The ICRC Position on Autonomous Weapon Systems is the formal institutional position of the International Committee of the Red Cross — the custodian of international humanitarian law (IHL) — on autonomous weapon systems (AWS). First issued on 12 May 2021 as a background paper published in the International Review of the Red Cross Vol. 102 No. 915, it calls for a new legally binding international instrument and stands as an international-organization counterpart to national-policy instruments such as DoD Directive 3000.09. The position was reinforced by the ICRC's March 2024 submission to the UN Secretary-General pursuant to UNGA Resolution 78/241 and the 3 May 2024 Vienna Conference statement by ICRC President Mirjana Spoljaric.
Summary of the position
The ICRC calls for a new legally binding international instrument comprising three elements: a prohibition on unpredictable AWS, a prohibition on anti-personnel AWS, and strict regulation of all other AWS to ensure meaningful human control. President Spoljaric's 2024 Vienna statement set the target of negotiating such an instrument by 2026.
Definition of AWS
The ICRC defines autonomous weapon systems as:
"Weapon systems that select and apply force to targets without human intervention. After initial activation or launch by a person, an autonomous weapon system self-initiates or triggers a strike in response to information from the environment received through sensors and on the basis of a generalized 'target profile'."
A consequence the ICRC draws from this definition is that the user does not choose, or even know, the specific target, timing, or location of the resulting application of force.
Concerns
The ICRC frames AWS through three concerns. The humanitarian concern addresses risks of harm to civilians and combatants and escalation risks, assessed through an "effects-based approach" the ICRC traces to 160 years of documenting weapons' humanitarian effects. The legal concern addresses IHL compliance under the principles of distinction, proportionality, and precautions, each of which the ICRC argues requires contextual human judgment at the time and place of the attack. The ethical concern is stated as the position that "ceding life-and-death decisions to machine sensors and software is a dehumanizing process that undermines our shared humanity."
Recommendations
The position advances three recommendations. The first is to prohibit unpredictable AWS, defined as those "designed or used in a manner such that their effects cannot be sufficiently understood, predicted and explained"; the ICRC notes this captures many machine-learning-based systems, and Spoljaric's 2024 statement specifically cites "opaque machine learning algorithms." The second is to prohibit anti-personnel AWS, defined as those "designed or used to apply force against persons," which the ICRC grounds in both IHL (distinction and proportionality with respect to persons) and independent ethical objections to machine killing of humans. The third is to strictly regulate all remaining AWS through limits on target type (military objectives by nature), limits on duration, geographical scope, and scale of use, limits on situations of use (for example, civilian-absent environments), and requirements for human–machine interaction (effective supervision, timely intervention, and deactivation).
Meaningful human control
The ICRC gives meaningful human control operational content: the human user must have sufficient information, sufficient time, and the ability to act — to activate, supervise, intervene, and deactivate. IHL commentators generally read this as stricter than DoD Directive 3000.09's standard of "appropriate levels of human judgment over the use of force," which does not mandate control over specific target selection. The ICRC has not formally superseded the 2021 position; the 2024 submission reinforces it, and no ICRC document expressly retires it.
Key claims
- The ICRC's formal position, first issued 12 May 2021, calls for a new legally binding international instrument with two prohibitions plus strict regulation, drawing on the primary-source ICRC website and the International Review of the Red Cross. (Confidence: high.)
- The 2024 UNSG submission and Spoljaric's Vienna 2024 statement reinforce and re-operationalize the 2021 position without departing from it; the 2026 negotiation target is new. (Confidence: high.)
- The ICRC regards existing IHL as applicable but insufficient to resolve the humanitarian, legal, and ethical questions AWS raise, and argues new binding rules are needed. (Confidence: high.)
- UNGA Resolution 78/241 (December 2023), which 152 states supported, created the UN-track vehicle the ICRC has since engaged alongside the stalled CCW GGE process. (Confidence: high.)
- The ICRC's framing of meaningful human control is stricter than DoD's "appropriate levels of human judgment." This is an inference from textual comparison, widely accepted among IHL commentators but contested by DoD. (Confidence: medium.)
- The 2021 position has not been formally superseded; the 2024 submission reinforces it, and no ICRC document expressly retires the 2021 position. (Confidence: medium.)
Provenance and standing
The ICRC is not a typical NGO. It holds a treaty-based mandate under the Geneva Conventions and is the custodian of IHL, and its positions carry quasi-institutional weight in international humanitarian law. The ICRC has led or materially contributed to previous weapons-regulation treaties, including the ban on anti-personnel mines (Ottawa 1997), cluster munitions (CCM 2008), and blinding laser weapons (CCW Protocol IV, 1995). The 2021 position was issued after the ICRC had engaged the CCW GGE process for seven years with limited progress. The target of negotiating a binding instrument by 2026, set in Spoljaric's Vienna statement of 3 May 2024, coincides with Austria-led UNGA-track discussions.
Relation to other pages
The position supplies the IHL and international-organization side of the AWS governance debate discussed on Autonomous Weapons. It stands as the principal international counterpoint to the US national-policy approach in DoD Directive 3000.09 — Autonomy in Weapon Systems (source summary): the ICRC advocates treaty-level categorical prohibitions for unpredictable and anti-personnel AWS, while the directive regulates via national policy and prohibits no category. The contrast runs along several axes — regulation versus prohibition, national policy versus treaty, and "appropriate human judgment" versus "meaningful human control." Relative to Clawed, Dean Ball's analysis of the Anthropic–Department of War conflict offers a republican-governance frame addressing who should set AI-use policy, while the ICRC position addresses the substantive limits that policy must respect; the two frames are complementary. The ICRC's ethical argument about dehumanization also intersects the concerns raised on AI and Authoritarianism about autonomous force exercised without human moral friction.
Relationships
- instance-of: Autonomous Weapons — the IHL / international-organization position.
- contradicts: DoD Directive 3000.09 — Autonomy in Weapon Systems (source summary) — regulatory national-policy approach vs. prohibitive international-law approach.
- related: Clawed — the ICRC's substantive IHL content complements Ball's procedural/governance critique.
- related: International Committee of the Red Cross (ICRC) — the issuing entity.
- depends-on: existing IHL (Geneva Conventions, Additional Protocols) as the legal framework the ICRC argues must be clarified and supplemented.