On May 29th, a drone struck a residential building in the Romanian town of Galați, sparking a fire and injuring two people. With a 50-kilogram high-explosive payload, the impact could have caused a tragedy of far greater proportions. Romanian President Nicușor Dan identified the drone as a Russian Geran-2 and suggested it had likely been hit by Ukrainian air defenses, which altered its trajectory into Romanian territory. In response, NATO Secretary-General Mark Rutte warned, “Russia’s reckless behavior is a danger to us all …. And last night showed yet again that the implications of their illegal war of aggression don’t stop at the border,” while European Commission President Ursula von der Leyen observed, “Russia’s war of aggression has crossed yet another line.”
But Russian President Vladimir Putin refused to accept responsibility for the incident, accused Romania of jumping to conclusions, and suggested the drone might have been Ukrainian. Dmitry Medvedev, Deputy Chair of Russia’s Security Council, issued a chilling message to “Citizens of EU countries”: “You should realize your authorities have unilaterally entered into a war with Russia. So be vigilant and don’t be surprised by anything. The peaceful sleep is over.”
Incursions into EU airspace have been frequent since Russia’s full-scale invasion of Ukraine began in February 2022, some of which have been destructive or injurious. For instance, the Institute for the Study of War reports that the May 29th incident was at least the 28th such penetration of Romanian airspace alone by Russian drones. Some of the incursions into European territory have involved non-Russian systems. For instance, in May, three Ukrainian drones crossed into Latvia, likely due to Russian electronic warfare countermeasures, with one exploding at an oil storage facility. And, as this article was being prepared for publication yesterday, reports emerged that fragments of a Ukrainian drone had landed in Moldova and that a French jet shot down a drone over Latvia, with the Latvian army claiming the incursion to be “a result of Russian electromagnetic warfare.”
The uncertainty over Russian intentions, or the lack thereof, in the Romanian (see here) and other intrusion incidents (see here) raises important questions about how unintended engagements between States, a term we use to denote incidents in which a State’s actions cause physical harm to another State unintentionally, are regulated under international law. Such engagements are not new and can take many forms. For instance, a State may intend to use force against a particular object but be mistaken about the object’s identity or the threat it poses. In an 1988 incident still studied in War Colleges today, the USS Vincennes shot down Iran Air Flight 655 over the Strait of Hormuz, killing all 290 civilians on board the airliner. The warship intended to shoot down the aircraft, believing it was a military aircraft in an attack profile, not a civilian airliner.
Or a State might intend to use force against one State, but unintentionally cause consequences in another, as has been hypothesized in the Romanian incident. That kind of possibility is especially acute in the cyber context, where effects can bleed over from a targeted system into another State’s cyber infrastructure in unforeseeable ways.
A State might also intend to act but not intend to cause a harmful consequence, as when an intelligence, surveillance, and reconnaissance (ISR) drone crashes, resulting in physical damage or injury on the ground. Or, an intervening cause, such as the electronic warfare defenses of another State, could cause or contribute to the consequences, as has been speculated vis-à-vis the Latvian incident (see discussion here). And a State may not have intended for an act to occur in the first place, for instance, when an AI-enabled system operates in circumstances not envisioned by those deploying it.
In this article, we ask whether a State’s conduct violates the prohibition on the use of force in international law, even when the conduct was not intentionally directed at the injured State. This question is increasingly significant in light of emerging technologies. The NotPetya ransomware attack, for example, reminds us of the risk of unintended, dispersed, global, and indiscriminate effects of cyber-enabled operations. And concerns about unintended engagements during the military use of artificial intelligence have dominated inter-governmental and multi-stakeholder discussions on autonomous weapons and AI-enabled operations.
Beyond its reputational repercussions, classifying an act as a prohibited use of force carries significant legal consequences. As an “internationally wrongful act” under the law of State responsibility, it can open the door to “countermeasures” by the injured State. That is no small matter, for it would allow the injured State to conduct actions that would otherwise be unlawful but for the fact that their purpose is to induce the “responsible State” to cease its unlawful conduct. The law of State responsibility also allows for reparations for harm caused and for the use of countermeasures to secure those remedies. Additionally, under the jus ad bellum, when a use of force qualifies as an “armed attack,” it triggers the right to use necessary and proportionate force in response under Article 51 of the UN Charter and customary law.
We do not address these or other consequences here. Nor do we take on the unsettled issues of the meaning of prohibited “force” and a potential “de minimis threshold” for qualification as a use of force. We are also setting aside the issue of intent in the context of an armed attack under jus ad bellum self-defense, as intent need not necessarily be treated in the same manner when assessing uses of force and armed attacks; we leave that matter for another time.
Instead, our analysis begins at the point where the requisite harm for a use of force violation has occurred. We conclude that the state of the law is unclear. In our estimation, while States have been navigating this ambiguity for decades, there are good reasons to seek further clarity on the boundaries of the prohibition on the use of force, especially in light of emerging technologies, such as artificial intelligence.
A requirement of intent in the prohibition on the use of force?
The law of State responsibility comprises the rules that determine when a State is legally responsible for violating an international obligation, in this case, the prohibition on the use of force against other States. In large part, the United Nations International Law Commission’s Articles on State Responsibility are regarded as reflecting those customary law rules. They treat responsibility as arising from the breach of a specific primary rule attributable to a State and impose no general requirement that the State have acted intentionally (art. 2 commentary). Whether intent matters, therefore, depends on the content of the specific primary rule.
The determinative question is therefore whether the prohibition on the use of force is breached solely by the objective fact of an act involving a projection of force by one State against another, or whether the rule contains a further subjective element that must be satisfied for the conduct to amount to a breach. According to some – notably, our friends and University of Reading colleagues Marko Milanovic and Russell Buchan – there are strong reasons to accept that the prohibition on the use of force establishes an objective standard, one devoid of a subjective requirement. Others, such as Christian Henderson, consider intent an essential characteristic of a prohibited use of force (The Use of Force and International Law, pages 125-128). Adopting a nuanced distinction between intended action and intended effects, Erin Pobjie suggests that while “the text of article 2(4) strongly indicates that at the very least, an intended action is required,” it “does not explicitly require or exclude an intended effect” (Prohibited Force, page 158).
This takes us to the rule itself, which is found in both Article 2(4) of the UN Charter and customary international law. For the purposes of the present analysis, we focus on the former, noting that the International Court of Justice (ICJ) has found them to be substantially the same (Paramilitary Activities, ¶¶ 176-179). Article 2(4) provides:
All Members shall refrain in their international relations from the threat or use of force against the territorial integrity or political independence of any state, or in any other manner inconsistent with the Purposes of the United Nations.
As a matter of treaty law, we must follow the customary methodology of treaty interpretation, as codified in the Vienna Convention on the Law of Treaties (VCLT), to establish the content of the prohibition.
Ordinary Meaning: The first step is to interpret the provision “in good faith in accordance with the ordinary meaning to be given to the terms of the treaty in their context” (art. 31(1)). In this regard, the Article 2(4) prohibition does not explicitly require intent, recklessness, negligence, or any other subjective element. But neither does it explicitly rule out such elements.
Certain textual clues point toward a subjective requirement. Note how the provision emphasizes that States must refrain from using force in their international relations. Though implicit, the provision embeds an element of inter-State interaction, presumably deliberate rather than happenstance. Textually, this interpretation is further evidenced by the provision’s prohibition on force “against the territorial integrity or political independence of any State, or in any other manner inconsistent with the Purposes of the United Nations.” The notion of being “against” implies an object against which the force is directed. As defined in the Oxford English Dictionary, the term refers to “[e]xpressing … action in opposition to someone or something,” as in “active hostility or opposition to; so as to fight with or attack.” That said, textual analysis alone is not dispositive.
Context: “Context” is also to be considered in interpretation, with context referring, in part, to “the text, including its preamble” (VCLT art. 31(1) and (2)). In this regard, the Charter’s references to the “armed attack” (art. 51) and “aggression” (arts. 1, 39, 53) are especially relevant when interpreting the prohibition on the use of force.
In Oil Platforms, the International Court of Justice took the position that the concept of armed attack is conditional on intent when it observed that “a Silkworm missile fired from (it is alleged) more than 100 km away could not have been aimed at the specific vessel, but simply programmed to hit some target in Kuwaiti waters.” Along the same lines, the Court observed, “There is no evidence that the minelaying alleged to have been carried out by the Iran Ajr, at a time when Iran was at war with Iraq, was aimed specifically at the United States; and similarly it has not been established that the mine struck by the Bridgeton was laid with the specific intention of harming that ship, or other United States vessels” (¶ 64) (emphasis added). And debates during the drafting of the UN General Assembly Resolution 3314, Definition of Aggression, illustrate how State representatives were keen to keep accidental and mistaken acts outside the scope of the prohibition on aggression (Corten, The Law against War, pages 136-137). Taken together, these contextual cues suggest that the Charter was intended to address situations in which one State intentionally uses force against another.
While these contextual considerations point to a subjective element in the prohibition on the use of force, we acknowledge that armed attacks and aggression, as well as the prohibition of “threats” in Article 2(4) (which definitionally are intentional, Nuclear Weapons, ¶ 47), may arguably be decoupled from the use of force. For instance, the prohibition on the threat of force – as a distinct rule of international law, separate from the non-use of force – may well be more demanding in terms of intent. Similarly, armed attacks and aggression, generally viewed as acts of greater gravity than the use of force (Paramilitary Activities, ¶ 191), might require intent even if the underlying prohibition on the use of force contains no such requirement. And in any case, regarding the notion of “armed attack,” some have criticized the ICJ’s subjective approach in Oil Platforms, instead suggesting an objective reading of the concept.
Object and Purpose: Not only must treaty provisions be interpreted in good faith according to their ordinary meaning and context, but they must be interpreted “in the light of [the treaty’s] object and purpose” (VCLT, art. 31(1)). The Preamble of the UN Charter most directly articulates the instrument’s object and purpose. It provides that the goal is “to save succeeding generations from the scourge of war, which twice in our lifetime has brought untold sorrow to mankind.” In light of the very deliberate uses of force that characterized the outbreak of World War II, the drafters certainly had in mind intentional uses of force against other States. They sought to counter such intentional force through the comprehensive prohibition in Article 2(4). (We reserve the question of whether the Security Council and States using force in self-defense can lawfully act against unintended projections of force for further research. In our view, an interpretation of Article 2(4) requiring a subjective element does not preclude this possibility.)
Subsequent Practice: Article 31(3)(b) of the VCLT provides that the “subsequent practice of States in the application of the treaty which establishes the agreement of the parties regarding its interpretation” … “shall be taken into account, together with the context” (art. 31(b)). When State practice is sufficiently dense to show that the Parties have reached such an agreement, it serves as an authentic means of interpretation. But if it is too limited or inconsistent to establish an agreed interpretation, it can still be used as a supplementary means of interpretation under Article 32, for example, to confirm an interpretation derived from Article 31 or to resolve ambiguity, obscurity, or manifestly absurd or unreasonable results that persist after applying the other means of interpretation. For instance, the ICJ recently relied extensively on State practice in accordance with Articles 31(3)(b) and 32 to interpret International Labor Organization Convention No. 87 (Advisory Opinion, Right to Strike under ILO Convention No. 87).
As highlighted in the literature (e.g., by Olivier Corten in The Law Against War, pages 137-138 and Russell Buchan in Artificial Intelligence and the Prohibition on the Use of Force: Intention and Causation), States have, on some occasions, condemned unintentional projections of force as Article 2(4) violations, and, on others, refrained from doing so. In addition to the difficulties of navigating this inconsistent practice, it bears emphasis that States may, for any number of reasons, choose to remain silent even when they consider themselves victims of prohibited force. Such silence could, for instance, be attributed to a desire to de-escalate or preserve alliances, or due to intimidation or coercion. Similarly, a State’s rhetorical labeling of an action as a prohibited use of force, or even of aggression, may be made for geopolitical or domestic reasons, and need not reflect its actual legal position. That said, such statements cannot be discounted, for they may, in fact, be the invocation of a legal position even if for ulterior reasons; the point is that they must be approached cautiously.
With that in mind, it remains noteworthy that no State has yet accused Russia of violating the prohibition on the use of force in relation to the incident in Romania. As reported by the BBC, “Romania’s military has sought to reassure the public that it was not an attack on the country but rather ‘a conflict at our border, with consequences for the local population.’” Also significant is the fact that after a Ukrainian air defense missile crashed in Poland in 2022, then-NATO Secretary-General Jens Stoltenberg emphasized that “this is not Ukraine’s fault” and that “Russia bears responsibility for what happened in Poland yesterday, because this is a direct result of the ongoing war and the wave of attacks from Russia against Ukraine yesterday.” And national governments, NATO, and EU leaders have not labeled the most recent incident in Latvia, which involved damage that may have been unintentionally caused, as a wrongful use of force. So far, the pattern is that governments and international organizations tend to speak in terms of airspace violations, provocations, threats, and the risk of escalation, rather than formally classifying the drone incidents as uses of force under the UN Charter. Again, there are compelling reasons not to do so, if only to avoid the conflict bleeding into the NATO space; but the omissions in official positions cannot be ignored.
In addition to statements about specific incidents, subsequent verbal practice in applying the UN Charter appears in national positions on the application of international law to cyberspace (available through the Cyber Law Toolkit); there is no question that they can be legally significant vis-à-vis Charter interpretation (National Position Handbook, page 18). While most States that have issued such positions have refrained from directly addressing the question of subjective elements in Article 2(4)’s prohibition, some have offered relevant comments. For instance, according to Australia, the analysis “involves a consideration of the intended or reasonably expected direct and indirect consequences of the cyber activity” (emphasis added).
The fact that most national and common positions do not specifically address our question could be interpreted in at least three ways. First, it could be construed as pointing toward the absence of a subjective element. Second, States may have decided to reserve their position on this question for the time being. Third, it may be that the prohibition on the use of force is so clearly engaged only when there is the relevant subjective attitude (of intent, or perhaps recklessness or negligence) that they saw no need to mention the subjective element.
Concluding Thoughts
Overall, no aspect of the treaty interpretation analysis yields an unambiguous answer to our question as to whether the prohibition on the use of force includes a subjective element of intent. What is clear is that a violation occurs when a State deliberately engages in forcible action against another State, intending to cause the requisite use of force consequences. Assuming, for the sake of discussion, that we are right in concluding that a subjective element resides in the prohibition, it is less clear whether mere recklessness, or perhaps even negligence, could satisfy that element. The language of “reasonable expectation” of effects, adopted in some positions on the use of force in cyberspace, could suggest that some States hold that view.
We reserve judgment on that issue pending further research and thought. But at first glance, such an interpretation may prove challenging. Consider, for instance, the fact that Russia is aware that Ukrainian electronic warfare defensive systems can, and are, causing its drones and missiles to sometimes be deflected in the direction of other European States. Does this mean that every time it employs such systems, it uses force against neighboring countries because of the reasonably foreseeable risk their use would cause harm there at the use of force level? Would the same apply to Ukraine, whose own weapons are being redirected by Russia toward the Baltic States? Perhaps those two scenarios would not count given the intervening acts of the other State. But then what about Ukraine firing air defense missiles close to another State’s border to protect its own civilians and civilian infrastructure from Russia’s relentless attacks on them? These are particularly difficult scenarios, given the fact that there is no requirement that the requisite effects of a use of force actually occur.
Sometimes it is advisable to adopt a more conservative position regarding the scope of international legal obligations, lest a broader view lead to undesirable legal, moral, and political consequences. Such caution seems implicit in the reactions of States and international organizations to the tragic incident in Romania and similar incidents that preceded and followed it. We find their approach wise, both legally and strategically.
But we close by emphasizing that none of this is to say that Russia did not violate international law when its drone crashed in Romania, narrowly avoiding a major catastrophe, or in the previous and subsequent incidents on European territory. Other rules of international law can capture the incidents and entail Russian legal responsibility, including for reparations. The question we address in this article is narrower, namely, the boundaries of one of the foundational rules of the international legal system – the prohibition on the use of force.
