Duty to act on knowledge: precautions, intelligence and the law of armed conflict

Abstract

As Einstein famously said ‘[t]hose who have the privilege to know have the duty to act…’ (Albert Einstein c. 1929). This article takes the precautionary obligations of the Law of Armed Conflict (LOAC) to argue that parties to an armed conflict are obliged to act on their intelligence to minimize the harm to civilians caught up in the conflict. As such, it assesses the relationship between intelligence and the precautions to be taken against attacks under LOAC. It uses the context of Improvised Explosive Devices (IEDs) to illustrate some of the challenges that may be faced in interpreting the law and to demonstrate how significant a thorough understanding of this obligation could be in protecting civilians. I argue that states have an obligation to use their actionable intelligence to protect civilians under their control and for more than merely militarily advantageous reasons. However, the critical challenge is in establishing the scope of the obligation in terms of control and temporality, with limited judicial handling reducing clarity. Nonetheless, I contend that there is a legal obligation on states, beyond the territorial state, to proactively use intelligence during armed conflict to protect civilians from IEDs as well as other evolving threats of armed conflict.

Introduction

The precautionary obligations of the Law of Armed Conflict (LOAC) require states to take all feasible precautions to protect civilians when conducting attacks, and to protect civilians from the effects of military operations. This article assesses the relationship between intelligence and the precautionary obligation to mitigate the effects of an attack to argue that a party to a conflict is obligated to act upon actionable intelligence regarding an attack by the opposing force, which is likely to injure or kill members of the civilian population with certain caveats.

Through detailed analysis of Article 58 of Additional Protocol I (API)¹ and customary law, this article suggests that the framing of the law for international armed conflict (IAC) has limited clarity for its use during non-IAC (NIAC). As a result, the precautions in mitigating the effects of attacks have often been overlooked by courts and scholars, with the precautions in attack outlined in Article 57 API receiving the greater scrutiny. By using other well-developed areas of LOAC, such as the precautions in attack and belligerent occupation, this article demonstrates that the obligation placed upon actionable intelligence during armed conflict is such that states are required to intervene to minimize harm to civilians under their control, when feasible.

While the obligation to protect civilians could have many practical applications, this work uses IEDs to develop and apply the legal discussion. Although reaching the height of attention during the conflicts in Iraq, Afghanistan, and Syria, the threat from IEDs remains with reports of their use during the recent conflicts in Ukraine and Gaza. In the decade to 2020, at least 135,800 civilians were injured or killed by IEDs, with Syria, Afghanistan, and Iraq witnessing the most devastating effects.² The challenges of IEDs are not novel, and states have invested significant resources into counter-IED (C-IED) initiatives, including extensive intelligence gathering and information operations. Accordingly, these challenges are illustrative of the wider balance of intelligence investments and civilian casualties, which should be addressed through the precautions outlined in LOAC.

The article will first discuss the relationship between intelligence and LOAC, before moving on to an analysis of the precautionary regime. It analyses the customary status of Article 58 and assesses its significant differences from Article 57. Notably, while temporality is broadened, there are difficulties presented by the control aspects of Article 58, which are assessed considering approaches from other areas of LOAC and from International Human Rights Law. Finally, this is applied to a specific IED attack in Afghanistan, and we consider its application in contested spaces. It is concluded that while there is an obligation for states to act on intelligence to protect civilians from harm, this is limited to cases where it is feasible, and where the state has sufficient control to intervene.

Intelligence’s relationship with LOAC

Intelligence is used throughout armed conflict by militaries to guide and direct operations, from the strategic level to specific tactical decisions, incorporating a vast array of data collected by both civilian and military organizations. Despite its critical importance of intelligence during armed conflict, it is not mentioned specifically within LOAC,³ thus it must be considered by its significance in enabling states to meet the core principles of LOAC: distinction,⁴ proportionality,⁵ necessity, humanity, and precautions.⁶ It is primarily an operational consideration and while targeting is an inherently intelligence-led process,⁷ even the links in this well-considered area are underdeveloped. However, it is only possible to comply with the principles of LOAC if those deciding upon an attack have knowledge of the real-time situation, situational awareness. This enables militaries to verify targets as military objectives, consider collateral damage and the proportionality of the proposed attack, and balance military necessity and humanity. I contend that this creates a direct link between the precautions of LOAC and intelligence.

The link between information and precautions is made clear by both the DOD and UK MOD in their military manuals, requiring commanders to consider the available information in their decision-making for targeting.⁸ Schmitt explores this further saying: ‘With regard to the specific precautionary obligations in attack, the feasibility of target verification depends on ISR [intelligence, surveillance, reconnaissance] assets….’⁹ One of the difficulties inherent in any study that discusses intelligence is the lack of a coherent definition or understanding of its scope. Particularly considering the difference between information and intelligence, Lowenthal suggests that intelligence always contains information, but information is not always intelligence.¹⁰ Even an examination of the specific terminology of intelligence partners, it is not wholly clear. For example, the USA considers intelligence to be the final product post collection and analysis steps. Whereas, due to the differences in the state architecture, the UK tends to merge the collection and analysis phases, resulting in a less precise definition of intelligence.¹¹ For the purposes of this article, it is the product of the information collection and analysis process that is important. NATO defines intelligence as ‘[t]he product resulting from the directed collection and processing of information … in order to identify threats and offer opportunities for exploitation by decision-makers.’¹² That said, there remains a tendency for the terminology of information and intelligence to be interchangeable, noting that it is phrased as the information available to commanders that is significant for API, rather than intelligence. Despite these terminological differences, the word intelligence is used in this article to mean the final product upon which decisions are, or can be, made.

Intelligence is critical to both the military and strategic efficacy of any armed conflict, as well as its lawful conduct. Its prominence has continued over recent decades with substantial development and investment in the methods and means of intelligence gathering, processing, and dissemination potentially granting greater insight and clarity during armed conflict. On this point, Biddle said: ‘The problem now is not putting a weapon on the aim point, but it’s figuring out the aim point… If you can tell me precisely that Osama bin Laden is at a certain longitude and latitude, we can put a lot of explosives on that point.’¹³ While the focus has mainly been on its role during targeting and attacks, intelligence has a far greater task, encompassing, amongst other things, force protection, strategic planning, mission support, and maintaining situational awareness.¹⁴ Therefore, the range and scope of intelligence held by states during armed conflict is broader than that which could be considered under the precautions in attack, whether in attack or defence.

The increasing intelligence developed by Western states could suggest that there is an increasing obligation to use this intelligence to mitigate harm to civilians during armed conflict. However, even if states hold intelligence, they will not necessarily act upon it, this is a complex decision, and even if they have high confidence in it, which itself is far from certain, they will consider factors such as risk to their own personnel, the possible repercussions, the wider context, and the risk of exposing intelligence-gathering techniques or sources. Furthermore, it is important to recall that not all intelligence provides enough information to act upon, with it ranging in quality, quantity, and detail. Therefore, the intelligence of most interest for establishing its link to precautions is that of actionable intelligence. Again, a universal definition is not forthcoming, but NATO defines this as ‘[i]ntelligence that can be used directly by commanders for the planning and execution of operations as situational information is available in due time and appropriately processed, especially for the tactical level.’¹⁵ Accordingly, it can be understood to be intelligence that provides sufficient detail, in a timely manner, to be acted upon at the tactical level.

An example of actionable intelligence could be that a state party discovers that a non-state armed group intends to place IEDs around a local town square to injure civilians visiting on market day. This state party knows the likely day of emplacement of IEDs, some of the identities of the IED group, and the location of their base. The level of information, while not directly relevant to the opposing force and their combat activities still seems to call for some kind of action. However, as mentioned, states may be reluctant to act upon this, so it is important to establish the scope of any legal obligation that requires action to protect those civilians. If this obligation were found through Article 58, then it would be a requirement to use actionable intelligence to protect civilians from the effects of military activities, which would in principle accord with the overall aims of LOAC. This would be a proactive use of intelligence, so the more commonly considered precautions in attack standards¹⁶ would not be applicable, as this situation does not concern the obligations for conducting an attack, but moreover focuses on proactive protection. Thus, to determine if there is such an obligation to protect civilians from IEDs, or indeed from broader, evolving threats, requires careful consideration of the precautionary obligations as determined by Article 58.¹⁷

Precautions of LOAC

The LOAC obligation on precautions is primarily found within Articles 57 and 58 of API. This test is such that parties to a conflict must take all feasible precautions to minimize the effects of military action on the civilian population. Article 57 is the most widely written on¹⁸ due to its focus on the obligations of the attacking party, such that precautions must be taken in the methods and means of attack. Article 58 is rather less well explored, as Dinniss reflects, authors have generally paid it little more than lip service.¹⁹ It provides the obligations for those who are subject to attack, and has largely been considered with respect to the location of military objectives within civilian areas²⁰ and the movement of civilians by their own government or state power.²¹ However, the two articles are clearly designed to be read in conjunction with each other. As Bothe, Partsch, and Solf write: ‘The obligation to take precautions to protect the civilian population and civilian objects against the collateral effects of attacks is a complementary one shared by both sides to an armed conflict in implementation of the principle of distinction.’²² As such, the duty of Article 58 can be understood to be a mirror obligation to that created and developed by Article 57.²³ However, there are critical differences presented by Article 58, most notably that of temporality. While Article 57 relates to obligations during an attack, and is thus limited to an extent by the end of that attack,²⁴ in contrast, the obligation on those subject to attack is a continuum not restricted by a specific period during conflict.

Therefore, although Article 58 may be less well developed, it creates a requirement that, in some ways, is as significant as the one created by Article 57. As Jensen determines: ‘Art. 58… places an affirmative obligation on those facing attack to either segregate or protect civilians and civilian objects to the maximum extent feasible in order to spare them from the effects of attacks’²⁵ (emphasis added). However, by virtue of its nature in API, it is, on the face of things, critically limited in its application only to IAC and to those state parties of API. The fact that it was placed in API and designed with IAC in mind also creates complications in simply exporting it to NIAC, as it is not clear where the obligation lies and what is meant by the phrasing ‘under their control’.

There are challenges in developing a clear understanding of the role of Article 58, primarily due to the fact that, unlike Article 57, a failure to comply with these obligations will not clearly entail individual criminal responsibility.²⁶ The imbalance in these precautions is such that ‘…while an attacker can commit a grave breach in at least five different ways under API, no similar provisions draw attention to the responsibility of the defender.’²⁷ Thus, restricted to state liability, there has been limited jurisprudential development of Article 58²⁸; nonetheless, the obligation is perhaps greater than at first glance. This is especially so when one considers the third part of Article 58, which states that parties must, to the maximum extent feasible, take ‘… the other necessary precautions to protect the civilian population, individual civilians and civilian objects under their control, against the dangers resulting from military operations.’²⁹

As such, it is valuable to disaggregate Article 58 from its renowned sibling and reflect on the scope of its obligation, and its own qualities. Initially, it is important to establish the customary status of the Article, as this is necessary to determine its application within NIAC and to understand its scope for non-signatory states. Additionally, given that the use of IEDs has predominantly been a tactic of asymmetric or guerilla warfare, its applicability to NIACs is of significant interest.

Article 58 beyond international armed conflict

The customary nature of the standard of care to be taken is advanced by the statements of the International Criminal Tribunal for the former Yugoslavia (ICTY) in Kupreškić.³⁰ The court stated that the principles described by Articles 57 and 58 are considered to be customary as they ‘specify and flesh out general pre-existing norms.’³¹ It furthered the view that although the standards delineated by Articles 57 and 58 allowed wide discretion to the attacking party, they ‘must be interpreted so as to construe as narrowly as possible the discretionary power to attack belligerents and, by the same token, so as to expand the protection accorded to civilians.’³² This is one of the few occasions that Article 58 has been handled by courts, and unfortunately while stating it to be customary, it does little more to aid the understanding of the obligations specifically created by Article 58.

In their customary law study, the International Committee for the Red Cross (ICRC) develops the rule, concluding that ‘[t]he parties to the conflict must take all feasible precautions to protect the civilian population and civilian objects under their control against the effects of attacks.’³³ Beyond the ICRC, the customary status of Article 58 has been generally accepted due to its intrinsic relationship with the overarching principles of IHL, on the basis of ‘common humanity’³⁴ and as a derivation of the principle of protection of civilians.³⁵ Therefore, it is widely agreed to also apply to NIACs.³⁶ This rule, while being different to Article 57, is intimately linked to the overarching purpose of LOAC, which is to minimize civilian casualties during armed conflict.³⁷ To fulfil the principles of distinction, proportionality, and necessity, it is critical to meet the standard of feasible precautions, both as established by Article 57 and by Article 58. The rule, as articulated by the ICRC, reflects the Article 58(c) ‘catch-all provision’,³⁸ stating that parties must ‘take the other necessary precautions to protect the civilian population, individual civilians and civilian objects under their control against the dangers resulting from military operations.’³⁹ The importance here is that it relates not only to dangers presented by attacks, but also to dangers ‘from military operations’. The US DOD Manual states: ‘Outside the context of conducting attack (such as when conducting defence planning or other military operations), parties to a conflict should also take feasible precautions to reduce the risk of harm to protected persons and objects from the effects of enemy attacks.’⁴⁰ This is notable in that it demonstrates that this obligation is ‘outside of attacks’ and continues during planning and other military operations. This broadens the scope of any military operation taking place in an area that places civilians in danger.

As such, there is a legal obligation that requires parties to take precautions to protect civilians from the effects of military operations during NIAC, but the scope and limits of this are not mirrored by Article 57. In both cases, states will need to rely upon intelligence, and for Article 58, this is required to establish the impact of military operations on the civilian population under their control, and then to act on it when it is feasible. Under Article 58, this goes further than the collateral damage assessments required for attacks with the temporal scope being broader than that of Article 57.

Time limits extended

The temporal scope of Article 57 is based on the considerations of the effects of attacks and what is known at the time of the attack, under the ‘circumstances prevailing at the time’.⁴¹ While this should also consider the reverberating effects of an attack, it is time-constrained.⁴² In contrast, Article 58 is an ongoing obligation to protect civilians, under the control of a party to the conflict, from the effects of military operations.⁴³ This almost open-ended temporal scope has a direct impact on the requirements for military intelligence. That is not to say that the military intelligence network is obligated to act beyond what is feasible under the circumstances prevailing at the time, but it does require an appreciation and understanding of the risks of operations to the civilian population. As such, should a party to the conflict identify that an attack is due upon their controlled area, then they have an obligation to take steps to protect the civilian population.

The nature of these steps is frequently understood to mean activities such as the issuing of warnings, the moving of civilians to safer areas, the building of shelters, or the creation of evacuation plans. Accordingly, should a state or military unit receive actionable intelligence of a likely IED attack on a particular stretch of road, or area, then that party is obliged, where feasible, to warn the local population or take other steps to protect them from the effects of this type of military operation. The importance of this was highlighted in Amnesty International’s report and subsequent legal analysis concerning Ukraine’s responsibility to protect its citizens from the effects of the conflict. Amnesty International claimed that Ukraine has undertaken a number of military operations, which directly breach Article 58, including launching strikes from civilian areas and using hospitals and schools as military bases.⁴⁴ While the co-location of the Ukrainian military in civilian structures is concerning, the legal response critiques the claim as being overly categorical, as the precautions required are fundamentally only those that are feasible.⁴⁵ Moreover, while providing warnings to the population is mentioned explicitly in Article 57, it is not repeated in Article 58, and so they merely form one of the methods of compliance.

These various methods of compliance are not detailed, and more than just warnings and air raid shelters are covered by Article 58 with the ‘other precautions’ being interpreted as going further. Jensen argues that these are proactive obligations, and so, in the context of cyber-attacks, governments must act in advance of a potential attack.⁴⁶ This proactive interpretation can be used as a parallel for discussion concerning IED operatives and their networks. Therefore, when intelligence provides an identifiable risk to the civilian population under a party’s control from an IED network, they must take proactive steps to mitigate this. There are various ways in which this obligation could be met, and intelligence is gathered in many forms during (and before) conflict. For example, as the threat from IEDs increased, most notably during the conflict in Afghanistan, there was a substantial response on the part of the USA and their Western allies to develop intelligence surrounding IEDs, which continues to this day.

In 2006, the USA established the Joint Improvised Explosive Device Defeat Organization (JIEDDO) to develop methods to respond to the increasing threat of IEDs. At its height, JIEDDO had a budget of $4 billion, employing around 3,000 people, and in late 2016 transitioned to be the Joint Improvised-Threat Defeat Organization (JIDO).⁴⁷ This agency, while smaller in budget, staff, and acronym, continued the role of JIEDDO to ‘attack the network, defeat the device and train the force’.⁴⁸ Their mission is somewhat broader than purely C-IED, also encompassing support for ‘counter-terrorism, counter-insurgency, and other related mission areas…’.⁴⁹ Nonetheless, it is without question that a significant amount of resources have been deployed in countering the IED threat, and a substantial amount of technological and intelligence development have been driven by the need to provide protection for soldiers encountering these devices.⁵⁰ This has not been restricted to the USA, with states and regional groups continuing to develop capabilities independently.⁵¹ It is also reflective of the wider context of state investment in intelligence collection, analysis, and assessment with billions spent annually.⁵² Therefore, it is important to understand if LOAC requires any of this to be used in the protection of civilians caught up in conflict.

Inevitably, despite significant investments in C-IED technologies, the efficacy of IEDs and the ability to change design and employment methods quickly and cheaply has meant their appeal endures.⁵³ Their sophistication has developed to the extent that IEDs are made on an industrial scale, with reports that ISIL was using off-the-shelf commercial drones to carry explosive payloads to conduct their own form of aerial bombardment.⁵⁴ The use of IEDs as aerial weapons has also been common during Ukraine–Russia conflict,⁵⁵ and the UK MOD has reported that Russian forces have been using old armoured vehicles packed with explosives as vehicle-borne IEDs.⁵⁶ Civilians tend to be disproportionality affected by IEDs, with 1,456 civilian casualties in January–June 2023 alone.⁵⁷ As part of the Western response to the increasing IED sophistication, not only were new technologies developed,⁵⁸ but intelligence efforts were also increased. This was intended to understand the nature of the threat, the scope of the networks involved, and to develop intelligence using a range of sources, from raw tactical reports⁵⁹ through to the complete dossiers provided by the wider intelligence community.⁶⁰ These gave greater insight into the networks of IED operatives, as well as one would expect, their production facilities, from both strategic and tactical viewpoints.

Therefore, there is an ongoing intelligence effort to respond to the threat, in this case from IEDs. This ongoing intelligence operation is then linked to the ongoing obligation provided by Article 58, in that, when a party has actionable intelligence of a threat to civilians from IEDs, they are required to act. While this intelligence is part of the wider context of conflict and remains important for considerations under Article 57, it is required to be acted upon under Article 58. However, this is not an all-encompassing rule, and there are some key limitations that need to be established before we can deduce the point at which actionable intelligence should be utilized in the protection of civilians during armed conflict.

Feasibility in protection

The first of these limitations is the requirement of feasibility, which under Article 58 is stated to be to the maximum extent feasible. This differs from the phrasing used in Article 57 and could be suggested to present a higher standard than that required by ‘all feasible precautions’. However, Boothby reasons that the obligations are complementary and equal in bringing about the protections for those entitled to it.⁶¹ This parity is supported by the Tallinn Manual, which adopts a similar standard, confirming that ‘maximum extent’ merely ‘…emphasizes the importance of taking the requisite measures…’ that are practically possible.⁶² This would align with the understanding that has developed concerning the feasibility of precautions in attack, such that while the obligation to verify is high, it is not absolute.⁶³ As Dinstein states: ‘… there is no way to ascertain with absolute certainty the military character of an objective selected for attack, [but] there is an obligation of due diligence and acting in good faith.’⁶⁴ The good faith and common-sense approach is confirmed by the ICRC,⁶⁵ reflected in the statements made in military manuals,⁶⁶ and by the Committee of Experts in their Final Report to the Prosecutors following the NATO bombing campaign over the former Yugoslavia.⁶⁷ Thus, the standard is one of reasonable endeavours within the context of the ongoing conflict, rather than absolute certainty. As such, the obligation to intervene based on actionable intelligence relating to the potential harm to civilians is limited by what is practically possible.

This understanding of practical measures to gain information for precautions has been supported by Cryer, who in reviewing action within the Afghanistan conflict, states that although ‘… attacks occurred before intelligence gatherers were deployed in the field [this] does not necessarily mean that all practicable steps were not taken. What is practicable is limited by the circumstances…’.⁶⁸ I contend that this is equally important for the precautions against the effects of an attack, as the complementary obligations of precautions are granted parity within LOAC. Thus, there is an obligation on all parties, whether in attack or defence, to maintain an intelligence-gathering network to the extent that it is feasible. The scope of that obligation will depend on several factors, not least of which is the element of control required. The dynamic nature of control during NIAC will mitigate the standard of which is practicable, but it may not remove the obligation. The feasibility criterion is contextual and is based on the information available at the time.⁶⁹

Therefore, the standard of intelligence required concerning ongoing hostilities that may affect the civilian population is contextual, and only that which is feasible under the circumstances prevailing at the time. As such, a lot will depend on what level of control is maintained by the parties, and over what period intelligence gathering was possible. This is particularly relevant for the precautions against the effects of attacks during NIAC, as the control exercised by forces is likely to vary significantly over time. Take, for example, Aleppo, which prior to 2011 was Syria’s largest city and controlled by the Syrian government. From mid-2012, the city was divided into areas of regime and rebel control, and bombing campaigns by the Syrian government, backed by the Russians, reduced areas to rubble. By late 2016, the regime had driven the rebels from the city.⁷⁰ This is not an unusual picture, with conflicts around the world seeing an ever-changing roll call of control over villages, cities, and regions. Therefore, given the requirement of Article 58, and indeed the customary rule, that this obligation is only relevant to aspects ‘under their control’, it is important to understand what level of control this refers to.

Caveats of control

As previously noted, API was drafted with the intention of applying to IAC, and as such, the notion of being under a state’s control was generally less problematic. The Diplomatic Committee Report for Article 58 (draft Article 51), comments ‘[i]t was clearly understood that article 51 applies to all territory under the effective de facto control of a party, that is, including both its own national territory which is under its control and any foreign territory which it occupies.’⁷¹ As such, it can be understood as a de facto standard for IAC, reflecting the provisions in IAC relating to belligerent occupation.

However, in the following period, the lack of judicial coverage of the precautions against the effects of attacks, combined with a lack of wider consideration of the obligation, and as such there is little written on the understanding of this point.⁷² As it is now established as being customary and thus applicable in NIAC, then the determination of ‘under their control’ in this context becomes critical to understanding applicability. It also operates to limit temporality, which, as discussed, is substantially broader than that found in Article 57.

The challenges of establishing that a state has reached the legal standard of meeting control of areas of contested territory within an NIAC are myriad. Moreover, the changing methods and means of conflict, such as cyber warfare, suggest that traditional interpretations of territorial control may not be wholly appropriate for application to modern warfare.⁷³ In consideration of this, it is suggested that effective de facto control for precautions should be interpreted in terms of an ability to intervene, reflecting the Article 57(2) obligations regarding control during attacks. This would mean that not only would the intelligence need to be actionable, but also that the state would need to have the ability to intervene to mitigate the impact on the civilian population at risk and finally only when it was feasible. The legal standard for control has received several discussions in other aspects of LOAC which can be utilized to progress this argument and to develop the dimensions and understandings of de facto control for precautions from the effects of attacks. The standard has also received judicial consideration for the purpose of extraterritorial application of human rights obligations, which can be drawn upon for the present discussion.

There are various methods of considering control⁷⁴; in this context, Article 58 provides for civilians and civilian objects, and obliges parties to: ‘endeavour to remove [them]… from the vicinity of military objectives’.⁷⁵ This could be interpreted as control over the civilian population, but equally over ‘civilian objects’, which could be buildings, installations such as energy generation, water sources, and potentially computer networks. This scope is recognized in part (b), which obliges states to ‘avoid locating military objectives within or near densely populated areas.’⁷⁶ This was the primary aspect debated during the drafting.

Therefore, it would be reasonable to suggest that the control element foreseen by the drafting of the provision relates not only to civilians but also to civilian objects and territory held by the force.⁷⁷ This would support the purpose and aim of the principle, which is fundamentally to take all feasible precautions to protect the civilian population from the effects of military activity, whether that be against military targets co-located, in response to the presence of military objectives in that location, or as part of a wider military effort, perhaps even attacks directed at civilians.

Notably, then for the consideration of obligations relating to networks of IED operatives, it is important to establish at what point a party is said to be ‘in control’. As mentioned, territory can change hands frequently during a conflict, with the most contested areas perhaps rarely reaching a standard that could be said to be under anyone’s control. Indeed, military forces ‘may invade or enter an area in order to pass through it to its intended goal and it may leave that area without establishing any effective control’.⁷⁸ This issue has also been raised in the context of the extraterritorial application of human rights obligations, albeit with a different aim.⁷⁹ As such, to determine the nature and quality of the control required, it is valuable to refer to other aspects of LOAC.

Effective control for belligerent occupation

The de facto standard of control was used by the Diplomatic Committee during the drafting process of Article 58, with its roots in the law of belligerent occupation. However, it is important to recall that the law of belligerent occupation ‘must have an international texture: two or more States must be pitted against each other in an armed conflict’.⁸⁰ Thus, it does not apply to NIAC and is not an aspect of LOAC that would apply to the protections given to civilians in relation to dangers of military activities for NIAC.⁸¹ However, its development of the understanding of control for the application of law is of considerable value when looking at precautions.

The law of belligerent occupation is found primarily in the Regulations respecting the Law and Customs of War on Land, appended to the 1899 and 1907 Hague Conventions. For the purposes of control of territory, Article 42(1) states: ‘Territory is considered occupied when it is actually placed under the authority of the hostile army.’⁸² It develops this further stating that: ‘The occupation extends only to the territory where such authority has been established and can be exercised.’⁸³ The determination here is not subjective but, moreover, an objective standard based on the de facto submission of the territory to hostile forces.⁸⁴

This is a factual test based on whether the party to the conflict can exercise their authority over territory and assume the administrative burden.⁸⁵ As such, effective control of the territory is a conditio sine qua non for belligerent occupation,⁸⁶ and this should be understood as effective de facto control.⁸⁷

The elements of control are considered to be a non-consented presence of military forces, the ability of these forces to exercise governmental functions, and the inability of the territorial state to exercise those powers.⁸⁸ These are qualified by Hague Reg. 43, such that the occupier ‘… shall take all the measures in his power to restore, and ensure, as far as possible, public order and safety…’⁸⁹ For this to be possible, the occupier needs to have some level of stability such that they can exercise their control to meet this responsibility.⁹⁰ With respect to modern warfare, this is developed by scholars to suggest that pure air superiority would be insufficient to reach the level of control that would be required to exercise their authority.⁹¹ It is also insufficient for the position to be one of a vacuum of control, whereby the territorial authority can no longer exercise their control but no one else can either.⁹² The critical question for a nature and quality assessment of effective de facto control for an assessment of the precautionary requirements though is further aggravated by the discourse concerning whether or not this authority needs to be exercised.

In the Armed Activities Case,⁹³ the ICJ determined that the substitution of the authority of the occupier must be effected on the ground, and it was insufficient for it to be merely be able to do so. This has been criticized by scholars who suggest that simply holding the authority to do so should be sufficient.⁹⁴ However, this view is supported by the UK MoD manual,⁹⁵ which comments that the ‘occupying power is in a position to substitute its own authority…’⁹⁶ and does not require that the occupying power has exerted this authority. Dinstein disputes this position, and concurs with the ICJ, by recourse to Article 42, stating that ‘the actual and the potential establishment of authority are cumulative… and it is not enough to satisfy only one part of the equation.’⁹⁷ For the purposes of our discussion concerning the control standard for the application of Article 58, this may not be the case, given the understanding of effective control for the application of Additional Protocol II (APII)⁹⁸ to non-state armed groups. Nevertheless, effective de facto control can be shown to be an objective standard with requirements determined by the ability of a state to exert this control.

Effective control and APII

As we are primarily concerned about the application of precautions against the effects of attacks and their scope for NIAC, it is important to recall the control elements presented by APII.⁹⁹ The applicability of APII to NIAC is based on the quality of control in terms of territory by non-state armed groups. This is determined by Article 1, which determines its application to conflicts ‘…in the territory of a High Contracting Party between its armed forces and dissident armed forces or other organized armed groups which, under responsible command, exercise such control over a part of its territory as to enable them to carry out sustained and concerted military operations and to implement this Protocol.’¹⁰⁰ (emphasis added) This itself has generated substantial discussion concerning the size and scale of the territory required to be held for the Protocol to be in effect.¹⁰¹

The starting point for this difficulty was the understandings raised by states at the Diplomatic Conference, with suggestions varying from ‘non-negligible’ to ‘substantial’¹⁰² in terms of the scale of territory held. Some even considered that any requirement for territorial control to potentially make the requirement unworkable in practice,¹⁰³ with Egypt stating that the requirement ‘… was too restrictive in view of the nature of modern, and in particular guerrilla warfare. In armed conflict situations characterized by high mobility, territorial control continuously changed hands, sometimes alternating between day and night, to the point of becoming meaningless’.¹⁰⁴ This is the familiar position we find ourselves in when attempting to determine if an area is ‘under their control’ for the purposes of the application of Article 58 precautionary obligations and the requirement to use intelligence to protect civilians.

The ICRC recognize this challenge saying that ‘[i]n many conflicts there is considerable movement in the theatre of hostilities; it often happens that territorial control changes hands rapidly… However, there must be some degree of stability in the control of even a modest area of land for them to be capable of effectively applying the rules of the Protocol.’¹⁰⁵ Therefore, the stability required must be such that the armed group is able to apply the rules of the Protocol, a similar requirement to that of belligerent occupation. The key to understanding of the APII application is in its objective nature¹⁰⁶ and the phrase to ‘exercise such control over a part of its territory as to enable’ the group to apply the Protocol. As such it is directly linked to a group’s ability to ‘maintain sustained and concerted military operations’, rather than whether they have, in fact, applied the Protocol.

The International Criminal Tribunal for Rwanda addressed this challenge in Akayesu, further demonstrating the link between the abilities of the armed group and the territorial control required. They determined that the armed forces ‘… must be able to dominate a sufficient part of the territory so as to maintain sustained and concerted military operations and to apply APII. In essence, the operations must be continuous and planned.’¹⁰⁷ Thus, the territorial control element is both objective and qualitative, rather than quantitative. It is not the scale or size of the territory that is under the control of the armed forces that is critical, but rather the ‘intensity of the violence and the ability to comply with the substantive law that is determinative’.¹⁰⁸ This supports the notion that for precautions, the standard of control is directly related to an ability to apply the protocol, or in the case of a proactive action based on actionable intelligence, an ability to intervene to mitigate harm to civilians.

While the standard for effective control is not clearly delineated within API, it enables interpretation on a case-by-case basis, supporting the de facto standard also established for belligerent occupation. This would suggest that the standard of control required for Article 58’s application as a customary rule for NIAC is also an objective and de facto qualification, and does not require a specific quantitative element. Moreover, it is based on the nature of that control and the ability of the armed forces to apply the Protocol; so for precautions, this is feasibility and ability to intervene.¹⁰⁹

Control and international human rights law

Aside from the analysis of ‘under their control’ from within LOAC, it is also beneficial to assess how this problem has been analysed and addressed by International human rights law (IHRL) when considering the extraterritorial application of human rights treaties. It is well established that, during armed conflict, the obligations of IHRL do not cease, except through relevant derogations.¹¹⁰ In discussing precautions under APII, the Diplomatic Conference noted the view that for NIAC ‘[a]n approach placing emphasis on the protection of human rights, rather than on the conduct of military operations, should, it was argued, be preferred.’¹¹¹ The interpretation of LOAC with IHRL has also been substantiated by the ICTY, stating:

…[b]ecause of the paucity of precedent in the field of international humanitarian law, the Tribunal has, on many occasions, had recourse to instruments and practices developed in the field of human rights law. Because of their resemblance, in terms of goals, values and terminology, such recourse is generally a welcome and needed assistance to determine the content of customary international law in the field of humanitarian law.¹¹²

Thus, the use of IHRL to determine the scope and provide interpretative guidance for LOAC is well established.

The difficulty in establishing control under IHRL has primarily been one of establishing state liability for breaches of human rights treaties outside their territory. This extension of jurisdiction is seeking a different purpose than we require for the application of the precautions of LOAC, but nonetheless, the de facto standard is also maintained. The case of Loizidou provides the basic approach, in which the court found that jurisdiction would apply ‘… when as a consequence of military action—whether lawful or unlawful—it exercises effective control of an area outside its national territory. The obligation to secure, in such an area, the rights and freedoms set out in the Convention, derives from the fact of such control whether it be exercised directly, through its armed forces, or through a subordinate local administration.’¹¹³ The point regarding territory is now well-settled for the extra-territorial application of human rights obligations.¹¹⁴

The test of effective control was outlined in Al-Skeini in which the court concluded that it is a ‘question of fact whether a Contracting State exercises effective control over an area outside its own territory…’,¹¹⁵ determining this ‘by reference to the strength of the State’s military presence in the area’.¹¹⁶ Thus, akin to the operation of LOAC for belligerent occupation and APII applicability, IHRL determines effective control as a de facto standard. However, the court has traditionally been strict on this standard, with Banković¹¹⁷ perhaps being the high-water point. This case concerned airstrikes conducted during NATO’s operation over Kosovo, specifically relating to civilian victims of the bombing of the RTS station in Belgrade. Here, the court was not persuaded that air superiority could establish effective control for the enaction of jurisdiction for the European Convention on Human Rights (ECHR),¹¹⁸ refusing to divide and tailor the obligations of the Convention to reflect the circumstances.¹¹⁹ The court ruled that air superiority and the bombing campaign alone could not create effective control for the purposes of establishing jurisdiction for the ECHR. This position would align with the LOAC rules of belligerent occupation, in that air power alone is insufficient to demonstrate control for occupation. However, it is arguable whether this same standard does, or should, apply to the obligation to use intelligence obtained to mitigate harm on civilians, as this is not the same as the need to extend human rights obligations extraterritorially. While the LOAC view for air superiority takes the same approach, this is again focusing on something rather different to the precautions against the effects of attacks. Indeed, for belligerent occupation, there is a requirement that the occupying state should be able to exercise authority over the territory, in the case of precautions, there is no presumed requirement on, for example, establishing law and order, only in mitigating harm from the effects of armed conflict.

Over time, the strict interpretation found in Banković has been eroded, further calling into question the interpretation of effective control, and has expanded the interpretation beyond control of territory to also encompass individuals.¹²⁰ This is interesting for Article 58, as it would be counterintuitive for a state to be found in effective control of individuals, thus be obliged under the ECHR, but not to have the obligation to take precautions to protect those individuals through the use of intelligence. Thus, the decision in Al-Skeini, which determined that the UK had control for security purposes in south-eastern Iraq, also implies that the UK in this region was required to meet the standards of Article 58 and use their intelligence to protect those civilians from the effects of military operations. This could in turn relate to activities such as warning civilians about IED attacks or indeed taking direct military action against IED networks.

Nevertheless, the consideration of effective control for extraterritorial jurisdiction of the ECHR continues to develop but remains a de facto standard. As Milanović contends, the test of effective control is a functional one, such that ‘…[i]ts stringency in the degree of effectiveness required depends foremost on the consequences that attach to the fact of such effective control.’¹²¹ As such, while decisions concerning IHRL have a different objective, the core aspects reflect the effective de facto standard found in LOAC. Significantly, the ECtHR appear unwilling to overturn the decision in Banković, so that contexts such as air warfare remain beyond the scope of effective control for the extraterritorial application of human rights. However, there is a distinction to be drawn between air superiority and the varied situations found in modern conflicts, as well as the crucial disparity in the aims and purposes of IHRL and LOAC. Thus, while effective de facto control remains the consistent standard across LOAC and IHRL, it is not necessarily appropriate to limit the precautionary obligation in an identical manner. Each approach analysed here is responding to the circumstances that it is intending to govern, from occupation by belligerents to the extension of human rights obligations. They all also require some form of ability to demonstrate effective control; for belligerent occupation, this is in the ability to exercise authority, for APII, the need to be able to apply the Protocol. As such, it is reasonable to expect that for the control requirements in precautions, a similar ability would be indicative of effective de facto control.

Control in contested spaces

While the standard required for the application of Article 58 is de facto effective control, it is less clear if this is useful when considering operations in contested spaces. While there are many places that could be used to demonstrate this problem, Afghanistan will be the focus here. That is primarily because of the longevity of the conflict and thus the amount of information that can be found about territorial movements.¹²² It is also illustrative of the complexities that will be found in attempting to apply this legal obligation to wider conflicts. Nonetheless, it is quite simply not possible for an outside observer to be able to say with any certainty that a specific IED explosion was potentially avoidable due to intelligence and control held by the parties to a conflict. This is due to several factors: the sheer quantity of IED incidents, discrepancies in the recording of territorial control,¹²³ the dynamic nature of conflict, and of course the understandable national security issues with knowledge of the deployment and purpose of troops within certain areas, and the challenges in establishing what was known by parties to a conflict at a certain point in time.

A specific example can illustrate the challenges presented and demonstrate how the law could be applied. On 31 July 2019, a roadside IED explosion killed 34 people, including women and children, who were travelling on a bus in Farah province.¹²⁴ This is reported as occurring on the Herat to Kandahar Road, which traverses the far east of the province.¹²⁵ For Article 58 to apply, it becomes necessary to establish who was in control of the region or the civilian population at the time. This is problematic. At the time of the incident in 2019, the US DOD had stopped collecting and reporting data concerning stability within the country. This ceased in October 2018, with the Special Inspector General for Afghanistan Reconstruction (SIGAR) reporting that ‘The DSA [district stability assessment] was of limited decision-making value to the RS [Resolute Support] Commander.’¹²⁶ They noted that there was no other ‘product or forum through which district-level control data is communicated to command’.¹²⁷ The final report on this was in 2018, with SIGAR reporting that ‘about 65% of the population lived in areas under Afghan government control or influence’.¹²⁸ At this time, around 12% of the population were under Taliban control, with the remaining population living in contested areas. This equated to 229 districts out of 407 being under the ‘influence’ or ‘control’ of the Afghan government and their allies. These figures and the preceding years’ data have routinely been challenged, as Cordesman is cited as saying ‘[t]here is no way to be sure of any figure like 70% [government-controlled territory], or to accurately estimate the size and location of the Afghan population.’¹²⁹ Amongst other reasons, he posits a lack of ‘forward presence’ and distrust in Afghan government estimates as being primarily responsible.

Other sources of information may give some indication of the security situation in Farah province at the time of the IED explosion in our example. The Long War Journal suggests that at this time the region was either controlled by the Taliban or contested,¹³⁰ despite SIGAR indicating that it was government controlled.¹³¹ The discrepancies here can often be down to the government having control of a district centre, but limited control beyond that specific area. Therefore, from the data available, it is not possible to know who had control of this specific stretch of road at the time the IED exploded or indeed was placed. However, for the law to apply, we need to establish, on a case-by-case basis, de facto control, considering both the nature of control and the ability of a party to the conflict to apply the law.

As such, it is important to establish whether the government or its allies controlled that stretch of road at that moment in time, but also whether they had any ability to intervene in the transit of the road by the civilian bus or against the individuals involved, what intelligence they had concerning the attack, and finally what was feasible at the time. In considering the legal obligation to protect civilians from the effects of attacks or other military operations, the requirements are not solely determined by the control of the site of the attack but can be argued to be somewhat wider. The placement of that IED may have happened at any point preceding the explosion, the network of operators required to source, design, and build the IED will also be somewhat removed in time and space from the explosion.

Within contested spaces, particularly during NIAC, it will always be a complex calculation, and at times it may be more accurate to say that no party has the ‘civilian population under their control’. However, this may not render the precautions against the effects of attacks impotent, as it can be argued that this is also about the ability to apply the law, or indeed to intervene, take steps, or act upon potential threats to civilians, rather than one of geographic control or even control of people at a specific point in time.

In contextualizing control as the ability to intervene, the obligation is reflective of the LOAC principles found for precautions in attack concerning suspending or cancelling an attack. This is described by the ICRC as ‘Control during the Execution of Attacks’¹³² and is the customary law expression of Article 57 (2)(b), which states that: ‘an attack shall be cancelled or suspended if it becomes apparent that the objective is not a military one … or that the attack may be expected to cause incidental loss of civilian life… which would be excessive in relation to the concrete and direct military advantage anticipated’.¹³³ This rule of control over ongoing attacks is considered customary¹³⁴ with the UK stating that they understood this obligation ‘…only extends to those who have the authority and the practical possibility to do so…’.¹³⁵ Therefore, it would seem appropriate to conceptualize the control requirements for Article 58 in the same manner, and considers control to be understood as one of an ability to intervene, with the practical possibility to do so. This would also concord with the approach taken by Milanović in assessing requirements for IHRL.¹³⁶

This conceptualization of Article 58 results in the applicability of the obligation to take precautions to protect civilians from the dangers of military operations, limited to the extent that it is feasible and when the control is such that the party to the conflict has the ability to intervene. This then would indicate that what is critical for legal compliance in the above example lies in establishing if allied forces had not only intelligence on the placement of IEDs along that major route from Herat to Kandahar but also if they had the ability to intervene. As such, intelligence alone is insufficient; moreover, that intelligence would need to have a reasonable level of specificity for it to be actionable. Consequently, general knowledge of an IED threat in the area would be insufficient. Even specific intelligence may be insufficient to raise challenges for Article 58 if the allied party had no ability to intervene or if it was not feasible under the circumstances prevailing at the time. In this, there is no obligation on states to risk their own forces or operations, and as such, decisional factors would likely include the risk to intelligence sources or methods of collection.¹³⁷

Nonetheless, this still grants Article58 a proactive obligation on the use of intelligence in minimizing or mitigating the harms on the civilian population. For the purposes of intelligence, states are obliged to act upon actionable intelligence when the area is under their effective de facto control, such that they are able to intervene, and when it is feasible to do so. Through this, the temporality of Article 58 is limited and the obligations caveated. However, it extends the protective regime to civilians by placing requirements on states to utilize their extensive intelligence resources not only in the conduct of hostilities but also to minimize the effects of military activity on the civilian population.

Conclusion

This article demonstrates that the obligations on precautions during armed conflict have a direct relationship to the usage of military intelligence. It argued that to adhere to the requirements provided by Articles 57 and 58 states are required to develop and maintain a flow of intelligence in order that they can meet the requirements of LOAC. Due to the lack of judicial consideration of Article 58 and the limited wider discussions, the parameters governing its application are not well established. Moreover, given the original treaty was drafted with the sole purpose of governing IAC, this has compounded the difficulties in finding the scope of the obligation when applied to NIAC.

However, this article argues that despite these difficulties there is an obligation on states to use their intelligence to mitigate the effects of attacks on the civilian population and as such to use their intelligence in a proactive and protective manner, albeit with some important caveats. While the temporal scope encompasses the whole period of conflict, it is limited by the requirement for the civilian population, objects, or territory to be under effective de facto control. This is understood to be qualitative rather than quantitative and to reflect the ability of the state to intervene in the circumstances prevailing at the time. It is also limited by feasibility in line with precautions in attack.

States make considerable investments in the machinery of intelligence, both to secure their home and to cut through the ‘fog of war’. While much is kept beyond the public eye, this does not, in most cases, prevent states from acting upon it. To that extent this article argues that states are in fact obliged to act on their intelligence when civilians are likely to be harmed, but only when feasible, when they hold sufficient control, and when they can intervene.

The challenge of IEDs is used to demonstrate just one area in which there is a substantial role that states should play in removing and mitigating harm to civilians. However, this argument is equally applicable to evolving and unanticipated threats. This then is a potentially significant obligation in its application to NIAC, which extends responsibilities beyond territorial states. Furthermore, it requires states to use their considerable intelligence investments to not only further their military and security goals but also to provide protection to the local populations caught up in conflict.

Footnotes