Introduction
The convergence of AI and armed conflict presents one of the most prominent legal challenges of the twenty-first century. International law is struggling to keep pace with technology as militaries around the world rush to develop autonomous weapons systems (AWS), lethal autonomous weapons systems (LAWS) and AI-assisted targeting platforms. The structural problem lies in the very foundations of international humanitarian law (IHL), which was created in an era when decisions in war were made by the human mind. Its core principles of distinction, proportionality, precaution and military necessity rest on human judgment and experience, and AI systems may not be able to apply them consistently.
The Martens Clause fits into this uncertain legal environment in a unique way. First drafted in the preamble to the 1899 Hague Convention II as a diplomatic compromise between smaller and great powers, and later carried into the wider body of IHL, the Clause has increasingly been read by some as a substantive standard with independent legal force.1 It has been acknowledged by the International Court of Justice (ICJ) and the International Committee of the Red Cross (ICRC). The use of AI in warfare poses a challenge under the Martens Clause in three important respects. First, weapons systems that can select and engage targets without meaningful human control create acute accountability gaps that existing treaties have not addressed. Second, algorithmic decision-making in targeting, especially when based on historically biased data sets, can systematically undervalue civilian life, running counter to the principle of humanity. Third, the lack of transparency in AI systems (the “black box” issue) hinders the very review and human oversight that IHL demands.
The remainder of this paper is organized into five parts. Part II explores the historical evolution and application of the Martens Clause. Part III explores the current state of AI in warfare, including the types of systems and capabilities that raise legal concerns. Part IV examines the application of the Clause to autonomous weapons systems, including accountability gaps, limitations based on dignity, and public conscience. Part V considers the Clause’s contribution to the emerging international regulatory agenda. Part VI concludes with recommendations for translating the Clause’s normative content into binding legal obligations.
The Martens Clause: historical development and legal interpretation
A. Origins and textual evolution
The Martens Clause owes its name to Fyodor Fyodorovich Martens (also known as Friedrich von Martens), the Russian jurist and diplomat who introduced its foundational language at the 1899 Hague Peace Conference. Its inclusion resolved a contentious debate over the legal status of irregular combatants, a question on which small and great powers could not agree. Rather than leave the matter unaddressed, Martens proposed a clause acknowledging that gaps in treaty law would be filled by principles external to the treaty text itself.2
The Clause has been included in a number of instruments over the past century. The 1899 text referred to principles of international law “as they result from the usages established between civilized nations, from the laws of humanity, and the requirements of the public conscience.”3 In the English text of Hague Convention IV of 1907, the same French words read “as they result from the usages established among civilized peoples, from the laws of humanity, and the dictates of the public conscience.”4 The Clause then reappeared in the denunciation provisions common to the four Geneva Conventions of 19495 and in Article 1(2) of Additional Protocol I of 1977, which speaks of the principles of international law “derived from established custom, from the principles of humanity and from the dictates of public conscience.”6
The spirit of the Clause is also incorporated in Additional Protocol II, governing non-international armed conflicts, whose preamble recalls that “in cases not covered by the law in force, the human person remains under the protection of the principles of humanity and the dictates of the public conscience.” The Clause is also reflected in the preamble of the 1980 Convention on Certain Conventional Weapons, and the preamble of the 1997 Ottawa Mine Ban Treaty stresses “the role of public conscience in furthering the principles of humanity,” demonstrating the Clause’s cross-cutting normative reach.7
B. Competing interpretations
The Martens Clause has been one of the most controversial provisions of IHL, even in the face of its widespread textual observance. Three main interpretations have developed in practice and scholarship. The minimalist view understands the Clause as a hermeneutic cue to look beyond the text of the treaty to custom and general principles, and thus does not give it independent legal meaning.8 Those who support this position highlight state sovereignty and the consensual basis of international law, which they claim cannot impose binding obligations unless they are supported by their own source of normativity.
An intermediate approach, championed by the ICRC, treats the Clause as a dynamic criterion that adapts to evolving circumstances, including technological developments.9 It draws support from the ICJ’s 1996 Nuclear Weapons Advisory Opinion, in which the Court observed that the Clause “has proved to be an effective means of addressing the rapid evolution of military technology,”10 although the Court did not expressly adopt any one interpretation. On this view, the Clause has considerable regulatory potential in areas of legal lacunae. A maximalist approach, taken by a group of writers whom Antonio Cassese surveys, holds that the Clause has expanded the sources of international law, so that the laws of humanity and the dictates of public conscience can themselves generate obligations for States, even for those not party to a treaty.11 Cassese himself rejects that reading as the most radical of the three, and confines the Clause to two roles within the existing sources: a guide to interpreting humanitarian rules, and a means of loosening the requirement of usus while raising the rank of opinio.12 The maximalist interpretation has been challenged as difficult to apply in court, but it best represents the Clause’s most salient potential restrictions on new weapons technologies.
C. Judicial recognition
The ICJ’s Advisory Opinion on the Legality of the Threat or Use of Nuclear Weapons (1996) is the most significant judicial consideration of the Clause. The Court treated the Clause as an expression of pre-existing customary law binding on all States,13 and pointed to it, “whose continuing existence and applicability is not to be doubted, as an affirmation that the principles and rules of humanitarian law apply to nuclear weapons.”14 The Court did not clarify the normative scope of the Clause, which leaves room for its application to AI systems.
With regard to emerging technologies, the treatment of the dictates of public conscience in Prosecutor v. Kupreškić (2000) is especially relevant. The Trial Chamber accepted that the Clause may not be taken to elevate the principles of humanity and the dictates of public conscience to the rank of independent sources of international law.15 It nevertheless held that, as a result of the Clause, “principles of international humanitarian law may emerge through a customary process under the pressure of the demands of humanity or the dictates of public conscience, even where State practice is scant or inconsistent.”16 That reasoning suggests that, even before States have adopted comprehensive regulation in this area, robust and consistent public moral condemnation of AI-driven atrocities may help to crystallize binding custom.
Artificial intelligence in contemporary warfare: a legal taxonomy
A. Autonomous weapons systems and LAWS
Autonomous weapons systems comprise a range of technologies with varying degrees of human control. In the ICRC’s description, they “select and apply force to targets without human intervention.” The most legally problematic are fully autonomous LAWS (human-out-of-the-loop systems, in which lethal decisions are made entirely autonomously).17 Reported examples of weapons with autonomous functions include the South Korean SGR-A1 sentry robot deployed in the demilitarized zone; the Israeli Harpy loitering munition, which can detect and destroy radar installations; the Turkish Kargu-2 drone, reportedly used in Libya in 2020; and the U.S. Navy’s Phalanx close-in weapon system (CIWS), which can autonomously engage incoming missiles.18 These and other systems are promoted as enhancing military precision, but critics point to the basic problem of how to program legal judgment into an algorithmic system.
B. AI-assisted targeting and predictive analytics
Beyond standalone systems, AI is also being used in human-assisted targeting workflows. The reported use by the Israel Defense Forces of the AI system “the Gospel” (Habsora) for targeting in the Gaza operations that began in 2023 attracted international attention. Reports indicated that the system could generate targeting recommendations at unprecedented speed, and critics argue that this reduces the time available for legal review.19 The U.S. Project Maven initiative, which uses machine learning to process intelligence, surveillance and reconnaissance data, also highlights the acceleration of the targeting cycle through AI, which compresses the time for legal review. This hybrid human-machine model creates specific IHL concerns. The pace of AI-driven targeting recommendations can leave human operators unable to take real ownership of the decision-making process, leading to what critics describe as “automation bias”: the tendency of human operators to accept AI recommendations without independent thought.20
C. The accountability gap
The “accountability gap” is a potential challenge for the law of war that arises when an autonomous system commits what would be a war crime, yet no human actor is held accountable.21 Traditional IHL accountability mechanisms assume a chain of command: an individual commits a war crime, a commander is held accountable for subordinates’ actions, and a State is held accountable for war crimes committed by its organs. Each link in that chain is stressed when an autonomous system makes a wrong call that results in the killing of civilians. The software developers who created the algorithm might not have considered the particular environment in which it is used. The commander who activated the system might have trusted in assurances of reliability that she was technically unable to check. The manufacturer can point to design specifications. The consequence is a diffusion of responsibility that leaves every actor effectively immune, an effect highly counterproductive to the aims of IHL. This accountability gap is not just a theoretical one. The history of drone warfare, where critics argue that existing oversight has not secured accountability for civilian deaths, shows how accountability can be undermined in practice. These pressures will only grow with increasingly autonomous AI systems.22
The Martens Clause as applied to AI in warfare
A. Principles of humanity and algorithmic decision-making
In the absence of specific rules, the “principles of humanity” in the Martens Clause contain a fundamental affirmation of the dignity of the human person that limits the methods and means of war. On this reading, these principles act as a floor for IHL: because the ICJ has treated the Clause as customary law binding on all States, the duties it carries do not depend on membership of a particular treaty.23 When considered in the context of AI systems, the principles of humanity raise at least two specific legal obligations. First, they insist that the taking of human life in armed conflict retain a fundamentally human character. This is not just a matter of procedure (that a human presses a button) but a substantive one: the decision must rest on real human deliberation that takes account of the moral nature of taking life, the specific situation of the particular target, and the proportionality calculus that IHL requires. However advanced the algorithms used in making decisions, they are not moral agents.24 Second, humanity demands that weapons systems treat people as ends in themselves, not as means to military ends. This Kantian strand of humanitarian law is manifested in the ban on superfluous injury and in the growing understanding that there are dignity-based limits on psychological and cognitive warfare. AI systems that are incapable of understanding human suffering or of making real-time morally responsive adjustments do not meet this standard. The ICRC’s 2021 position on autonomous weapon systems reflects this concern: it identifies the “loss of human control and judgement in the use of force and weapons” as a serious concern, and recommends that unpredictable autonomous weapon systems, and the use of autonomous weapon systems to target human beings, be ruled out.25 The ICRC’s positions are not legally binding, but they carry great weight given its standing in IHL.
B. Dictates of public conscience and emerging norm formation
The “dictates of public conscience” element of the Martens Clause is simultaneously the most legally uncertain and the most politically potent component of the provision. It arguably introduces a form of democratic legitimacy into IHL, acknowledging that the moral sensibilities of affected communities, and not merely of States negotiating at diplomatic conferences, shape the content of international law.26
Contemporary public opinion on autonomous weapons has been consistent. Opinion surveys commissioned by the Campaign to Stop Killer Robots, governmental statements at meetings of the Convention on Certain Conventional Weapons (CCW), and the 2018 protest by Google employees over Project Maven, in which thousands expressed concern and about a dozen reportedly resigned, all reflect a transnational public conscience opposed to the delegation of lethal decision-making to machines.27
Under the Kupreškić reasoning, this consistent public moral rejection may contribute to the crystallization of customary law norms even ahead of formal state practice. The UN Secretary-General’s 2023 policy brief A New Agenda for Peace recommended concluding, by 2026, a legally binding instrument to prohibit lethal autonomous weapon systems that function without human control or oversight and cannot be used in compliance with IHL.28 A joint statement by 70 States at the General Assembly in October 2022, and the adoption in December 2023 of General Assembly Resolution 78/241 with 152 votes in favour, suggest that public conscience is already generating institutional momentum.29
Critics of this expansive reading argue that “public conscience” is too indeterminate a concept to generate legal obligations and that it risks importing majoritarian politics into a legal framework designed to constrain state behaviour. These concerns are legitimate and counsel caution. However, the appropriate response is not to discard the dictates-of-public-conscience standard entirely but to develop criteria for identifying when it is sufficiently widespread and consistent to bear normative weight. Factors such as cross-regional consensus, articulation by credible non-state actors, and resonance with recognized humanitarian values can provide the necessary discriminating criteria.
C. The accountability gap and the Martens Clause
The Martens Clause’s application to the accountability gap is perhaps its most practically significant dimension. Where treaty law has not resolved attribution and responsibility for AI-enabled violations, the principles of humanity and public conscience can supply the normative foundation for developing new accountability frameworks.
At a minimum, the Clause supports three accountability-related propositions. First, States bear an obligation under the principles of humanity to ensure meaningful legal review before deploying novel weapons systems, an obligation reflected in Article 36 of Additional Protocol I, which requires review of new weapons to determine their conformity with international law. The principles of humanity can be interpreted to require that Article 36 reviews of AI systems include not only technical assessments but also ethical review informed by public conscience standards.30
Second, the principles of humanity support a presumption in favour of human control over lethal targeting decisions. This presumption does not preclude all autonomous systems, but it places the burden of justification on those who would deploy fully autonomous LAWS. In the absence of a demonstrated capability to comply with IHL requirements, deployment should be prohibited.
Third, the dictates of public conscience support the development of command responsibility doctrines adapted to AI systems, including the proposition that commanders who deploy systems known to be incapable of IHL compliance bear individual criminal responsibility for resulting violations. This extension of command responsibility is not without doctrinal challenge, but the Martens Clause provides normative grounding for its development.31
The Martens Clause and the international regulatory agenda
A. The CCW process and its limitations
The principal multilateral forum for LAWS regulation has been the Group of Governmental Experts (GGE) on Lethal Autonomous Weapons Systems, established under the CCW in 2016. The GGE has produced a series of “guiding principles” affirming human responsibility and accountability, the applicability of IHL, and the importance of human-machine interaction. However, the GGE has not produced a binding instrument. Its final report, adopted on 4 September 2026, agreed a characterization of LAWS, without prejudice to any future understanding, and a set of elements of an instrument “without prejudging its nature,” which it referred to the Seventh Review Conference of the CCW in November 2026.32
The CCW’s consensus-based decision-making procedure allows just a few States, or even one, to block an agreement, and critics note that a handful of military powers, including the United States, Russia and Israel, have rejected proposals for a new treaty.33 This structural limitation has led a growing coalition of States, non-governmental organizations and legal scholars to advocate a new legally binding treaty on autonomous weapons negotiated outside the CCW framework, analogous to the processes that produced the 1997 Ottawa Mine Ban Treaty and the 2008 Convention on Cluster Munitions.34
The Martens Clause is directly relevant to this strategic choice. The ICJ’s treatment of the Clause as customary law supports the argument that, even in the absence of a new instrument, the principles of humanity and public conscience constrain state behaviour with respect to LAWS, although that argument is contested.35 This normative baseline strengthens the hand of advocates for a new treaty by supporting their position that States cannot simply wait out the regulatory process: on this view, they are already bound by standards that flow from the Clause.
B. Regional and national frameworks
In the absence of a global instrument, regional and national regulatory initiatives are emerging. The European Union’s Artificial Intelligence Act, proposed in 2021 and adopted in 2024, is confined to non-military AI: Article 2(3) excludes systems used exclusively for military, defence or national security purposes. It nevertheless establishes a risk-based framework that includes the prohibition of certain AI practices, and that logic may offer a model for military applications.36 The EU’s strategic autonomy agenda has generated internal discussions about autonomous military AI that may produce binding European-level standards.
At the national level, U.S. Department of Defense Directive 3000.09 (2023 update) requires that autonomous weapons systems be designed to allow commanders and operators to exercise “appropriate levels of human judgment over the use of force.”37 The United Kingdom has similarly stated that “there must always be human oversight and authority in the decision to strike” and that responsibility lies “with the commanders and operators.” China has called for a treaty banning the use of lethal autonomous weapons systems, but not their development or production, while also being among the States most advanced in pursuing such systems, a gap between declaratory and operational policy that the Martens Clause’s public conscience standard helps to expose.38
National legal review mechanisms under Article 36 of Additional Protocol I provide another avenue for Martens Clause standards to shape AI weapons development. Several States have established formal legal review processes for new weapons; incorporating Martens Clause criteria, including the principles of humanity and public conscience, into such reviews would operationalize the Clause at the procurement stage, before deployment.
C. Toward a binding instrument: the contribution of the Martens Clause
The Martens Clause’s most enduring contribution to the regulatory agenda may be its capacity to serve as the normative foundation for a new binding instrument. Any treaty on autonomous weapons will require agreement on at least three elements: a definitional framework for regulated systems; substantive behavioural rules (what is prohibited and what is permissible with conditions); and accountability mechanisms for violations.
A reading of the Clause that focuses on the substance of human control, and not merely its form, offers a principled basis for distinguishing regulated from non-regulated systems. A definition based on the principles of humanity would go beyond technical parameters, which can easily be circumvented by design choices, to focus on whether a system can make the moral-legal judgments that IHL requires. On the substantive rules, the Clause supports a prohibition of fully autonomous LAWS that cannot comply with IHL, and meaningful human control of targeting in all other instances. Elaborating the content of “meaningful control” would require a treaty but, in the absence of one, the Martens Clause provides a minimum content: control must be substantive and not just formal; it must allow for genuine deliberation; and it must be exercised by people who have a technical understanding of the system at stake as well as an understanding of the operational context. On accountability, the Clause’s reference to the principles of humanity supports holding commanders accountable for using non-compliant AI systems and holding States accountable for failing to maintain adequate legal review processes. Some of these doctrines can be developed in the course of negotiating a treaty, but the Martens Clause already provides a normative basis for them.
Conclusion and recommendations
The Martens Clause was originally designed as a diplomatic tool but has become one of the most formative instruments of IHL. It refers to the principles of humanity and the dictates of public conscience, which are not completely captured by treaty-specific rules but represent a moral and legal reality. This commitment is being put to the ultimate test as AI is used in war. Autonomous weapons systems that choose and launch attacks without human input, legal review processes that may be assisted by AI but must be completed in fractions of a second, and algorithmic systems trained on data that may systematically undervalue certain lives all challenge the principles of humanity that the Martens Clause is meant to help preserve. This paper has suggested that the Clause, if correctly interpreted, gives rise to three types of obligation for the use of AI in warfare: a substantive obligation, based on the principles of humanity, to retain a meaningful human element in targeting decisions; a norm-generative role in producing binding customary restrictions on LAWS; and a normative role in supporting accountability frameworks that fill the gap left by existing treaty law.
The following recommendations are based on this analysis:
1. States should formally integrate Martens Clause considerations, such as consistency with the principles of humanity and public conscience, into their weapons review processes for AI-enabled weapons.
2. The international community should develop an instrument on autonomous weapons, in addition to the CCW, with the norms of the Martens Clause reflected in both its preamble and its substantive provisions.
3. An authoritative legal study, commissioned by the ICRC and the UN Secretary-General, should further develop the content of the Martens Clause’s requirements as applied to specific AI weapons capabilities, to inform state practice and judicial interpretation.
4. National courts and international criminal tribunals should develop a command responsibility doctrine applicable to AI systems, grounded in the Martens Clause’s requirement to observe the principles of humanity, which bind all without distinction and irrespective of the machinery and equipment used.
5. Civil society organizations, professional organizations and academic institutions should continue to express and record the “dictates of public conscience” concerning autonomous weapons, noting their potential role in the formation of custom.
The Martens Clause cannot on its own resolve the regulatory issues surrounding the use of AI in warfare. But it offers something just as essential in a time of rapid technological evolution: a reminder that the role of law is not simply to record the promises States have made, but to safeguard what human dignity demands. In that regard, the compromise of 1899 remains as relevant as ever to the technologies it is now needed to manage.
*****
Footnotes
1. Convention (II) with Respect to the Laws and Customs of War on Land pmbl., July 29, 1899, 32 Stat. 1803.
2. Theodor Meron, The Martens Clause, Principles of Humanity, and Dictates of Public Conscience, 94 Am. J. Int’l L. 78, 80 (2000); Antonio Cassese, The Martens Clause: Half a Loaf or Simply Pie in the Sky?, 11 Eur. J. Int’l L. 187, 193–98 (2000).
3. Convention (II) with Respect to the Laws and Customs of War on Land, supra note 1, pmbl.; see also Meron, supra note 2, at 80.
4. Convention (IV) Respecting the Laws and Customs of War on Land pmbl., Oct. 18, 1907, 36 Stat. 2277; see Cassese, supra note 2, at 187–88 & n.1 (French text and English translation).
5. Geneva Convention (I) for the Amelioration of the Condition of the Wounded and Sick in Armed Forces in the Field art. 63, Aug. 12, 1949, 75 U.N.T.S. 31; Geneva Convention (II) for the Amelioration of the Condition of Wounded, Sick and Shipwrecked Members of Armed Forces at Sea art. 62, Aug. 12, 1949, 75 U.N.T.S. 85; Geneva Convention (III) Relative to the Treatment of Prisoners of War art. 142, Aug. 12, 1949, 75 U.N.T.S. 135; Geneva Convention (IV) Relative to the Protection of Civilian Persons in Time of War art. 158, Aug. 12, 1949, 75 U.N.T.S. 287.
6. Protocol Additional to the Geneva Conventions of 12 August 1949, and Relating to the Protection of Victims of International Armed Conflicts (Protocol I) art. 1(2), June 8, 1977, 1125 U.N.T.S. 3.
7. Protocol Additional to the Geneva Conventions of 12 August 1949, and Relating to the Protection of Victims of Non-International Armed Conflicts (Protocol II) pmbl., June 8, 1977, 1125 U.N.T.S. 609; Convention on Prohibitions or Restrictions on the Use of Certain Conventional Weapons Which May Be Deemed to Be Excessively Injurious or to Have Indiscriminate Effects pmbl., Oct. 10, 1980, 1342 U.N.T.S. 137; Convention on the Prohibition of the Use, Stockpiling, Production and Transfer of Anti-Personnel Mines and on Their Destruction pmbl., Sept. 18, 1997, 36 I.L.M. 1507.
8. See Yoram Dinstein, The Conduct of Hostilities under the Law of International Armed Conflict 12–13 (3d ed. 2016).
9. Int’l Comm. of the Red Cross, Autonomous Weapon Systems: Implications of Increasing Autonomy in the Critical Functions of Weapons 82 & n.72 (2016); see also Int’l Comm. of the Red Cross, A Guide to the Legal Review of New Weapons, Means and Methods of Warfare: Measures to Implement Article 36 of Additional Protocol I of 1977 § 1.2.2.3 (2006).
10. Legality of the Threat or Use of Nuclear Weapons, Advisory Opinion, 1996 I.C.J. 226, ¶ 78 (July 8).
11. Cassese, supra note 2, at 190–92.
12. Id. at 187, 193, 212–14; see generally Antonio Cassese, International Law 168–69 (2d ed. 2005).
13. Legality of the Threat or Use of Nuclear Weapons, supra note 10, ¶ 84.
14. Id. ¶ 87.
15. Prosecutor v. Kupreškić, Case No. IT-95-16-T, Judgement, ¶ 525 (Int’l Crim. Trib. for the Former Yugoslavia Jan. 14, 2000).
16. Id. ¶ 527.
17. Int’l Comm. of the Red Cross, ICRC Position on Autonomous Weapon Systems (May 12, 2021), https://www.icrc.org/en/document/icrc-position-autonomous-weapon-systems.
18. Panel of Experts on Libya, Final Report, ¶ 63, U.N. Doc. S/2021/229 (2021) (Kargu-2); Int’l Comm. of the Red Cross, Autonomous Weapon Systems: Technical, Military, Legal and Humanitarian Aspects 65 nn.40, 43, 68 n.56 (2014) (Phalanx, SGR-A1 and Harpy); Human Rights Watch, Losing Humanity: The Case Against Killer Robots (2012), https://www.hrw.org/report/2012/11/19/losing-humanity/case-against-killer-robots (SGR-1 along the demilitarized zone; Harpy); see generally Human Rights Watch, Stopping Killer Robots: Country Positions on Banning Fully Autonomous Weapons and Retaining Human Control 3 (2020), https://www.hrw.org/report/2020/08/10/stopping-killer-robots/country-positions-banning-fully-autonomous-weapons-and.
19. Yuval Abraham, ‘A Mass Assassination Factory’: Inside Israel’s Calculated Bombing of Gaza, +972 Magazine (Nov. 30, 2023), https://www.972mag.com/mass-assassination-factory-israel-calculated-bombing-gaza/ (last visited May 30, 2026).
20. Paul Scharre, Army of None: Autonomous Weapons and the Future of War 178–82 (2018).
21. Robert Sparrow, Killer Robots, 24 J. Applied Phil. 62, 69–72 (2007).
22. Sarah Kreps, Drones: What Everyone Needs to Know 89–95 (2016).
23. Legality of the Threat or Use of Nuclear Weapons, supra note 10, ¶ 84; see generally Meron, supra note 2, at 88–89.
24. Peter Asaro, On Banning Autonomous Weapon Systems: Human Rights, Automation, and the Dehumanization of Lethal Decision-Making, 94 Int’l Rev. Red Cross 687, 694, 708 (2012).
25. Int’l Comm. of the Red Cross, ICRC Position on Autonomous Weapon Systems, supra note 17.
26. Rupert Ticehurst, The Martens Clause and the Laws of Armed Conflict, 37 Int’l Rev. Red Cross 125, 132–33 (1997).
27. Ipsos, Global Survey Highlights Continued Opposition to Fully Autonomous Weapons (Feb. 2, 2021), https://www.ipsos.com/en-us/global-survey-highlights-continued-opposition-fully-autonomous-weapons; Human Rights Watch, Stopping Killer Robots: Country Positions on Banning Fully Autonomous Weapons and Retaining Human Control, supra note 18; Tajha Chappellet-Lanier, Google Employees Resign in Protest Against Air Force’s Project Maven, FedScoop (May 14, 2018), https://fedscoop.com/google-employees-resign-project-maven/.
28. U.N. Secretary-General, Our Common Agenda Policy Brief 9: A New Agenda for Peace 27 (July 2023), https://www.un.org/sites/un2.un.org/files/our-common-agenda-policy-brief-new-agenda-for-peace-en.pdf.
29. G.A. Res. 78/241 (Dec. 22, 2023); Stop Killer Robots, The Story So Far, https://www.stopkillerrobots.org/the-story-so-far.
30. Protocol I, supra note 6, art. 36; Int’l Comm. of the Red Cross, A Guide to the Legal Review of New Weapons, Means and Methods of Warfare, supra note 9, § 1.2.2.3.
31. See Michael N. Schmitt & Jeffrey S. Thurnher, “Out of the Loop”: Autonomous Weapon Systems and the Law of Armed Conflict, 4 Harv. Nat’l Sec. J. 231, 276–78 (2013) (accountability of programmers, operators and commanders); but see id. at 275–76 (arguing that the Clause applies only in the absence of treaty law and that its significance has diminished).
32. Rep. of the 2019 Sess. of the Grp. of Governmental Experts on Emerging Technologies in the Area of Lethal Autonomous Weapons Systems, Annex IV, U.N. Doc. CCW/GGE.1/2019/3 (Sept. 25, 2019); Rep. of the 2024-2025-2026 Sess. of the Grp. of Governmental Experts on Emerging Technologies in the Area of Lethal Autonomous Weapons Systems, ¶¶ 23, 25, 45–46, U.N. Doc. CCW/GGE.1/2026/3 (Sept. 14, 2026).
33. Human Rights Watch, Stopping Killer Robots: Country Positions on Banning Fully Autonomous Weapons and Retaining Human Control, supra note 18.
34. Id.; Convention on Cluster Munitions, Dec. 3, 2008, 48 I.L.M. 357 (2009).
35. Legality of the Threat or Use of Nuclear Weapons, supra note 10, ¶ 84; see generally Geoffrey S. Corn et al., The War on Terror and the Laws of War: A Military Perspective 201–04 (2d ed. 2015); but see Kupreškić, supra note 15, ¶ 525; Schmitt & Thurnher, supra note 31, at 275–76.
36. Regulation (EU) 2024/1689 of the European Parliament and of the Council of 13 June 2024 Laying Down Harmonised Rules on Artificial Intelligence (Artificial Intelligence Act), arts. 2(3), 5, 2024 O.J. (L 2024/1689).
37. U.S. Dep’t of Def., Directive 3000.09, Autonomy in Weapon Systems (Jan. 25, 2023).
38. Human Rights Watch, Stopping Killer Robots: Country Positions on Banning Fully Autonomous Weapons and Retaining Human Control, supra note 18.