russian Federation), with 11 abstentions including China, North Korea and Israel. It requests the Secretary-General to seek the views of member states and other stakeholders on these systems.
Source: UN document A/RES/78/241.
Autonomous, semi-autonomous and human-operated
Debate about artificial weapons is mostly a debate about how much of the targeting decision stays with a person. Policy discussions commonly describe three positions:
- Human in the loop: a person must approve each engagement.
- Human on the loop: the system can engage on its own while a person supervises and can intervene.
- Human out of the loop: the system selects and engages without a person supervising that engagement.
These are descriptive labels, not legal categories. The ICRC and DoD texts above define the autonomy of the critical functions (select and engage) rather than using these labels. See the glossary for related terms such as HITL and LAWS.
Examples of systems in this category
- Phalanx CIWS: a close-in weapon system that automatically detects, tracks and engages incoming missiles and aircraft. It is a defensive system that operates autonomously within a narrow task.
- IAI Harpy: a loitering munition designed to loiter over an area and attack radar emitters.
These are two long-established examples. Newer systems add AI to targeting, navigation and swarm coordination. Browse the full systems directory, with sources, by category and country.
Where to follow artificial weapons on this site
- Weapons database: systems by category and country
- Map and conflict map: where AI-enabled weapons are in use
- Treaties and rules: where international rules stand
- Company profiles and defense AI budgets
- Research articles and the glossary
Frequently asked questions
What are artificial weapons?
"Artificial weapons" is an informal term for weapon systems that use artificial intelligence or autonomy. Official bodies use the terms autonomous weapon systems (AWS) and lethal autonomous weapon systems (LAWS). The core idea in the ICRC, US Department of Defense and UN texts is the same: a weapon that can select and attack targets without a human deciding each time.
Is "artificial weapons" an official term?
No. It is not defined in international law or in US policy. The formal terms are "autonomous weapon system" (ICRC, US DoD Directive 3000.09) and "lethal autonomous weapons systems" (UN General Assembly Resolution 78/241). This site uses "artificial weapons" to cover AI-enabled and autonomous weapons in general.
How does the US Department of Defense define an autonomous weapon system?
DoD Directive 3000.09 (effective January 25, 2023) defines it as a weapon system that, once activated, can select and engage targets without further intervention by a human operator. The directive requires that such systems be designed to allow appropriate levels of human judgment over the use of force.
How does the ICRC define autonomous weapons?
The ICRC describes them as any weapon system with autonomy in its critical functions: a system that can select (search for or detect, identify, track, select) and attack (use force against, neutralize, damage or destroy) targets without human intervention.
Has the United Nations acted on autonomous weapons?
Yes. On 22 December 2023 the UN General Assembly adopted Resolution 78/241 on lethal autonomous weapons systems by a recorded vote of 152 in favour to 4 against (Belarus, India, Mali and the Russian Federation), with 11 abstentions, including China, North Korea and Israel. It asks the Secretary-General to seek the views of states and other stakeholders.
Are all drones artificial weapons?
No. A drone flown by a remote pilot who selects and fires on each target is not autonomous in the sense used by the ICRC or DoD. A drone or munition that can search for, select and attack targets on its own is. Many systems sit between these two, which is why definitions focus on the targeting and engagement functions.
Sources
- ICRC: What you need to know about autonomous weapons
- US Department of Defense Directive 3000.09, Autonomy in Weapon Systems (January 25, 2023)
- UN General Assembly Resolution 78/241, Lethal autonomous weapons systems (22 December 2023)
Last updated 2026-10-09. Quotes are reproduced from the sources above; everything else on this page is our summary of them.