Revisiting Omission Liability at the ICC in the Wake of Evolving Weapons Technology

[Shashiskala Gurpur is the Director of Symbiosis Law School and Dean, Faculty of Law, Symbiosis International (Deemed) University. Aakarsh Banyal is a student of law at Symbiosis Law School, Pune, India.]

Introduction

Facebook CEO’s apology over the company’s failure to remove posts which incited widespread violence in Myanmar stands testament to the possibility of mass violence being perpetuated through the use of technology. Such instances demonstrate that reprehensible consequences may also flow from an individual’s failure to act/omission, especially in the context of evolving technology. The Fourth Industrial Revolution pushes the technological boundaries even further where Artificial Intelligence in the area of weapons systems become a practicable reality.

Kai Ambos, in his Treatise on International Criminal Law, specified that the Rome Statute [hereinafter, ‘ICCSt.’]  does not make a reference to general omission liability because a consensus could not be reached by delegates during the deliberations over content of the ICCSt. Therefore, omission liability in the ICCSt. stays restricted to Art. 28 and some genuine crimes of omission (eg. starvation under Art. 8(2)(b)(xxv), ICCSt.).

As a result, only in specific cases, which fall under the purview of Art. 28, ICCSt. (pertaining to command/superior responsibility), an individual can be held responsible for her failure to act. Aside from this, individual criminal responsibility has generally been reserved for positive acts/commission under Art. 25(3), ICCSt. Never before has the need for a general omission liability based on an individual’s failure to act at the ICC been more strongly felt than now with the rapid advance of technology.

The authors attempt to lay out a workable interpretative schema and the possible pitfalls that ought to be avoided in order to effectively tackle the issue of AI-driven atrocities at the ICC. Part II offers a context to the article and emphasises the need for a general omission liability. Part III establishes the scope for the inclusion by the ICC of a general omission liability through reliance on the drafting history of the ICCSt. Parts IV and V explore the case for such inclusion by weighing some of the sources of law the ICC may rely on, ie., customary international law and general principles, to derive a general omission liability for individuals within international criminal law.

1. Necessity of General Omission Liability vis-à-vis Autonomous Weapons Systems

The heightened significance of incorporating a general omission liability within the ICCSt. is inextricably linked to the emergence of advanced weapons technology such as Lethal Autonomous Weapons Systems (“LAWS”). These weapons are capable of making decisions without any human involvement. Military powers such as China, Israel, Russia, South Korea, the United Kingdom, and the United States are increasingly investing in the development of various autonomous weapons systems, as per the recent Human Rights Watch report. Furthermore, Russia, US and UK have expressed significant opprobrium towards negotiating a treaty which seeks to ban such autonomous weapons systems.  They cite humanitarian and strategic advantages that accompany increased automation, inter alia.

Most conventional weapons systems have a human at the helm of operation and decision-making (humans-in-the-loop). However, the current discourse sways towards reducing human control to the point of mere “abortive” measures (humans-on-the-loop). In such cases, an individual’s failure to abort an attack by these weaponscan lead to outcomes that merit individual criminal responsibility. Hence, imputing criminal responsibility to individuals who do not abort such attacks takes pre-eminence.

The difficulty in attaching criminal responsibility in this fashion is that omission liability exists within the confines of Art. 28, ICCSt. An accountability gap thus gets created since, ‘LAWS-human’ relationship is not synonymous with a ‘human superior-human subordinate’ relationship, the latter being key requirement to trigger responsibility under Art. 28, ICCSt. In addition, the ICC can only try natural persons and hence, the possible entry of private corporations also widens this accountability gap by virtue of nonexistence criminal responsibility at the level of corporation and Art. 28, ICCSt.’s ineffectuality at the level of individuals.

Thus, general omission liability stands to provide the wherewithal required to hold individuals responsible for committing crimes through the deployment of LAWS. To instantize, if an autonomous robot were about to shoot indiscriminately and the operator would not call off such an attack, the ICC could ascribe liability to the operator via general omission liability, provided other requirements are duly met (jurisdiction, elements of crimes, etc.). A limitation of the authors’ approach is a paradigm where LAWS are devoid of any human intervention (humans-out-of-the-loop) because in this case, pinning liability requires extension of ICC’s personal jurisdiction to include LAWS themselves.

Revisiting the Preparatory Works on Omission Liability

The recognition of a general omission liability was deliberated upon by the drafters of the ICCSt. The 1995 Draft Statute for an ICC recognised commission by omission outside of command/superior responsibility. In addition, the 1998 Report of the Preparatory Committee referred to an “intentional omission” liability demanding a pre-existing legal obligation. Despite these indications, a general liability for omissions was not included in the final draft of the ICCSt. Saland (then-Chairman, Working Group on General Principles of Criminal Law) justified this exclusion as a bid to allow scope for future case law of the ICC to develop this concept.

To that end, it is clear that the non-inclusion of omission liability is not indicative of the intention to exclude such a mode of liability. This observation is corroborated by the ICC’s dictum in Prosecutor v. Katanga (see para. 287 ) and Prosecutor v. Lubanga (see para. 351), whereinthe possibility of culpability through actions as well as omissions was obliquely read into the ‘conduct’ of the accuseds.

One might contend that Art. 28, ICCSt. constitutes lex specialis (law governing specific subject matter) in relation to omission liability, thereby rejecting its recognition in a more general form. To this end, Karadžić’s motion (see para. 2, 19-20) which put forth this argument was dismissed by the International Criminal Tribunal for the former Yugoslavia

[hereinafter, ‘ICTY’]

Appeals Chamber on the ground that omission is merely a contour of the actus reus of a crime. Thus, the preparatory works favour the growth of omission liability within the ICC’s jurisprudence, as Saland remarked.

A case to incorporate general omission liability must be made via reliance on the recognised sources of law in the ICCSt. Article 21 of the ICCSt. recognises customary international law and general principles as sources which the ICC may resort to during trial. The following sections will weigh the benefits and drawbacks of each of the aforementioned sources vis-à-vis omission liability.

Customary International Law vis-à-vis Omission Liability

As noted by Schabas, Art. 21(1)(b), ICCSt. contains an expression of customary international law, which has been traditionally described as being comprised of State practice and opinio juris. As observed in the North Sea Continental Shelf case, “[n]ot only must the acts concerned amount to a settled practice, but they must also be such, or be carried out in such a way, as to be evidence of a belief that this practice is rendered obligatory by the existence of a rule of law requiring it.”

Even so, the threshold of evidence required to establish State Practice and opinio juris is not set in stone. Based on Kirgis’ ‘sliding scale’ approach, a case-by-case analysis has to be undertaken to establish a customary rule and the weight accorded to each of the two elements can differ, provided that evidence of these elements is incontrovertible.

The pronouncements of the British Military Court and the U.S. Military Tribunal are proof of State practice inasmuch as they recognised liability for omissions beyond command/superior responsibility (Art. 28, ICCSt.), thereby favouring the interpretation in the direction of general omission liability. However, the pronouncements referenced hereinabove are dated and quite possibly, inadequate to firmly institute State Practice.

With respect to opinio juris, Art. 86, Additional Protocol I [hereinafter, ‘AP I’] states that the High Contracting Parties “shall repress grave breaches, […] which result from a failure to act when under a duty to do so”. The ICTY has previously relied on Art. 86, AP I to ascertain opinio juris. However, this provision cannot be regarded as a true reflection of opinio juris for the purpose of general omission liability as it is limited to “grave breaches” and not all international crimes.

The palpable obstacle in arguing that omission liability is a rule of customary international law lies in the sparse evidence of State practice, combined with the limited utility of Art. 86(1), AP I as opinio juris. Hence, controvertible State Practice and opinio juris would increasingly add to the fallibility of such an argument.

General Principles vis-à-vis Omission Liability

Art. 21(1)(c), ICCSt. paves the way for the ICC to interpret law based on “general principles of law derived by the Court from national laws of legal systems of the world…”. A scrutiny of principal legal systems reveals that omission liability has a firm foundation as a general principle, as evidenced by Duttwiler’s findings on the point.

Duttwiler reviewed a number of States in common law, civil law, Sharia law, and “Far-East” jurisdictions. Broadly, and across numerous jurisdictions, liability for omissions existed when there was a legal duty to act. The legal orders of countries such as Germany, Spain, Japan, China, UK, USA, etc. were relied on. In an effort to supplement Duttwiler’s findings, the authors have attempted to study the legal systems of Finland, Latvia, Lithuania, Greece, Fiji, and Israel where omission has been criminalised, provided there exists a legal duty to act. The overall survey reveals a tenable case for general omission liability to be made via Art. 21(1)(c), ICCSt.

In reality however, any such case has to be developed within the rigours of an ongoing trial, and not as an academic exercise. Previously, the ICTY inProsecutor v. Erdemović (see para. 56-71) embarked on an investigation into relevant legal systems to derive a general principle concerning duress as a defence wherein a sample of about 20 States was used to determine a general principle. The apparent shortcoming of this approach is that national laws of only 20 States were reviewed to ascertain a general principle, excluding the possibility of a thorough investigation into multiple legal orders.

As pointed by Neha Jain, the inherent limitations with respect to language, availability of resources, and time may cause the Court to engage in cherry-picking national laws, which “as a practical matter, [are] accessible to [them]” (Prosecutor v. Erdemović, para 57). Thus, the procedure employed in Erdemović hints at “convenience sampling”, a technique in which only those samples are chosen which are convenient to assess and thus, may breed bias.

In addition, reliance on legislations to infer a legal system’s position on a given proposition also impedes a comprehensive effort to arrive at a principle. For instance, Erdemović’s case majorly looked at legislations, which means judicial and scholarly opinions which also carry the potential to alter the final outcome may go unnoticed.

To the same effect, Werle and Jeßberger posit that “the determination of general principles of law […], particularly if it is to be carried out with the appropriate methodological sophistication, is not practical under conditions of an ongoing legal case.” Thus, given that the development of omission liability has been left to the case law of the ICC (see Part III), an exhaustive assessment of national legal systems under Art. 21(1)(c), ICCSt. will need to be undertaken cautiously.

Concluding remarks

The ICC is often confronted with a diverse range of facts and circumstances. It follows that in order to effectively address the legal issues that arise in light of the impending tech-based atrocities, there must be scope for such omissions to be penalised. The defensive stance taken by the military powers over a complete ban coupled with the highly destructive power of LAWS merit individual criminal responsibility for individuals deploying these LAWS at the ICC. Even though all stages of LAWS’ development may not be covered by this approach, omission liability ensures accountability at the level of operationalisation of LAWS in the battlefield.

Thus, despite the silence of the ICCSt. in this regard, the ICC could fill this gap by drawing from the various sources of applicable law. On balance, it appears that reliance on customary international law might be misplaced as the elements of State practice and opinio juris cannot be strongly established. A more viable route to institute general omission liability is through general principles, with care being taken to avoid the sampling biases.