1 Berkowitz, Bruce, “Sea Power in the Robotic Age”, Issues in Science and Technology, Vol. 30, No. 2, 2014Google Scholar.

2 Barrett, Tim, The Navy and the Nation: Australia's Maritime Power in the 21st Century, Melbourne University Press, Melbourne, 2017, pp. 53–55Google Scholar; Clark, Bryan, The Emerging Era in Undersea Warfare, CSBA, Washington, DC, 2015Google Scholar; US Department of Defense (DoD), Unmanned Systems Integrated Roadmap: FY2013–2038, Washington, DC, 2014 (UMS Roadmap)Google Scholar; Navy, US, The Navy Unmanned Surface Vehicle (USV) Master Plan, Washington, DC, 2007 (USV Master Plan)Google Scholar.

3 Sparrow, Robert and Lucas, George, “When Robots Rule the Waves?”, Naval War College Review, Vol. 69, No. 4, 2016Google Scholar, offers the most comprehensive discussion of the ethical issues raised by the use of UMS in war at sea of which we are aware to date. The most extensive discussion of the legal issues of which we are aware is Capt. Norris, Andrew, Legal Issues Relating to Unmanned Maritime Systems, US Naval War College, Newport, RI, 2013Google Scholar. See also Gogarty, Brendan and Hagger, Meredith, “The Laws of Man over Vehicles Unmanned: The Legal Response to Robotic Revolution on Sea, Land and Air”, Journal of Law, Information and Science, Vol. 19, No. 1, 2008Google Scholar; Henderson, Andrew H., “Murky Waters: The Legal Status of Unmanned Undersea Vehicles”, Naval Law Review, Vol. 53, 2006Google Scholar; McLaughlin, Rob, “Unmanned Naval Vehicles at Sea: USVs, UUVs, and the Adequacy of the Law”, Journal of Law Information and Science, Vol. 21, No. 2, 2011CrossRefGoogle Scholar. The use of autonomous weapon systems in war at sea would, of course, raise issues discussed in the larger debate about the ethics of the development and deployment of autonomous weapon systems. For a recent survey and further sources, see Sparrow, Robert, “Robots and Respect: Assessing the Case against Autonomous Weapon Systems”, Ethics and International Affairs, Vol. 30, No. 1, 2016CrossRefGoogle Scholar.

4 Geneva Convention (II) for the Amelioration of the Condition of Wounded, Sick and Shipwrecked Members of Armed Forces at Sea of 12 August 1949, 75 UNTS 85 (entered into force 21 October 1950), Art. 18. This expectation is both longstanding as an article of law (for example, Hague Convention (X) for the Adaptation to Maritime War of the Principles of the Geneva Convention, 1907, Art. 16(1)), and broadly cast as an obligation, as in, for instance, International Committee of the Red Cross (ICRC), Commentary on the Second Geneva Convention: Convention (II) for the Amelioration of the Condition of the Wounded, Sick and Shipwrecked Members of Armed Forces at Sea, 2nd ed., 2017 (ICRC Commentary on GC II), para. 1619Google Scholar. Importantly, of course, an obligation to rescue also exists in peacetime: see International Convention for the Safety of Life at Sea, 1184 UNTS 278, 1 November 1974. See also Davies, Martin, “Obligations and Implications for Ships Encountering Persons in Need of Assistance at Sea”, Pacific Rim Law and Policy Journal, Vol. 12, No. 1, 2003Google Scholar; Papanicolopulu, Irini, “The Duty to Rescue at Sea, in Peacetime and in War: A General Overview”, International Review of the Red Cross, Vol. 98, No. 902, 2016CrossRefGoogle Scholar; Peltz, Robert D., “Adrift at Sea – The Duty of Passing Ships to Rescue Stranded Seafarers”, Tulane Maritime Law Journal, Vol. 38, No. 2, 2014Google Scholar; Tondini, Matteo, “The Legality of Intercepting Boat People Under Search and Rescue and Border Control Operations with Reference to Recent Italian Interventions in the Mediterranean Sea and the Ecthr Decision in the Hirsi Case”, Journal of International Maritime Law, Vol. 18, No. 1, 2012Google Scholar.

5 DoD, DoD Dictionary of Military and Associated Terms, Washington, DC, 2017, p. 246Google Scholar.

6 UMS Roadmap, above note 2, p. 8.

7 USV Master Plan, above note 2, p. 7.

8 US Navy, The Navy Unmanned Undersea Vehicle (UUV) Master Plan, Washington, DC, 2004, p. 4Google Scholar. Beyond conventional hull configurations, USVs “include hydrofoils and semi-submersible (i.e., continuously snorkeling) crafts”: see USV Master Plan, above note 2, p. 7.

9 UMS Roadmap, above note 2, p. 109.

10 Some information about the British, Russian and Chinese programmes is available at: www.royalnavy.mod.uk/search?q=Maritime+Autonomy+Surface+Testbed; “United Kingdom Naval Drones”, Naval Drones, available at: www.navaldrones.com/United-Kingdom-Naval-Drones.html; “China's Naval Drones”, Naval Drones, available at: www.navaldrones.com/China.html; “Russia's Naval Drones”, Naval Drones, available at: www.navaldrones.com/Russian-Naval-Drones.html.

11 See “Fire Scout”, Northrop Grumman, available at: https://tinyurl.com/y9zkjojz.

12 Sam LaGrone, “Pentagon to Navy: Convert Uclass Program into Unmanned Aerial Tanker, Accelerate F-35 Development, Buy More Super Hornets”, USNI News, 1 February 2016, available at: https://tinyurl.com/hwk9gtx; Kris Osborn, “Navy Awards MQ-25 Stingray Tanker Deal”, Defense Systems, 24 October 2016, available at: https://defensesystems.com/articles/2016/10/24/stingray.aspx.

13 UUV Master Plan, above note 8, p. 9.

14 “Sea Stalker UUV”, Naval Drones, available at: www.navaldrones.com/Sea-Stalker-UUV.html; “Sea Maverick UUV”, Naval Drones, available at: www.navaldrones.com/Sea-Maverick.html. The Sea Stalker and Sea Maverick programmes were respectively initiated in 2008 and 2009, with both undergoing final demonstration in 2010: see US Defense Science Board (DSB), The Role of Autonomy in DoD Systems, Office of the Under Secretary of Defense for Acquisition, Technology and Logistics, Washington, DC, 2012, p. 89Google Scholar.

15 DSB, above note 14, p. 88.

16 For a detailed outline of current and future operations, see UMS Roadmap, above note 2, pp. 109, 88.

17 See DSB, above note 14, pp. 85–86.

18 USV Master Plan, above note 2, pp. 11, 38.

19 UUV Master Plan, above note 8, pp. 7–15.

20 The US DSB has reiterated the centrality of combat missions to the research and deployment of UMVs: see DSB, above note 14, p. 17.

21 See ibid., p. 86.

22 See Richard Scott, “ONR to Swim Ahead on ASW Package for Large UUV”, IHS Jane's Navy International, 20 November 2014.

23 B. Clark, above note 2, p. 13.

24 The Sea Hunter is now considered a medium-displacement USV, a characterization that reveals the US Navy's interest in extending its surface combatant capabilities. See Richard Scott, “Surface Navy 2017: Sea Hunter Trials to Inform Unmanned Debate for Next Surface Combatant”, IHS Jane's Navy International, 11 January 2017; Geoff Fein, “Sea Hunter Begins Operational Testing, Readies for 2017 Colregs Certification”, Jane's International Defence Review, 5 December 2016; Richard Scott, “Talons Raised Aloft in USN Testing”, IHS Jane's Navy International, 17 November 2016.

25 “Protector USV”, Naval Drones, available at: www.navaldrones.com/protector.html; “Protector Unmanned Surface Vehicle (USV), Israel”, Naval Technology, available at: www.naval-technology.com/projects/protector-unmanned-surface-vehicle/.

26 Huw Williams, “Rafael Launches Spike Missiles from Protector USV”, Jane's International Defence Review, 8 March 2017.

27 “Protector Unmanned Surface Vehicle (USV), Israel”, above note 25.

28 Singer, Peter Warren, Wired for War: The Robotics Revolution and Conflict in the 21st Century, Penguin, New York, 2009, pp. 126–128Google Scholar; R. Sparrow, “Robots and Respect”, above note 3.

29 There are intimations in the literature that recent technological breakthroughs have significantly increased the capacity of submersibles to communicate with other vessels and shore installations without revealing their location (see, for instance, B. Clark, above note 2, p. 14). In particular, short-burst transmissions from deployed devices, timed to transmit after the submarine has left the area, are within the current inventory. Nevertheless, we think it is unlikely that any such technology will allow continuous tele-operation of a UUV under combat conditions without jeopardizing the safety of the UUV.

30 Adams, Thomas K, “Future Warfare and the Decline of Human Decision Making”, Parameters, Vol. 31, No. 4, 2001Google Scholar; R. Sparrow, “Robots and Respect”, above note 3.

31 Doswald-Beck, Louise (ed.), San Remo Manual on International Law Applicable to Armed Conflicts at Sea, Cambridge University Press, Cambridge and New York, 1995CrossRefGoogle Scholar. It is possible that eventually, once the technology becomes available, most commercial shipping will also be unmanned, in which case it is only the presence of troop transports or commercial passenger vessels that would establish a risk of persons being left shipwrecked, wounded or sick.

32 Just war theory is a tradition of moral, political and philosophical argument regarding the ethics of war, which has strongly influenced the development of IHL. Classically, just war theory is concerned both with the question of when States are justified in resorting to war (jus ad bellum) and with the question of how wars may permissibly be fought (jus in bello). The most influential contemporary source on just war theory is Michael Walzer's Just and Unjust Wars: A Moral Argument with Historical Illustrations, published by Basic Books and now in its fifth edition (2015). For an account of the history of this tradition, see Johnson, James T., Ideology, Reason and Limitation of War: Religious and Secular Concepts, 1200–1740, Princeton University Press, Princeton, NJ, 1981Google Scholar.

33 Declaration Renouncing the Use, in Time of War, of Explosive Projectiles under 400 Grammes Weight, 1868; Corn, Geoffrey S et al. , The Law of Armed Conflict: An Operational Approach, Wolters Kluwer Law and Business, New York, 2012Google Scholar.

34 Protocol Additional (I) to the Geneva Conventions of 12 August 1949, and relating to the Protection of Victims of International Armed Conflicts, 1125 UNTS 3, 8 June 1977 (entered into force 7 December 1978) (AP I), Art. 48.

35 Geneva Convention (I) for the Amelioration of the Condition of the Wounded and Sick in Armed Forces in the Field of 12 August 1949, 75 UNTS 31 (entered into force 21 October 1950) (GC I); Geneva Convention (IV) relative to the Protection of Civilian Persons in Time of War of 12 August 1949, 75 UNTS 287 (entered into force 21 October 1950).

36 I. Papanicolopulu, above note 4.

37 ICRC Commentary on GC II, above note 4, para. 1619.

38 It is true that there are sometimes significant inconveniences associated with the transport and care of prisoners of war. Nevertheless, these are clearly outweighed by the benefits of the practice to those who would otherwise be left to drown.

39 Boss, Pauline, “Families of the Missing: Psychosocial Effects and Therapeutic Approaches”, International Review of the Red Cross, Vol. 99, No. 905, 2017CrossRefGoogle Scholar.

40 Rousseau, Jean-Jacques, The Social Contract, trans. Cranston, Maurice, Penguin, London, 2003, p. 56Google Scholar. The phenomena of civil wars and counter-insurgencies problematizes Rousseau's claim, but even in cases of non-international armed conflicts the actors must be collective and “State-like” in order to justify the description of a conflict as “war” rather than as another less organized form of political violence, such as civil unrest or banditry.

41 Sparrow, Robert, “‘Hands up Who Wants to Die?’: Primoratz on Responsibility and Civilian Immunity in Wartime”, Ethical Theory and Moral Practice, Vol. 8, No. 3, 2005CrossRefGoogle Scholar.

42 The legal status of (former) combatants who have become non-combatants by virtue of being hors de combat is more complicated given that, for instance, they may be taken as prisoners of war where civilian non-combatants may not. Legally speaking, therefore, although it is prohibited to attack them, such persons remain “enemy” nationals. See, inter alia, Article 3 common to the four Geneva Conventions; GC I, Art. 12; GC II, Art. 12; AP I, Arts 10, 41(2); Protocol Additional (II) to the Geneva Conventions of 12 August 1949, and relating to the Protection of Victims of Non-International Armed Conflicts, 1125 UNTS 609, 8 June 1977 (entered into force 7 December 1978), Arts 4, 7.

43 GC II, Art. 18.

44 ICRC Commentary on GC II, above note 4, para. 1636.

45 The Laconia Order was issued by Grand Admiral Karl Dönitz of the German Navy in 1942, subsequent to an Allied attack on a German U-boat involved in rescuing the survivors of a German attack on the Royal Mail steamer Laconia. The order instructed the commanders of the U-boat fleet to cease conducting rescue operations. Dönitz's having issued the Laconia Order was one of the matters at stake in Dönitz's trial for war crimes at Nuremberg. See Nuremberg Trial Proceedings, Vol. 13, 125th Day, 9 May 1946, available at: http://avalon.law.yale.edu/imt/05-09-46.asp; Bennett, G. Harry, “The 1942 Laconia Order, the Murder of Shipwrecked Survivors and the Allied Pursuit of Justice 1945–46,” Law, Crime and History, Vol. 1, No. 1, 2011Google Scholar; Maurer, Maurer and Paszek, Lawrence J., “Origin of the Laconia Order”, Royal United Services Institution Journal, Vol. 109, No. 636, 1964CrossRefGoogle Scholar; Fenrick, William J, “The Exclusion Zone Device in the Law of Naval Warfare”, Canadian Yearbook of International Law, Vol. 24, 1986, p. 103Google Scholar. See also note 53 below.

46 Alternative formulations would be to insist either that the belligerent parties still have an obligation but would have no way of fulfilling it or that they would have an excuse for not fulfilling their obligations. Either way, though, they would in practice have no obligation to perform any particular action that might benefit those in need of rescue in such a case.

47 Kahn, Paul W., “The Paradox of Riskless Warfare”, Philosophy & Public Policy Quarterly, Vol. 22, No. 3, 2002Google Scholar, has argued that, alongside other systems involved in high-tech warfare, UMS systematically transfer risks from combatants to non-combatants in ways that threaten traditional justifications for the use of force. To our knowledge, however, we are the first to notice this particular way in which such a shift might occur.

48 Arkin, Ronald C., “Lethal Autonomous Systems and the Plight of the Non-Combatant”, AISB Quarterly, No. 137, 2013Google Scholar.

49 Bryan Bender, “Attacking Iraq, from a Nev. Computer”, Boston Globe, 3 April 2005; P. W. Singer, above note 28, pp. 330–333; David L. Ulin, “When Robots Do the Killing”, Los Angeles Times, 30 January 2005. For critical evaluation of this claim, see Sparrow, Robert, “Robotic Weapons and the Future of War”, in Wolfendale, Jessica and Tripodi, Paolo (eds), New Wars and New Soldiers: Military Ethics in the Contemporary World, Ashgate, Surrey and Burlington, VA, 2011Google Scholar. For a discussion of the implications of the geographical distance between the operators and the actions of remotely controlled systems for the extent to which the operators can cultivate the martial virtues, see Sparrow, Robert, “War without Virtue?”, in Strawser, Bradley Jay (ed.), Killing by Remote Control, Oxford University Press, Oxford, 2013Google Scholar.

50 Engineers and weapon designers have almost always been a long way from the front lines, of course. What is distinctive about UMS is that the design and programming of these weapons encompasses more and more aspects of their use. Indeed, the logical end point of the development of UMS is the creation of systems that only need to be launched before proceeding to conduct combat operations entirely autonomously. In this case, the designers of these systems would become responsible for all of the decisions previously made by the combatants who formally carried out this combat role. For a discussion of the ethical issues that arise in the course of the design of UMS more generally, see Sparrow, Robert, “Building a Better Warbot: Ethical Issues in the Design of Unmanned Systems for Military Applications”, Science and Engineering Ethics, Vol. 15, No. 2, 2009CrossRefGoogle ScholarPubMed.

51 Distinguishing between weapons, weapon systems and agents remains a significant challenge in terms of both law and ethics when assessing compliance with obligations. See Sparrow, Robert, “Twenty Seconds to Comply: Autonomous Weapon Systems and the Recognition of Surrender”, International Law Studies, Vol. 91, 2015Google Scholar. Regrettably, compliance with the ongoing obligation upon States to conduct weapons reviews on all new means and methods of warfare (AP I, Art. 36) is far from universal. See, generally, ICRC, “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”, International Review of the Red Cross, Vol. 88, No. 864, 2006Google Scholar.

52 Strike aircraft do typically possess the capacity to notify nearby surface vessels of the existence of survivors in the water after an engagement, and are morally obligated to do so when they do possess this capacity.

53 Historically, the conventions and protocols regulating maritime conflict, particularly relating to rescue, centre on the conduct of surface ships, posing distinct problems for submarines. First, as the military utility and strategic advantage of submersibles depends on their status as “stealth vehicles”, detection uniquely threatens their military capability and makes them highly vulnerable to enemy attack. Consequently, regulations that oblige participation in surface operations distinctively jeopardize submersible vehicles. Second, submarines have few crew and limited space and resources, all of which restricts their ability to perform rescue. Accordingly, it has been unclear historically what is required of these vehicles in providing for the safety of non-combatants. See Legro, Jeffrey, Cooperation under Fire: Anglo-German Restraint During World War II, Cornell University Press, Ithaca, NY, 1995, pp. 35–40Google Scholar. During the Nuremberg Trials, Germany argued that as the “security of the submarine is, as the first rule of the sea, paramount to rescue”, and because of the “unusual additional danger” that rescue presented to submarines, there was cause for an exception to the rescue duty of submersible vehicles. Further, they argued that for the reasons we have mentioned here (space, crew, stealth), the submarine was “subject to special considerations” as rescue “prejudices the military mission”. Subsequently, while it was accepted that on 17 September 1942 Grand Admiral Dönitz of the German Navy had forbidden all rescue efforts by submarines, the sentencing of Dönitz for war crimes “was not assessed on the ground of his breaches of the international law of submarine warfare”. See Nuremberg Trial Proceedings, Vol. 18, 179th Day Tuesday, 16 July 1946, available at: http://avalon.law.yale.edu/imt/07-16-46.asp; Judgement: Doenitz, available at: http://avalon.law.yale.edu/imt/juddoeni.asp.

54 It might be argued that because mines tended to be deployed near coasts and at choke points, and thus were never far from observation, manned systems have always been available to conduct rescue operations for those whose vessels were sunk by mines – with the consequence that the question of the ethics of the use of naval mines in relation to the duty of rescue did not arise. Similarly, torpedoes have a limited life when launched. The fact that free-floating mines are required by IHL to render themselves (or be rendered) inert after one hour and that torpedoes are expected to become inert at the end of their run might further be adduced in support of this claim. See Hague Convention (VIII) Relative to the Laying of Automatic Submarine Contact Mines, 18 October 1907, Art. 1. However, we believe the legal expectations when it comes to the use of free-floating mines and torpedoes are better explained as arising from a concern for distinction, and that the empirical claim about the availability of manned systems is disputable. See, generally, Levie, Howard S., “Submarine Warfare: With Emphasis on the 1936 London Protocol”, International Law Studies, Vol. 65, 1993Google Scholar; Stephens, Dale and Fitzpatrick, Mark, “Legal Aspects of Contemporary Naval Mine Warfare”, Loyola of Los Angeles International and Comparative Law Journal, Vol. 21, No. 4, 1999Google Scholar; Letts, David, “Naval Mines: Legal Considerations in Armed Conflict and Peacetime”, International Review of the Red Cross, Vol. 98, No. 902, 2016CrossRefGoogle Scholar; US Navy, US Navy Commander's Handbook on the Law of Naval Operations, NWP 1-14M, 2017, section 9.2.

55 UMS that destroy themselves in the course of an attack cannot make any contribution to search and rescue operations. Where systems are operated remotely by personnel on board manned vessels nearby, it is less important that the unmanned system is able to contribute to search and rescue operations because presumably the manned vessel would have the capacity to do so. However, where UMS are capable of extended operations and travelling long distances, it increases the chance that they might be the only vessel in the vicinity of people in need of rescue, which in turn increases the force of the case that they should be provided with the capacity to contribute to search and rescue operations. Similarly, larger vessels have more of a capacity to carry life rafts and/or take seamen aboard, so it would be especially egregious if they were not provided with such functionality. For further discussion of the extent to which it is reasonable to expect that different sizes and sorts of UMS should be provided with the capacity to contribute to search and rescue operations, see the section “Designing UMS for Rescue”, below.

56 Matthias, Andreas, “The Responsibility Gap: Ascribing Responsibility for the Actions of Learning Automata”, Ethics and Information Technology, Vol. 6, No. 3, 2004CrossRefGoogle Scholar; Roff, Heather M, “Killing in War: Responsibility, Liability, and Lethal Autonomous Robots”, in Allhoff, Fritz, Evans, Nicholas G and Henschke, Adam (eds), Routledge Handbook of Ethics and War: Just War Theory in the 21st Century, Routledge, Milton Park, 2013Google Scholar; Sparrow, Robert, “Killer Robots”, Journal of Applied Philosophy, Vol. 24, No. 1, 2007CrossRefGoogle Scholar.

57 Hellström, Thomas, “On the Moral Responsibility of Military Robots”, Ethics and Information Technology, Vol. 15, No. 2, 2013CrossRefGoogle Scholar.

58 R. Sparrow and G. Lucas, above note 3. See also the discussion in I. Papanicolopulu, above note 4, pp. 495–497, 504, on the references in the 1989 International Convention on Salvage and the 1910 Salvage Convention to the duties of the masters of vessels, which, Papanicolopulu argues, persist in wartime.

59 ICRC Commentary on GC II, above note 4, para. 1630.

60 B. Gogarty and M. Hagger, above note 3, pp. 114–116; A. H. Henderson, above note 3, p. 66; R. McLaughlin, above note 3, p. 112; A. Norris, above note 3; von Heinegg, Wolff Heintschel, “Submarine Operations and International Law”, in Engdahl, Ola and Wrange, Pål (eds), Law at War: The Law as It Was and the Law as It Should Be, Koninklijke Brill, Leiden, 2008, p. 146Google Scholar.

61 Law of the Sea Convention, 1982, Arts 91 (“Nationality of Ships”), 92 (“Status of Ships”).

62 R. McLaughlin, above note 3.

63 Insofar as “ought implies can”, the obligations on UMS to conduct search and rescue obligations – or, more precisely, on the designers of UMS to provide them with the capacity to contribute to search and rescue operations – is a function of the capacity of systems to facilitate search and rescue operations. Our reasons for thinking that larger systems have a greater capacity to contribute to search and rescue operations are provided immediately below.

64 Unless they can communicate with a human controller, UMS cannot be switched to “active” mode at the beginning of a conflict and risk continuing to engage in combat after a conflict has ended. The capacity to detect objects in the water is essential to most of the roles for which naval UMS are intended, and especially to combat operations in and under the water. Detecting individual persons in the water is a formidable challenge, but detecting objects such as life rafts or the presence of life jackets that emit light or signals may well be plausible for a sophisticated unmanned system located nearby. UAVs could presumably transmit video footage to a human controller who could assess the nature of the post-conflict environment and the likelihood that it contains persons in need of assistance.

65 It is possible that UUVs should be held to have a limited obligation to conduct search and rescue operations by analogy with the case of manned submersibles. However, as suggested in the section “The Case for a Duty of Rescue for UMS” above, it is also plausible to think that they are under more of an obligation to do so than manned submersibles insofar as they are more expendable than manned submersibles. In any case, if there are any circumstances in which they have such an obligation, the issues discussed here will arise.