Above: People look at drones the Ukrainian government claims it recovered in eastern Ukraine that prove direct Russian involvement in the fighting between Ukrainian troops and pro-Russian separatists .(Photo by Sean Gallup/Getty Images)
This article is the latest in our Fog of Law series that examines the gray zones in international law and conflict that can be exploited by states. The series comes in advance of an expert workshop on the topic at the US Naval War College’s Stockton Center that Just Security is cosponsoring with the Naval War College and the International Committee of the Red Cross.
The current global security environment is characterized – as a thousand commentators have said during the past few years – by a return to geopolitics and the rise of security competition among sovereigns that are technologically advanced peers of each other. The United States, for its part, is seen as stepping back from its post-WWII and especially its post-Cold War role as global hegemon – whether voluntarily or by reason of the increasing relative power of other sovereign actors, notably China and Russia, or some combination of both. The outcome for the global security order is either a world of Great Powers with no global hegemon, no implicit guarantor of last resort of rough international security and stability – or else a world in which the US maintains its role as the only sovereign with truly global security reach, but unable prevent Great Power challengers from gaining dominance, perhaps even their own weak hegemonies, within a given geographic region.
China and Russia, in their UN General Assembly speeches last fall, each argued in favor of a Great Power global order, with no global hegemon. China, for its part, appears to seek both a Great Power geopolitics and a rules-based international order – a global governance model grounded on sovereignty and its prerogatives, particularly non-interference in a sovereign’s internal affairs. Russia affirms the same – in 2016 the two countries’ foreign ministers issued a joint declaration on the principles and meaning of international law – but in practice seems far more opportunistic. Russia, it might be said, seems to be playing geopolitics on a transaction-by-transaction basis. China, by contrast, seems to be playing a long game, one that takes an interest in the overall structure of the “rules-based international order.”
Whether Russia or China, however, their view (shared in part by the United States in hegemonic role) is that the “rules-based international order” is not the same thing as “international law,” at least not as international law has been understood by many in the “international community” since the end of the Cold War in 1990. Without going into that very large topic, one could say that Russia and China have, or might turn out to have, their own strongly state-centric views of international law and governance. They might, and likely do, have their own views on aspects the law of armed conflict (LOAC) and its binding interpretation, including such key LOAC topics as targeting and weapons law.
Thus, for example, though it seems patently obvious to me that in important instances, Russian aerial targeting in Syria has been indiscriminate under LOAC and that to deny it is merely bad faith, we should not ignore what is frankly the far more important possibility that the Russian Federation believes that its targeting is lawful. On this view, gradually tightening standards of interpretation under treaty or customary law of targeting reflect a false consensus that consisted of the Western democracies and their friends among the (Western) NGOs agreeing among themselves. Again speaking on this view, Russia and China stood outside that closed epistemic circle, not powerful enough to make their voices matter, and yet silently disagreeing (or at least refusing to accept the apparent “consensus” as the only binding interpretation).
These differences in interpretation of international law matter, of course, in a world of competitive geopolitics in which competition among Great Powers might escalate into actual conflict. Challenges by Russia and China to important status quo positions undergirded by US hegemony – the status of Russia’s “near abroad,” or China’s claims to large swathes of the South China Sea, for example – have largely been successful. Success naturally invites fresh challenges to the status quo security order, and whatever the long-run aims of either Russia or China (there are other important examples, not just these two) in terms of the international security order, interstate conflict is an increased global risk.
At the same time, none of these leading sovereign competitors wants anything remotely approaching open conflict or war. China and the United States are far too intertwined economically for that to be anything other than a disaster for them and everyone else; Russia is less intertwined with the United States, but it is economically deeply intertwined with NATO and EU countries.
One result of all these political pressures is the rise of what, as a whole category, might be called the “liminal” forms of conflict or near-conflict: Proxy wars, whether through sovereign proxies or non-state actors; so-called “hybrid warfare”; certain forms of cyberattack or “information manipulation” cyber operations; uses of force calculated to be below the threshold at which the United States or its key allies would respond with force, or other than symbolic force – there are other forms of “murky” conflict. One commonality among them, however, is that they are characterized by uncertainty, vagueness, ambiguity, and deniability as matters of deliberate strategy. They are “liminal” in the sense that they destabilize, undermine, and blur categories otherwise used to define forms of conflict and the legal regime applicable to each. The United States, it should be added, has often had recourse to all of these liminal forms of conflict. This is evident in how it has defined and conducted counterterrorism operations since 9/11, but was equally true, taking all sorts of forms of conflict, during the Cold War, where the risk of open conflict might escalate to nuclear war. Seen this way, the rise of liminal forms of warfare is less a “rise” than a “return.”
One reason why these forms of conflict seem to be new, however, is that the categories of conflict – how to group and characterize ways of engaging in conflict – have become increasingly rigid and, well, categorical. Moreover, the rigidity is driven in considerable part because the categories have been increasingly characterized (especially since 9/11 and the rise of the hegemon’s wars with non-state armed groups within and across sovereign borders) ever less by strategic concerns about how to fight in order to achieve one’s strategic aims, and ever more by normative concerns, claims about international law and especially claims about international human rights law (IHRL). The rigidity is expressed, in part, by the conversion of flexible positions – literal UN Charter language notwithstanding – on circumstances for a state to use force, short of full-on armed conflict but nonetheless conducted under the rules of jus in bello, into rigid legal categories – legal binaries, in particular.
Hence, for example, the legal view (widely held, of course, in the international law community) that a conflict with a transnational non-state armed group (if that is even legally possible) only justifies the rules and permissions of LOAC if it rises to a certain level of intensity, otherwise it is necessarily governed by international human rights law. The United States does not take this binary position, and some states that formally accept it appear to do so with pragmatic tweaking of what justifies hostilities-level force in IHRL. The point here is not to argue the substance, but instead to point out that this hardening of categories arose in large part during a period of asymmetric conflict involving a powerful sovereign, but also a hegemonic sovereign with a remarkably loose and accommodating attitude toward other parties’ views of international law – versus a weak, but also legally null, non-state actor. In a new world in which Great Powers jostle and compete, but also want to avoid full-on, direct conflict, however, all these liminal forms of war take on greater strategic importance – including for the United States. In that case, arguably, such legal binaries do not seem legally or strategically apt.
One tendency of the strict legal binaries, after all, is potentially to increase possibilities of open, direct Great Power conflict – by making principals strictly responsible for an ever-wider range of acts of its proxy agents, for example, but thereby making the use of intermediaries less legally relevant. One tends to doubt, to be sure, that sovereigns design their strategic plans for using force according to the second-order incentives of interpretations of legal rules rather than the material conditions of force, but that is only to say that the legal rules as envisioned by many in the international community do not receive the deference they are intended to have. This is, obviously, a very large debate about international law. For present purposes, however, the point is that insofar as these rigid categories are taken to express binding views of international law, both jus ad bellum and jus in bello, they disfavor liminal forms of conflict. This is so even though sovereigns not infrequently conclude, reasonably or not, in good faith or not, that recourse to these forms is the best way to achieve their strategic aims, while limiting the potential fall-out that might occur from more open and direct forms of conflict.
The liminal forms of conflict are not going to go away, in my view. Moreover, the United States and its allies are also likely to have increasing recourse to them – it is not just about Russia in Ukraine, in other words. The “liminality” of rising forms of conflict among states is leveraged upwards, in addition, by the introduction of new technologies of conflict (conflict in the broad sense, ranging from ordinary “peacetime” intelligence activities to actual armed conflict) made possible by advances in AI technologies.
The lion’s share of public attention to AI technologies in national security has been given to (AI-enabled) autonomous weapon systems (AWS). Far less attention has been given to national security applications of AI that are “pure” AI software, rather than a physical robot such as AWS. Yet it is increasingly plausible that AI software agents will have far greater national security impact than robotic weapon systems –sooner and more broadly. Moreover, many of the most consequential applications of the most advanced forms of AI, machine learning (ML) in its several forms, will not be weapon systems in the legal sense (though they might also be components of weapon systems), but instead tools of intelligence gathering and analysis, tools with potentially profound and leveraged impacts.
In what ways do these new AI technologies leverage liminal forms of conflict? Partly through the discrimination and focus upon particular targets (in all senses, not just targeted killing) enabled by new tools for surveillance, individual identification, and predictive analytics. These intelligence tools will be a work in progress, to be sure, and the United States will not necessarily lead in these AI applications. ML at its core is about extracting patterns from large amounts of data, in ways that allow the algorithm to “learn” and improve its performance. Large – very large – data sets are at its core. For good and important civil liberties reasons in democratic societies, however, neither the United States nor many of its democratic allies will come close to what China, notably, is gathering (and intends to gather over time – a relevant factor in the development of data sets) through such program as its Total Information Awareness systems of surveillance.
Such constraints on democratic societies might or might not make a decisive difference in the development of successful national security AI software agents; mass surveillance is not the only basis or function of national security AI applications. But it matters a great deal that China and Russia believe that they have many advantages that make them (or will make them over time) equal to or superior to the United States and its allies in AI development, advantages in which the willingness to engage in mass surveillance is one. Another is their corollary belief that AI technologies are not just dual-use across the national security and general civilian sectors – the civilian and security uses of these technologies are not truly separate, as a matter of technological development and design parameters. And, finally, a belief that AI research, development, and deployment is amenable to central government industrial policy, as China’s recently announced national plan for AI development in effect says.
Whether these beliefs on the part of China, Russia, and other technological sovereign peers are correct or not, or whether they will unfold as hoped or predicted, it matters today that Russia and China believe they are likely to be true. Given the strategic reasons for avoiding direct conflict among the Great Powers; given that many of the consequential applications of AI are likely to be about intelligence functions that favor liminal forms of conflict; and given that Russia and China believe they have (or will have) technological superiority over the United States and its allies, the result is that there is both an increased emphasis on AI by these Great Powers and an increased emphasis on liminal forms of conflict. This conclusion is only as good, of course, as the correctness of these premises.
Although in a sense, national security (especially intelligence) applications of AI tend to favor non-direct, non-open forms of conflict at lesser levels of the use of force, these emerging forms of data-driven, ML-driven, AI software agents have a destabilizing effect on status quo international security politics. This is not a matter of strategic logic, in the way that some nuclear weapon technologies, for example, rationally favor first strikes. Rather, the instability is a function of the belief of both Russia and China, but perhaps China most of all, that they have technological advantages over the United States in national security and intelligence AI applications. An advantage, in particular, over the US’s legacy weapon systems and military structure; much as the emergence of smart phone on cellular networks allowed many in the developing world to leapfrog over expensive and inferior legacy landline telecommunications, Russia and China appear to believe that these technologies enable them to leapfrog fleets of aircraft carriers and jet fighters. It is conjoined with another crucial understanding – that AI in national security applications and cyber are functionally part of the same military and national security system.
Such beliefs, true or not, invite sovereign overconfidence and overreach. They invite destabilizing and dangerous bets against the United States and its allies, often through these “liminal” forms of proxy, hybrid, and less than full-on conflict uses of force (which have the effect, to be sure, of hedging the bet, by avoiding open conflict). But the nature of Great Power competition favors the use of these forms of conflict, and the emergence of new AI technologies leverages them.
There are important consequences in all this for international law and LOAC. One is that sticking strictly by legal binaries and other rigid legal lines that emerged in the period of US hegemony and in the conflicts with cross-border non-state armed groups is not likely to end well, either for the US (which is one reason why it has never signed on to the full extent of these rigid categories) or its allies. It is also not likely to promote the integrity and universal application of LOAC. LOAC has not had to engage, since 9/11 (but, more accurately, since the end of the Cold War) with the meaning, interpretation, and application of the law of armed conflict under conditions of jostling and competing Great Power sovereigns. These sovereigns believe (particularly by reason of their belief of their prowess in emerging AI technologies) that they are no longer quite as constrained as they once were.
Put differently, LOAC – along with many other aspects of post-Cold War law and institutions – is having to address what it means to exist in a world in which it is no longer automatically sheltered under broad, rough global security umbrella that, throughout the post-Cold War, the United States has provided in its role as hegemon. In one sense, the questions confronting LOAC today are relatively narrow ones, about hybrid warfare, about new AI technologies that (as is typically the case) are likely to prove less world-shattering than first promised. In another sense, however, the questions today confronting LOAC are those that necessarily emerge in the post-post-Cold War.