Canada

nwc_leftCan the laws of war constrain robot warriors? Is international humanitarian law adaptable to the use of weapons that possess artificial intelligence? To what extent can such weapon systems determine who is, and who is not, a combatant? To what extent must humans control the decision to kill the enemy?

These questions and others fostered a fascinating discussion at “Legal Implications of Autonomous Weapon Systems,” a workshop at the Naval War College in Newport, Rhode Island, this past Thursday and Friday. We four dozen or so attendees were drawn from the armed forces of the United States, Australia, Britain, Canada, and Israel, from the International Committee of the Red Cross, and from a global array of academic institutions.

As one who reserves just a couple days for the topic in my Laws of War course, I came to the workshop with more questions than answers about the actual and potential uses in armed conflict of robots, the shorthand term I’ll use here for “autonomous weapons systems.” The military, characteristically, prefers an acronym: AWS.

The actual use of such weapons already is significant. Smart missiles called JDAMs deliver munitions to a target, while a WALL·E-looking machine called SWORDS has, as the U.S. Department of Defense wrote in 2004, “march[ed] into battle” alongside troops.

In fact, such machines tend not to be used in a fully independent manner (though with a little reprogramming, some could be). They are, we were told, semi-autonomous – humans are kept “in” or “on” the loop leading to choice of target and other decisions.

This mention of human supervision, like the WALL·E-on-the-march metaphor above, pointed to a pivotal workshop topic:

nwc_right►  Is it appropriate, as a matter of law or of ethics, to indulge in the human tendency to anthropomorphize these machines?

Apparently, some lab robots can recognize – or at least can mimic the act of recognizing – themselves in a mirror. Does this mean they are, or soon will be, sufficiently human-like to conduct operations wholly without oversight by actual humans? Might human-like robots evolve an ability to refuse programmed orders – orders that limited action to the boundaries of international humanitarian law? The answers to these questions, like many at the workshop, seemed to be “perhaps yes, perhaps no.”

At one end of the spectrum, this uncertainty has spurred a call for an outright ban. Emblematic is the headline of a notice about the November 2012 release of the Human Rights Watch report, Losing Humanity:

‘Ban ‘Killer Robots’ Before It’s Too Late: Fully Autonomous Weapons Would Increase Danger to Civilians’

At the other end of the spectrum, some would prefer to let the technology develop before the onset of any new legal regulation.

Many seem to fall in between. Acknowledged were some challenges; for instance:

► Does compliance with the precautions requirement of Article 57 of the Additional Protocol I (1977) to the four Geneva Conventions (1949) preclude the use of a fully autonomous weapon?

► Would the robotic commission of a war crime be susceptible to sanctions by global justice mechanisms like the International Criminal Court, and if not, what effective sanctions and deterrents would there be?

Persons falling in the vast middle of the regulatory spectrum harbored concerns about such questions, yet seemed to lean toward the view that if due care is taken, international humanitarian law can – and should – be applied. Documents discussed in this vein included the:

► U.S. Department of Defense Directive 3000.09, ¶ 4(a) (November 12, 2012), which states as “DoD policy” the following:

‘Autonomous and semi-autonomous weapons systems shall be designed to allow commanders and operators to exercise appropriate levels of human judgement over the use of force.’

heyns► April 9, 2013 report to the U.N. Human Rights Council by University of Pretoria Law Professor Christof Heyns, who’s served since 2010 as the Special Rapporteur on extrajudicial, summary or arbitrary executions. At ¶ 108 of his report, Heyns termed the 2012 Defense Directive as “imposing a form of moratorium” with respect to what he termed “lethal autonomous robotics,” or LARs. Heyns’ 2013 U.N. report (¶ 35) favored a broader scope for delay:

‘The present report … calls on States to impose national moratoria on certain activities related to LARs.’

A reprise of such issues likely will occur at the Meeting of Experts on Lethal Autonomous Weapons Systems set for May 13 to 16 in Geneva under the auspices of the 1980 Convention on Certain Conventional Weapons. Named in full the 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 as amended on 21 December 2001, this treaty has 117 states parties, including the United States.

The Naval War College International Law Department workshop’s vital and timely discussion exposed many avenues for study – study sooner rather than later, so that the legal regulatory framework may be determined before fully autonomous robots are fully deployed.

Notable in the just-released White House recap of its efforts to prevent mass atrocities is the foregrounding of 2 actions this year:

► Enactment in January of “bipartisan legislation to enhance our ability to offer financial rewards” – up to $5 million – “for information that helps to bring to justice” selected international indictees, among them “Joseph Kony and other senior leaders of the Lord’s Resistance Army, as well as Sylvestre Mudacumura from the Democratic Forces for the Liberation of Rwanda….”

► Developments in March, when “[t]he United States facilitated the voluntary surrender of Bosco Ntaganda” to stand trial “for war crimes and crimes against humanity….”

g8What’s notable is that both actions – like others noted in this commentary by Professor David Kaye – come to the aid of the indicting organization, the Hague-based International Criminal Court.  The same is true of an action not mentioned in the recap; that is, the Declaration on Preventing Sexual Violence in Conflict adopted at a mid-April London meeting. Paragraph 5 of the Declaration concludes:

‘Ministers emphasised the need for further funding support for victims and called on the international community, including the G8, to increase their efforts to mobilise such funding, including to programmes such as the ICC Trust Fund for Victims and its implementing partners.’

“Ministers” refers to the Foreign Ministers of the European Union and all members of the G-8. Thus joining the Declaration were 2 countries not party to the ICC’s Rome Statute: Russia and the United States. (credit for AFP photo made at the G-8 meeting of Foreign Ministers – from left, minsCatherine Ashton, European Union; John Baird, Canada; Laurent Fabius, France; John Kerry, United States; William Hague, Britain; Sergey Lavrov, Russia; Guido Westerwelle, Germany; Fumio Kishida, Japan; and Mario Monti, Italy)

These actions prompt examination of the potential extent of U.S. support for the ICC – in particular, given the G-8 Declaration, U.S. support for the Trust Fund for Victims. Would U.S. financial contributions to the Trust Fund for Victims contravene the American Service Members Protection Act? A preliminary look at the question indicates that they would not.

Section 2004 of the Act prohibits the giving of various forms of “cooperation,” “support,” and “appropriated funds” to the “International Criminal Court.” Section 2013(6) has the following definition:

‘INTERNATIONAL CRIMINAL COURT – The term “International Criminal Court” means the court established by the Rome Statute.’

The definition gives rise to a question: Does the Trust Fund fall within that statutory term “the court”?

Despite some writings on the workings of the Trust Fund (e.g., here), there appears to be little in-depth scholarship on the organizational relationship between it and the Court. An expert on international organizations ought to take this on.

Still, ICC documents seem to set the Fund apart from the Court; that is, the Trust Fund is established to benefit victims of crimes in the Court’s jurisdiction, but is not expressly itself within the Court’s jurisdiction. It is governed by the Assembly of States Parties, as is the Court, but there is an argument that it is not part of the Court. Indeed, a 2007 amendment permits earmarking of voluntary contributions to the Trust Fund in a way that sets such donations outside the frame of the Court. Both are within the “Rome Statute system,” as ICC Prosecutor Fatou Bensouda and others term it, but they arguably are separate entities within that system.

Note too that the founding resolution does not seem to limit membership on the Trust Fund board of directors to nationals of states parties. Assuming later-promulgated regulations do not change this, that would make this board different from other elected positions, like ICC judge and ICC prosecutor, and again suggests a different status.

Now consider Section 2004(f) of the American Service Members Protection Act:

‘PROHIBITION ON USE OF APPROPRIATED FUNDS TO ASSIST THE INTERNATIONAL CRIMINAL COURT – Notwithstanding any other provision of law, no funds appropriated under any provision of law may be used for the purpose of assisting the investigation, arrest, detention, extradition, or prosecution of any United States citizen or permanent resident alien by the International Criminal Court.’

The phrasing begs the question whether there is a ban on U.S. funding in any instance in which the suspect or accused is not a “United States citizen or permanent resident alien.” It seems even more clear that funding reparations – aiding victims, without regard to perpetrators – is something wholly outside the scope of “investigation, arrest, detention, extradition, or prosecution.” By this reasoning, U.S. financial contributions to the Trust Fund for Victims, which has no role at all in “investigation, arrest, detention, extradition, or prosecution,” are not prohibited by the American Service Members Protection Act. (See further limits on the Act’s scope – including reference to unofficial reports of a confidential Office of Legal Counsel memo on the issue – at pages 6-11 and 17 of a 2010 American Society of International Law compilation of white papers, Beyond Kampala: Next Steps for U.S. Principled Engagement with the International Criminal Court.)

The above interpretation of the Act’s funding rules cannot be extended to subsequent legislation, by which Congress imposed a blanket ban on using appropriated funds “for use by, or for support of, the International Criminal Court,” unless and until the now-unlikely event that the United States ratifies the Rome treaty following 2/3 approval by the Senate. The latter amendment, however, defines “International Criminal Court” in much the same way as the above-quoted Section 2013(6) of the American Service Members Protection Act; thus the question lingers whether the Trust Fund falls within the scope even of the latter amendment’s ban.

dosA final point respecting the American Service Members Protection Act: U.S. contributions to the Trust Fund would not contradict the intent of Congress, as it may be inferred from the Section 2002 Findings with which the Act begins. Donating to the Trust Fund for Victims in no way would enable the Court to pursue U.S. nationals or other “covered” individuals. It would have no relation to the ICC offense that appeared to give Congress most concern, the not-yet-fully-punishable crime of aggression. And with regard to Finding #4, which quotes the 1998 statement in which then-Ambassador David Scheffer opposed the Rome Statute on the ground that “‘[w]e are left with consequences that do not serve the cause of international justice,'” one discerns a congressional willingness to support institutions (such as ad hoc tribunals, which are exempted from the Act) that in fact serve that cause. U.S. contributions to reparations, via the Trust Fund for Victims, would meet that criterion.

Consideration of such contributions would further the United States’ current policy of positive engagement with the Rome system of international criminal justice. Reconsideration of all federal statutory barriers, a move supported by a range of U.S. experts (among them, Professor Kaye and former State Department Legal Adviser John B. Bellinger III), seems yet another logical next step.