By: Emma Staller ‘26
Volume XI – Issue I – Fall 2025
I. INTRODUCTION
The 1949 Geneva Conventions and their Additional Protocols are international treaties that establish the core rules limiting the barbarity of war. [1] They protect people who do take part in the fighting, including civilians, medics, and aid workers, as well as those who can no longer fight, such as wounded, sick, and shipwrecked troops and prisoners of war. Many scholars consider the Geneva Conventions one of humanity’s most significant achievements of the last century, as they form the foundation of International Humanitarian Law (IHL), the body of international law regulating armed conflict and seeking to mitigate its effects. [2] The Conventions and their Protocols prescribe measures to prevent or halt breaches, define “grave breaches,” and outline consequences for those who commit them. [3]
As artificial intelligence (AI) technology continues to advance, opportunities for its large-scale application in warfare have expanded dramatically, creating novel and precarious challenges for IHL. Among the greatest threats posed by AI in warfare is the rise of Lethal Autonomous Weapon Systems (LAWS), often called 'killer robots' for their destructive potential. [4] Machines, and by extension killer robots, currently cannot bear responsibility for violations of international law. [5] As it stands, when a machine decides who lives and who dies, the Geneva Conventions become ceremonial relics, not enforceable law. IHL assumes humans make decisions with intention and conscience; autonomous weapons systems disrupt this assumption, creating a legal “black hole” whereby nobody can be held accountable. [6]
II. CONTEMPORARY USE & DEBATES SURROUNDING AUTONOMOUS WEAPON SYSTEMS
Autonomous Weapon Systems (AWS) are weapons that, once activated, can identify and engage one or more targets without further human intervention. [7] After activation or launch, an AWS responds to information gathered from its environment through sensors and applies force based on a generalized “target profile.” [8] Because the process is automated, the operator who initiated the system neither selects nor necessarily knows the specific targets, nor the exact timing or location of the resulting strike. [9]
In contemporary military practice, AWS are primarily employed to engage military objects, such as enemy missiles, projectiles, radars, or warships—targets that, under IHL, are classified as military objectives by nature. Most deployments occur in contexts where civilians and civilian objects are absent, or where measures, such as physical barriers, are implemented to prevent civilian presence in areas where AWS are active. [10] Many AWS are stationary rather than mobile, and nearly all are monitored in real time by human operators who retain the ability to intervene and suspend an attack if necessary. [11] Militaries have long deployed defensive systems, including human-supervised Close-In Weapon Systems and Active Protection Systems, which allow operators to halt or cancel engagements as circumstances require. [12] While these operational safeguards and regulations demonstrate efforts to constrain AWS use and support compliance with IHL, there are still many opposed to the usage of AWS.
Arguments in favor of AWS use generally fall into two categories: military advantages and moral justifications. [13] From a military perspective, these “killer robots” act as force multipliers, reducing the number of human warfighters required for a given mission. They also enable operations in environments that would otherwise be inaccessible to humans. Additionally, the deployment of AWS has the potential to reduce casualties by removing human personnel from high-risk situations. The Department of Defense’s Unmanned Systems Roadmap: 2007–2032 notes that robots are particularly well-suited for missions that are “dull, dirty, or dangerous.” [14] For example, a “dull” mission might involve long-duration sorties, a “dirty” mission could expose humans to harmful radiological material, and a “dangerous” mission might involve explosive ordnance disposal. [15]
Moral justifications for AWS use are more abstract, yet they are supported by several military and robotics experts. American roboethicist Ronald C. Arkin argues that killer robots could behave more “humanely” on the battlefield because they are not vulnerable to emotions such as fear or hysteria that can cloud human judgment. [16] Lt. Col. Douglas A. Pryer of the U.S. Army supports this view and points to neuroscience research indicating that extreme stress can shut down the neural circuits responsible for self-control. [17] According to Pryer, this breakdown can contribute to acts such as sexual assault and other forms of misconduct that soldiers might otherwise be less likely to commit. [18]
One of the dominant oppositions posed against the usage of AWS is based on moral grounds. In an open letter announced on July 28th, 2015 at the opening of the International Joint Conference on Artificial Intelligence (IJCAI), an impressive list of signatories signed their support on the fight against AI technologies and AWS military use. [19] With signatories such as physicist Stephen Hawking , professor Noam Chomsky , and businessman Steve Wozniak , the letter warns, “Artificial Intelligence (AI) technology has reached a point where the deployment of such systems is—practically if not legally—feasible within years, not decades, and the stakes are high: autonomous weapons have been described as the third revolution in warfare, after gunpowder and nuclear arms.” [20] Among the various concerns surrounding AWS, the legal dimension is particularly pressing, as existing accountability mechanisms are fundamentally ill-equipped to address the potentially unlawful harms these systems could inflict.
III. AUTONOMOUS WEAPONS AND THE LIMITS OF INTERNATIONAL HUMANITARIAN LAW
The Geneva Conventions and their Additional Protocols form the core of modern International Humanitarian Law (IHL). [21] Adopted in 1949 after World War II and entering into force in 1950, they establish minimum protections, standards of humane treatment, and fundamental guarantees for individuals affected by armed conflict. [22] The Conventions govern the treatment of civilians, prisoners of war, and soldiers rendered hors de combat (a soldier who cannot fight due to being sick, wounded, captured or incapacitated). [23] Although the 1949 Conventions have been universally ratified, their Additional Protocols have not. [24] Together, these instruments also provide for universal jurisdiction, reflecting the principle that certain offenses such as genocide, crimes against humanity, torture, and war crimes are so grave that they implicate the fundamental interests of the international community. As a result, individuals accused or convicted of such crimes may be prosecuted by any state party, regardless of nationality or where the offense occurred.
Codified in the 1977 Additional Protocol I to the Geneva Conventions are the principles of distinction, found in Articles 48, 51(2), and 52(2), and proportionality, articulated in Article 51(5)(b). [25] The principle of distinction assets that parties to an armed conflict must “at all times distinguish between the civilian population and combatants and between civilian objects and military objectives and accordingly shall direct their operations only against military objectives.” [26] Put simply, forces may not conduct attacks or employ methods that fail to discriminate between combatants and noncombatants. The principle of proportionality prohibits attacks against military objectives which are “expected to cause incidental loss of civilian life, injury to civilians, damage to civilian objects, or a combination thereof, which would be excessive in relation to the concrete and direct military advantage anticipated”. [27] In effect, proportionality seeks to ensure that military actions do not cause harm that is disproportionate to the strategic or tactical benefit pursued.
Fully autonomous weapon systems face significant, and perhaps insurmountable, obstacles in reliably distinguishing between lawful and unlawful targets, placing them in direct tension with the principle of distinction under IHL. These systems lack the qualitative human capacities required to make targeting judgments on contemporary battlefields where combatants often conceal their identity. [28] Distinguishing an active combatant from a civilian or from a wounded or surrendering soldier demands not only sophisticated sensory and processing capabilities, but also the uniquely human ability to assess intention. This type of assessment relies on subtle, context-dependent cues such as tone of voice, facial expression, and body language, all of which current autonomous systems are unable to interpret. [29]
This legal issue is exacerbated when considering the principle of proportionality, which prohibits attacks in which expected civilian harm outweighs anticipated military advantage. According to the U.S. Air Force, “proportionality in attack is an inherently subjective determination that will be resolved on a case-by-case basis.” [30] Even the slightest shift in circumstances can significantly alter what qualifies as a lawful response. To pre-program autonomous weapons to handle an effectively infinite range of scenarios is therefore a nearly impossible task. Proportionality is ultimately “a question of common sense and good faith for military commanders.” [31] To comply with both distinction and proportionality would require autonomous systems to exercise the judgment of a “reasonable military commander.” [32] As it stands, AWS do not possess such judgment, nor is it expected that they will in the near future. [33]
While the inability of AWS to comply with the Geneva Conventions’ principles of distinction and proportionality is both clear and consequential, an even more urgent concern is the profound gap in accountability created by their use. Fully autonomous weapons cannot serve as responsible defendants in legal proceedings designed to enforce deterrence or deliver retribution. [34] Although the humans involved in deploying or designing these systems, including operators, commanders, programmers, and manufacturers, could theoretically be held liable, in practice a host of legal obstacles makes this outcome highly unlikely. Commanders and operators could rarely be assigned direct responsibility for the autonomous actions of these weapons, except in exceptional circumstances where intent and control could be definitively established. [35] Likewise, programmers and manufacturers cannot reasonably be criminally punished for harms they neither specifically intended nor could foresee. [36] This absence of clear legal responsibility underscores a fundamental tension. Autonomous systems may act on the battlefield with lethal effect, yet no human or legal mechanism can fully account for the consequences, creating a scenario the Geneva Conventions were never designed to address.
IV. POLICY AND LEGAL RECOMMENDATIONS
As it stands, current International Humanitarian Law creates a significant accountability gap when applied to the use of AWS in armed conflict. IHL assumes that humans make the decisions that govern armed conflict, yet fully autonomous weapon systems operate without human judgment or intent. This leaves a profound accountability gap, as current legal frameworks cannot hold anyone responsible for the unlawful or disproportionate harms these systems may cause. The following policy and legal recommendations examine potential approaches to addressing this gap, highlighting existing proposals and assessing their effectiveness in ensuring that AWS deployment protects civilians and upholds the principles of humanity enshrined in the Geneva Conventions.
A body of literature exists in which proponents of fully autonomous weapons argue that there may be limited circumstances in which such systems could be both militarily valuable and capable of conforming to IHL. While some proponents note that fully autonomous weapons could be used lawfully under “limited circumstances,” such as in attacks on “nuclear-tipped mobile missile launchers where millions of lives are at stake,” others cite that “not every battlespace contains civilians.” [37], [38] While these proposals take the form of legally binding instruments or gradually developed informal standards, they share one key characteristic: none call for a complete prohibition on the development, production, and use of fully autonomous weapons. Instead, they advocate for partial restrictions, applying only to specific circumstances in which such weapons are deployed. [39]
Partial restrictions are insufficient to address the full range of risks posed by fully autonomous weapons. As noted by Human Rights Watch and Harvard Law School’s International Human Rights Clinic (IHRC), it is nearly impossible to limit the use of such weapons to narrowly defined scenarios. [40] Once fully autonomous systems are developed and integrated into warfare, there is no guarantee that states will comply with International Humanitarian Law. The pressures of battle or extreme circumstances could lead to uses that increase the risk of violations. [41] In today’s increasingly polarized and violent world, with 2024 marking the highest number of state-based armed conflicts in over seven decades, this scenario is highly plausible. [42]
Consistent with the recommendations of Human Rights Watch and IHRC, a complete prohibition on the development, production, and use of fully autonomous weapons through an international legally binding instrument, accompanied by national laws implementing the ban, is necessary. [43] A comprehensive ban offers advantages that partial restrictions cannot: it maximizes protection for civilians, creates a clear mandate that is easier to enforce, and establishes a powerful stigmatizing effect that sets a global standard, influencing even states that do not formally join the treaty.
Regarding accountability, legalizing even limited uses of fully autonomous weapons would inevitably produce situations in which responsibility is impossible to assign. Operators, commanders, programmers, and manufacturers are unlikely to be held liable under current legal frameworks, and even a strict liability scheme would only generate compensation without achieving the moral and deterrent functions of accountability. [44] By instituting a full ban, any state attempting to deploy or develop these systems would be clearly acting in violation of international law, reinforcing the principles of deterrence, retribution, and humanity. [45] While the act of enforceability is intrinsically limited in international law due to the nature of the practice, codifying a categorical prohibition still establishes a vital legal and moral boundary.
V. CONCLUSION
The threat posed by autonomous weapon systems is a pressing one. Bypassing the need for human intervention in the selection and application of force, these systems pose considerable humanitarian risks while simultaneously raising profound legal questions regarding accountability. Under international humanitarian law, accountability is crucial both to bring justice for victims and to deter future violations. As it stands, modern international humanitarian law is ill-equipped to address the problems AWS could create if engineered and deployed on the battlefield. While military experts and academics cite a slew of advantages posed by the technology, such as force multiplication, access to dangerous or inaccessible areas, and suitability for dull, dirty, or dangerous missions, we must ask ourselves: do these advantages trump the protection of civilians and the principle of humanity? [46] The Geneva Conventions, the principal doctrine of international humanitarian law, were created to establish rules for the humane treatment of individuals during armed conflict. To ensure that these rules remain meaningful and not merely words on paper, the development, production, and use of fully autonomous weapons must be unequivocally prohibited through an international legally binding instrument accompanied by national laws implementing the ban, for without such a measure, civilians will suffer and the principles of humanity will be hollow.
Endnotes
[1] Int’l Comm. of the Red Cross, The Geneva Conventions and Their Commentaries (last visited Nov. 6, 2025).
[2 ]Int’l Comm. of the Red Cross, Geneva Conventions (Nov. 6, 2025).
[3] Office of the High Commissioner for Human Rights, Geneva Convention (IV) Relative to the Protection of Civilian Persons in Time of War, accessed November 18, 2025, https://www.ohchr.org/en/instrumentsmechanisms/instruments/geneva-convention-relative-protection-civilian-persons-time-war.
[4] United Nations Regional Information Centre, UN Addresses AI and the Dangers of Lethal Autonomous Weapons Systems, 06/01/2025.
[5] Human Rights Watch, Mind the Gap: The Lack of Accountability for Killer Robots, April 9, 2015, Human Rights Watch, https://www.hrw.org/report/2015/04/09/mind-gap/lack-accountability-killer-robots.
[6] Silvia Borelli, “Casting Light on the Legal Black Hole: International Law and Detentions Abroad in the ‘War on Terror,’” International Review of the Red Cross 87, no. 857 (March 2005): [page number], https://www.icrc.org/sites/default/files/external/doc/en/assets/files/other/irrc_857_borelli.pdf.
[7] Int’l Comm. of the Red Cross, Autonomous Weapon Systems and International Humanitarian Law: Selected Issues (Oct. 13, 2025).
[8] ICRC, Autonomous Weapon Systems and IHL.
[9] ICRC, Autonomous Weapon Systems and IHL.
[10] Stockholm International Peace Research Institute (SIPRI), Limits on Autonomy in Weapon Systems: Identifying Practical Elements of Human Control (Stockholm: SIPRI, 2020), https://www.sipri.org/sites/default/files/2020- 06/2006_limits_of_autonomy.pdf.
[11] United Nations General Assembly, Report of the Secretary-General on [Title of Report, if there is one], A/79/88 (Date), accessed November 18, 2025, https://docs.un.org/en/A/79/88.
[12] P.W. Scharre & M.C. Horowitz, An Introduction to Autonomy in Weapons Systems: Working Paper 12 (Ctr. for a New Am. Sec., Feb. 2015).
[13] Amitai Etzioni and Oren Etzioni, “Pros and Cons of Autonomous Weapons Systems,” Military Review, May– June 2017,https://www.armyupress.army.mil/Journals/Military-Review/English-Edition-Archives/May-June2017/Pros-and-Cons-of-Autonomous-Weapons-Systems/.
[14] James R. Clapper Jr. et al., Unmanned Systems Roadmap: 2007-2032 (Washington, DC: Department of Defense [DOD], 2007), 19, accessed 28 March 2017, http://www.globalsecurity.org/intell/library/reports/2007/dodunmanned-systems-roadmap_2007-2032.pdf.
[15] Clapper et al., Unmanned Systems Roadmap, 19.
[16] Ronald C. Arkin, “The Case for Ethical Autonomy in Unmanned Systems,” Journal of Military Ethics 9, no. 4 (2010): 332–41, doi: 10.1080/15027570.2010.536402.
[17] Lieutenant Colonel Douglas A. Pryer, “The Rise of the Machines,” Military Review, March-April 2013, 12–23, https://www.armyupress.army.mil/Portals/7/militaryreview/Archives/English/MilitaryReview_20130430_art005.pdf.
[18] Pryer, “The Rise of the Machines,” 12.
[19] International Joint Conference on Artificial Intelligence, IJCAI-15 Conference Website, accessed November 19, 2025, https://ijcai-15.org/
[20] “Autonomous Weapons: An Open Letter from AI [Artificial Intelligence] & Robotics Researchers,” Future of Life Institute website, 28 July 2015, accessed 8 March 2017, http://futureoflife.org/open-letter-autonomous-weapons/.
[21] International Committee of the Red Cross, “Geneva Conventions and Their Commentaries,” accessed November 19, 2025, https://www.icrc.org/en/law-and-policy/geneva-conventions-and-their-commentaries.
[22] Hans-Peter Gasser, “Geneva Conventions I–IV (1949),” Max Planck Encyclopedia of Public International Law (MPEPIL), last updated December 2015, Oxford Public International Law, https://opil.ouplaw.com/display/10.1093/law:epil/9780199231690/law-9780199231690-e300.
[23] International Committee of the Red Cross (ICRC), “Rule 47: Persons Who Are Hors de Combat,” Customary International Humanitarian Law, accessed November 19, 2025, https://ihl‑databases.icrc.org/en/customaryihl/v1/rule47.
[24] ICRC, “Geneva Conventions and Their Commentaries.”
[25] International Committee of the Red Cross, “Fundamental Principles of IHL,” ICRC Casebook, accessed November 19, 2025, https://casebook.icrc.org/a_to_z/glossary/fundamental-principles-ihl.
[26] International Committee of the Red Cross, “Distinction,” ICRC Casebook, accessed November 19, 2025, https://casebook.icrc.org/a_to_z/glossary/distinction.
[27] International Committee of the Red Cross, “Proportionality,” ICRC Casebook, accessed November 19, 2025, https://casebook.icrc.org/a_to_z/glossary/proportionality.
[28] United Nations, Letter dated 19 March 2025 from the Permanent Representative of the United States of America to the United Nations addressed to the Secretary-General, A/79/88 (March 19, 2025), https://docs.un.org/en/A/79/88.
[29] Human Rights Watch, Autonomous Weapons Systems and Digital Decision-Making: A Hazard to Human Rights, April 28, 2025, https://www.hrw.org/report/2025/04/28/hazard-human-rights/autonomous-weapons-systems-anddigital-decision-making.
[30] U.S. Air Force Judge Advocate General’s Department, Air Force Operations and the Law: A Guide for Air and Space Forces, 1st ed. (Washington, DC: U.S. Air Force, 2002), 27.
[31] HRW, Mind the Gap: The Lack of Accountability for Killer Robots.
[32] International Committee of the Red Cross (ICRC), Commentary on the Additional Protocols of 8 June 1977 to the Geneva Conventions of 12 August 1949 (Geneva: Martinus Nijhoff Publishers, 1987), 679, 682.
[33] HRW, Mind the Gap: The Lack of Accountability for Killer Robots.
[34] HRW, Mind the Gap: The Lack of Accountability for Killer Robots.
[35] Library of Congress, “Legal Aspects of Unmanned Systems, Part 2: Lethal Autonomous Weapons,” In Custodia Legis (blog), November 18, 2015, https://blogs.loc.gov/law/2015/11/legal-aspects-of-unmanned-systems-part-2- lethal-autonomous-weapons-2/.
[36] Eleni Nerantzi, “‘Hard AI Crime’: The Deterrence Turn,” Crime, Law and Social Change (2024), https://pmc.ncbi.nlm.nih.gov/articles/PMC11368829/.
[37] Paul Scharre, “Reflections on the Chatham House Autonomy Conference,” Lawfare (blog), March 3, 2014, accessed April 20, 2014, http://www.lawfareblog.com/2014/03/guest-post-reflections-on-the-chatham-houseautonomy-conference/.
[38] Michael N. Schmitt, “Autonomous Weapon Systems and International Humanitarian Law: A Reply to the Critics,” Harvard National Security Journal Features (2013), http://harvardnsj.org/wp-content/uploads/2013/02/SchmittAutonomous-Weapon-Systems-and-IHL-Final.pdf (accessed April 1, 2015), p. 11.
[39] Kenneth Anderson and Matthew Waxman, “Law and Ethics for Autonomous Weapon Systems: Why a Ban Won’t Work and How the Laws of War Can,” Jean Perkins Task Force on National Security and Law (2013), 22, accessed March 31, 2015, http://media.hoover.org/sites/default/files/documents/AndersonWaxman_LawAndEthics_r2_FINAL.pdf.
[40] International Human Rights Clinic (Harvard Law School), “’Killer Robots’ Threaten Human Rights During War, Peace,” accessed November 20, 2025, https://humanrightsclinic.law.harvard.edu/killer-robots-threaten-human-rightsduring-war-peace/.
[41] HRW, Mind the Gap.
[42] Peace Research Institute Oslo, “New Data Shows Conflict at Historic High as U.S. Signals Retreat from World Stage,” June 9, 2025, accessed November 20, 2025, https://www.prio.org/news/3616.
[43] Harvard Law Human Rights Clinic, “’Killer Robots’ Threaten Human Rights.”
[44] HRW, Mind the Gap.
[45] HRW, Mind the Gap.
[46] Clapper et al., Unmanned Systems Roadmap, 19.