About the author(s):
Yousuf Syed Khan is Manager, Law and Policy at Legal Action Worldwide in Geneva. He is also a Nonresident Senior Fellow with the Strategic Litigation Project at the Atlantic Council, and an Associate Fellow at the International Centre for Counter-Terrorism – The Hague. His work focuses on the use of siege warfare, attacks against objects indispensable to the survival of a civilian population, and forced displacement as a warring strategy.
Yousuf has over 15 years of legal experience dealing with complex conflict situations, with specific expertise on the contributions and practice of UN atrocity inquiries. He has served on four commissions/investigative accountability bodies established by the UN Human Rights Council, regarding situations in Syria, South Sudan, Belarus, and Ethiopia. In these capacities, he conceptualised and led the drafting of over a dozen public UN reports, including the first-ever report by a UN-mandated mechanism on starvation as a method of warfare.
He also worked in Ukraine supporting the Government to prosecute starvation crimes; on human rights with the UN in Afghanistan; led a team monitoring the post-ISIS administration of justice countrywide with the UN in Iraq; and served with the Trial Chambers of the International Criminal Tribunal for the former Yugoslavia (ICTY) in The Hague.
This post forms part of phase two of the Beyond Compliance Symposium: How to Prevent Harm and Need in Conflict, hosted by the Armed Groups and International Law blog. The introductory post can be found here. The symposium invites reflection on the conceptualisation of negative everyday lived experiences of armed conflict, and legal and extra-legal strategies that can effectively address both civilian harm and humanitarian need.
The starvation of civilians as a method of warfare is prohibited under treaty law (Art. 54 of Additional Protocol I and Art. 14 of Additional Protocol II), recognised under customary international humanitarian law (Rules 53 and 54), and criminalised in the Rome Statute (Art. 8(2)(b)(xxv) of the Rome Statute for international armed conflicts, extended to non-international armed conflicts by a 2019 amendment (Art. 8(2)(e)(xix))). Still, there has never been an international judgment interpreting starvation as a war crime (at the national level, see Public Prosecutor v. M.P. et al. (Croatia), and the on-going trial of Jihad A., Mahmoud A., Mazhar J., Sameer S., and Wael S. before the Higher Regional Court of Koblenz (Germany)).
For the first time, in November 2024, Pre-Trial Chamber I of the International Criminal Court (ICC) granted the Prosecutor’s request for arrest warrants in the Gaza situation, finding reasonable grounds to believe that the civilian population in Gaza was intentionally deprived of objects indispensable to survival (OIS) under Art. 8(2)(b)(xxv) of the Rome Statute, including food, water, medicine and medical supplies, as well as fuel and electricity. Several parameters of the war crime remain judicially untested, however, including: the requisite mental element (the relationship between deprivation and intent), the scope of what may constitute OIS (eg, fuel and electricity), and the treatment of aggregated conduct.
The last of these matters significantly, because starvation in contemporary armed conflicts has arisen through patterned forms of aggregated deprivation across the basic systems people require to survive. Across the sieges of Mariupol (Ukraine), Tigray (Ethiopia), and Gaza, for example, what emerged goes beyond isolated denials such as blocked aid convoys or attacks against markets and bakeries, or even their systematic deprivation. In each context, deprivation resulted from a combination of the conduct of hostilities, bureaucratic decisions, and economic restrictions that, taken together, dismantled the very systems through which civilians ordinarily accessed food, water, healthcare, and other essential items.[1] Analysing each act of deprivation in isolation obscures how starvation can actually be provoked in recent wars – so, too, does framing deprivation as systematic for a crimes against humanity assessment, which can capture the pattern but not necessarily how each act of deprivation compounds the others.
This is also a problem that compliance with international humanitarian law (IHL) alone cannot resolve, since, as the Beyond Compliance Consortium notes, the rules regulating the conduct of hostilities tolerate significant harm (pp. 17-18). A systems-based analysis of starvation similarly reveals that, even where individual acts of infrastructure destruction, administrative restriction, or access denial might each fall below the threshold of the starvation war crime or outside its ambit, the cumulative systems-level effect can and often does produce starvation.
How contemporary deprivation is produced
In practical terms, the systems through which civilians are able to access what they need to survive operate across three interdependent layers:
- The first comprises primary OIS. Though examples in the treaty text are non-exhaustive, Additional Protocol I, Article 54 and Additional Protocol II, Article 14 include foodstuffs, crops, and livestock, or the items whose deprivation directly threatens life.
- The second comprises the enabling physical infrastructure that makes the primary OIS accessible, with examples in the treaty text including drinking water installations and supplies and irrigation works. Here, and especially during siege warfare, energy grids, water treatment and distribution networks, cold chains, healthcare facilities, and transport routes are increasingly relevant.
- The third comprises the administrative, economic, and informational framework that coordinates access to primary OIS through the enabling physical infrastructure, such as banking, telecommunications, civil service payroll, commercial supply chains, humanitarian coordination mechanisms, and regulatory approvals. Tier 3 is not expressly captured by the war crime as drafted: AP I Article 54 and its Rome Statute analogues address physical objects and their enabling infrastructure, not the administrative and economic systems that govern access to them. Whether deliberate disabling of those systems can nonetheless be reached through the existing prohibition – as a method by which deprivation of Tier 1 OIS is produced – is an unresolved doctrinal question.
Still, each layer depends on the others, and, though the dependencies vary by conflict context, disabling any one layer can eliminate civilian access to primary OIS. Stated plainly, some food may exist in markets but cannot be purchased if banking is suspended, water infrastructure cannot function and water cannot be desalinated if the energy supply is severed, and healthcare facilities cannot operate without fuel, electricity, or meaningful medical resupply.
The civilian harm literature describes these kinds of downstream consequences as “reverberating effects.” The framework offered below draws on that but applies it differently, to account for the known dependencies between these layers and for forms of deprivation – banking freezes, salary suspensions, telecommunications shutdowns – that are not necessarily the consequences of an attack at all.
In reconstructing how starvation unfolds, and drawing on my own investigations, the three conflicts examined below illustrate how the interdependence of these layers has operated in Mariupol, Tigray, and Gaza, with each revealing a different pattern of their interdependence.
Mariupol, Ukraine
Between February and May 2022, Russian forces and affiliated actors laid siege to Mariupol City and its environs. Within the first week of the siege alone, pro-Russian forces destroyed all 15 electricity inputs leading to Mariupol, which demonstrated how quickly urban survival could unravel when a single service in the interdependent layers of the system was destroyed.
As the city’s water ran on electric pumps and its heating on boilers pumping hot water to heating outlets, including radiators, the large-scale destruction of electricity inputs meant that access to water, heating, and telecommunications were simultaneously eliminated, in temperatures reaching as low as -12°C (pp. 8, 15, 18-19, 30). Without telecommunications, residents were also unable to find remaining food/water distribution points or available shelter, both of which were then targeted in tandem, including famously the Mariupol Drama Theatre that was a makeshift food/water distribution point sheltering up to 1,300 displaced civilians.
Even where food was available, residents were unable to cook without gas and ended up burning furniture for cooking and warmth, and melting snow to drink water or drinking directly from defunct radiators. At the same time, humanitarian aid convoys were regularly blocked. The large-scale attacks against electricity inputs – coupled with the denial of humanitarian aid – meant that residents were deprived not just of food, but of the means to store, prepare, or reach it.
Hundreds of thousands of residents were trapped in Mariupol City during the siege while each deprivation compounded the last. Electricity was the single point on which the enabling infrastructure, telecommunications, and humanitarian coordination all depended. Destroying it disabled water, heating, and communications simultaneously, and the effects spread laterally within the second tier and across the second and third tiers at once. Pro-Russian forces then reinforced the resulting deprivation by destroying not only formal supply networks but also the self-help mechanisms through which civilians attempted to cope and subsist.
Tigray, Ethiopia
For two years, between November 2020 and November 2022, systems-based deprivation during the siege of Tigray was characterised by the deliberate disconnection of an entire region from the state-level banking, telecommunications, energy, and commercial networks on which the civilian population depended. On 3 November 2020, the day fighting began, the Federal Government cut off electricity, internet, telecommunications, and banking across Tigray, and began looting and destroying OIS in tandem, on what the United Nations International Commission of Human Rights Experts on Ethiopia had described as “an industrial scale” (para. 23). Where hospitals survived the systematic looting, they were rendered non-operational without electricity and fuel (paras. 208-210, 214).
After Tigrayan forces retook the capital, Mekelle, on 28 June 2021, and the federal military withdrew from much of the region, the Federal Government reimposed the blanket shutdown, including by freezing bank accounts, forcing the closure of more than 600 bank branches (para. 212), suspending salaries, and blocking fuel and commercial imports. Even where markets still functioned, people could not buy food because they had no access to their money. Likewise, humanitarian organisations, although present, could not deliver because only 15 per cent of required cash entered the region (paras. 221-224).
The siege of Tigray demonstrated that the administrative and economic systems of the third tier, namely banking, payroll, and telecommunications, could cut off access to primary OIS even where the enabling infrastructure of the second tier remained physically intact. The dependency did not run primarily through physical infrastructure destruction but rather through the deliberate disconnection of the systems that rendered the infrastructure usable.
Gaza
The “total siege” of Gaza, particularly between 8 October 2023 and 19 January 2025, exemplified how systems-based deprivation operates within a closed environment, where some 2.3 million residents were trapped under sustained bombardment due to the closure of the Rafah (Egypt), Kerem Shalom, and Erez (Israel) crossings.
The deprivation of primary OIS was reinforced from every direction. Israeli forces began by cutting electricity to the enclave within days of the attack by Hamas and affiliated armed group members on 7 October 2023. From then on, water desalination ceased, hospitals lost power, food cold chains failed, and sewage treatment stopped, contaminating remaining water sources and accelerating disease within an overwhelmed healthcare system that was itself collapsing from the same energy denial.
An 11-week total blockade in 2025 allowed zero aid trucks despite three provisional measures orders by the International Court of Justice (ICJ) (here, here, and here). By 3 April 2025, almost all flour distributions had been suspended, and all 25 UN-supported bakeries were forced to close, due to depletion of cooking gas and flour stocks.
In an open environment, disruption to one food source can be partially compensated by civilian movement to areas with improved access. Across Gaza, the inability to leave meant that displacement and starvation operated simultaneously and reinforced each other. Displacement deepened deprivation, as repeated directives to evacuate pushed populations into progressively smaller areas where infrastructure had already been destroyed, concentrating more and more residents into areas with less and less capacity to sustain them. In such a sealed environment, every act of deprivation became additive and irreversible (paras. 85, 110-126). This was further exacerbated by killings at aid sites operated by the Gaza Humanitarian Foundation (GHF). Overall, one modelling study estimated that life expectancy in Gaza fell from a prewar 75.5 years to 40.5 years during the first 12 months of the war alone, roughly between October 2023 and September 2024, representing a 46.3 per cent decline.
Gaza demonstrates how all three tiers can collapse simultaneously within a sealed environment. Primary OIS were restricted directly through closed crossings and depleted stocks of food, water, medicine, cooking gas, and flour. The second tier was disabled through the denial of electricity and fuel, taking down hospitals, bakeries, cold chains, water systems, and sewage treatment in tandem. The third tier – humanitarian coordination, evacuation directives, and crossing controls – determined whether anything still available could actually reach civilians. Deprivation did not move from one tier to the next but accumulated across all three at once, and the absence of exit routes left civilians with no fallback.
Notably, the ICC’s Pre-Trial Chamber I appears to have recognised this kind of interdependence, by listing fuel and electricity next to OIS even though neither are mentioned as illustrative examples of OIS in the treaty text, noting that cutting them “had a severe impact on the availability of water in Gaza and the ability of hospitals to provide medical care.” This opens the door to treating restrictions on fuel and electricity as central to the deprivation analysis.
The mental element and individual criminal responsibility
Under the Rome Statute, the war crime of starvation requires intentionally using starvation of civilians as a method of warfare by depriving them of OIS (Arts. 8(2)(b)(xxv) and 8(2)(e)(xix)). Article 30 of the Rome Statute provides the default mental element, requiring that the material elements of the crime be committed with intent and knowledge. In relation to a consequence, Article 30(2)(b) provides that a person has intent where that person means to cause that consequence or is aware that it will occur in the ordinary course of events. Whether this default standard suffices for starvation, or whether the crime demands a dolus specialis – ie, a specific purpose to starve, analogous to the special intent threshold for genocide – remains judicially unsettled, and the ICC has not interpreted it yet. The source of the debate is the Elements of Crimes for Article 8(2)(b)(xxv), which requires that “the perpetrator intended to starve civilians as a method of warfare.” Whether that phrasing displaces Article 30’s default consequence standard, so that nothing short of purpose will do, or whether it leaves the knowledge limb intact is what remains unresolved. This contribution does not seek to resolve that doctrinal debate, and argues only that a systems-based account is compatible with whichever standard of mens rea the Court ultimately adopts.
This is the hardest question for a systems-based account of starvation, because the systems analysis reveals patterns of deprivation that unfold across multiple actors, decisions, and timeframes. The risk is that identifying systemic causation collapses the mental element into a constructive knowledge or “should have known” standard, which is a line criminal law does not cross. The three-tiered framework described above provides a structure, but it does so as an evidentiary method for investigators to reconstruct how deprivation is produced, and not as a substitute for proving intent or knowledge on the part of an accused individual.
A related question concerns individual criminal responsibility. The three-tier framework above reconstructs how deprivation is actually produced across the tiers, but not which individuals could be responsible, or on what basis. That analysis runs through Article 25 of the Rome Statute and its modes of liability. The orchestrator of the pattern may be reached as a principal under Article 25(3)(a), whether through co-perpetration or indirect perpetration through an organisation, while lower-level actors who contribute to discrete acts would require separate analysis of their contribution and mental element under Article 25(3)(c) or (d). The three-tier framework does not render every administrative act within an aggregated pattern automatically criminal, but reconstructs how deprivation was produced, leaving modes of liability to do their conventional work.
Applying the mens rea
The clearest case involves the deliberate deprivation of first-tier OIS. Where a decision-maker orders the destruction of food stocks, the contamination of water sources, or the blocking of medical supplies, and does so with the purpose of depriving the civilian population, the mental element is straightforwardly satisfied.
The harder cases arise at the second and third tiers, where the deliberate severing of enabling infrastructure or governing systems deprives civilians of access to first-tier OIS. This is where the systems framework matters most, because the causal chain between conduct and consequence runs through the interdependencies between the tiers.
Certain indicators could be used by investigators assessing whether the mens rea can be inferred in systems-based deprivation cases, when four cumulative conditions, proposed below, are met.
First, the component must be functionally critical, meaning that, within the specific conflict context, the civilian population’s access to first-tier OIS depends on the continued functioning of that component.
Second, the component must have low substitutability, meaning that, at the time of the conduct, no reasonably available alternative existed through which the civilian population could access the first-tier OIS in question. In Mariupol, for example, the destruction of all 15 electricity inputs had low substitutability because the city had limited independent generation capacity and was simultaneously besieged. By contrast, if a population has realistic access to alternative water sources, the destruction of a water treatment plant, while potentially unlawful on other grounds, does not necessarily cross the starvation threshold because of the substitutability. Similarly, while low substitutability for electricity in a besieged city is almost self-evident, low substitutability for formal banking in a region where informal economies, barter, and humanitarian cash may partially compensate is a harder factual question that would require more assessment and granular proof. The framework therefore places the most evidentiary weight on the factual record of whether such alternatives were realistically available at the scale required, not merely present in principle.
Third, a deprivation threshold must be crossed, meaning that the disabling of the component must eliminate or reduce civilian access to first-tier OIS below the level necessary for survival of the affected population. This could be assessed against benchmarks such as caloric intake falling below emergency thresholds, water availability dropping below the recommended minimum of 15 litres per person per day, or the capacity of the healthcare system falling below the point at which treatable conditions become fatal.
Fourth, the accused must have actual awareness of these dependencies – that the component was functionally critical and lacked substitutes. For example, a 2008 Israeli government document on “Food Consumption in the Gaza Strip – Red Lines,” known as the “Red Lines” paper, had calculated the precise caloric relationship between supply and nutritional outcomes, which demonstrates institutional knowledge of how the system worked and what would happen when systems-level deprivation was carried out.
When all four conditions are satisfied, a systems-based analysis may support the relevant mens rea inference. It may support, at least, awareness under Article 30(2)(b) that starvation would occur in the ordinary course of events. If a specific purpose reading is adopted by the ICC, the same facts may be probative of purpose, although they would not remove the need to prove that starvation was deliberately used as the method. The burden for satisfying the four conditions varies by tier, and third-tier conduct requires more contextual evidence to establish functional criticality and low substitutability than second-tier infrastructure destruction.
One important limitation remains, as the framework does not extend to cases where starvation was merely a foreseeable risk among several possible outcomes.
What a systems lens offers
The main objection is that a systems lens risks treating every major urban siege as a potential starvation crime, since modern cities depend on interlocking infrastructure and any sustained campaign may degrade access to OIS. Here, the central claim is not that the starvation crime should be expanded to cover all disruption of civilian systems, but that where deprivation is produced through those systems, it cannot be assessed without reconstructing how civilians actually accessed OIS.
Concerning aggregated conduct, no single tier-3 administrative act in Tigray, for example, including the freezing of bank accounts or suspending civil service salaries, might independently satisfy the elements of the starvation war crime, standing alone. Each, too, could be defended on administrative, economic, or security grounds. When taken together, however, and over months, these acts dismantled the system through which 6 million Tigrayans accessed food, water, medicine, and the financial means to obtain them. The open question is whether the starvation war crime can capture such conduct.
Additionally, the prohibition requires that starvation be used “as a method of warfare,” which, too, has not been judicially interpreted yet. On a narrow reading, this requires that starvation be used instrumentally, that is, as a means of securing military advantage. On a broader reading, the intentional starvation of civilians is itself a method of warfare within the meaning of the provision, and the question of motive does not arise. The latter is in line with the ICRC commentary on AP I, Art. 54, which notes that using starvation “as a method of warfare would be to provoke it deliberately, causing the population to suffer hunger, particularly by depriving it of its sources of food or of supplies” (para. 2089). If the narrow reading prevails, the harder cases are those where deprivation appears punitive, coercive, or administrative rather than clearly connected to a further military advantage.
Under a systems analysis, where a belligerent disables one of the three interdependent layers knowing it will affect access across multiple survival functions and then reinforces the resulting deprivation through further acts of denial, the cumulative pattern may constitute a single method of warfare operating through the system, not a collection of isolated decisions each falling below the threshold of the crime. In this sense, the pattern is the method. Who can be held individually responsible for a pattern built from many actors’ decisions is a separate question, left to the modes of liability discussed above.
A third problem concerns military necessity. Under Additional Protocol I, dual-use objects such as electricity grids may qualify as military objectives where their destruction, capture, or neutralisation offers a definite military advantage (Art. 52(2)). Attacks against them, however, are prohibited where the expected incidental civilian harm would be excessive in relation to the concrete and direct military advantage anticipated (Art. 51(5)(b)). Even where OIS are used in direct support of the enemy’s military action, they may not be targeted where doing so may be expected to (i) leave the civilian population with inadequate food or water so as to cause starvation or (ii) create conditions that force their movement (Art. 54(3)(b)). Whether electricity itself could qualify as OIS remains unresolved, though the ICC’s inclusion of fuel and electricity within the deprivation analysis, again, suggests the Chamber recognises the interdependence. The framework does not resolve how to distinguish lawful dual-use targeting from deliberate starvation strategies in every case – that analysis requires the Article 51(5)(b) proportionality assessment and the Article 54(3) exception to be worked out on the facts – but it does provide a structure for identifying when the cumulative pattern of targeting and denial goes beyond what military necessity can plausibly support.
A systems analysis exposes a gap when starvation arises not from any one attack but from the cumulative degradation of the interdependent layers across multiple operations. Where a belligerent attacks a power grid and simultaneously blocks humanitarian access, prevents repair crews from restoring civilian supply, freezes bank accounts, and suspends the salaries of the civilian workforce – acts that carry no military justification – then the pattern demonstrates that the objective is not the military advantage from disabling the grid but the starvation effect from the resulting failure. A systems lens therefore gives investigators a basis for distinguishing between legitimate military operations with incidental effects on civilian survival and potential deliberate strategies that exploit the interdependencies to produce starvation.
The limits of a systems lens
As laid out, starvation in contemporary conflicts is often produced by cumulative deprivation across interdependent civilian survival systems, and any legal analysis that isolates acts risks missing how the crime is actually inflicted. A systems analysis of starvation is not meant to expand the war crime, but rather to make visible the causality between a deliberate act and the deprivation of what civilians need to survive. In Mariupol, Tigray, and Gaza, that pathway ran through the (known) interdependencies between physical infrastructure, administrative systems, and the primary objects on which civilian life depended. Disabling one layer predictably collapsed others.
Where decision-makers understood those dependencies and maintained or intensified the resulting deprivation rather than mitigating it, the systems analysis might support an inference of the required mens rea without necessarily lowering the legal threshold or broadening the definition of the starvation war crime.
The systems lens framework also marks its own boundary – where deprivation was foreseeable but not accepted or pursued, for example, where administrative systems produced starvation-like conditions without evidence that any individual intended or knowingly sustained them, the systems lens reveals a gap that cannot be closed by reinterpreting the starvation prohibition. It points instead to the broader project of civilian protection that the Beyond Compliance Consortium (BCC) describes, in which the question is not only whether the law was violated, but whether following it was enough.
[1] Under IHL, “essential supplies” refers broadly to items used for humanitarian relief consignments (AP I, Arts. 69-70; GC IV, Arts. 23, 55, and 59), while OIS is an explicitly protected category whose deliberate deprivation is prohibited (AP I, Art. 54(2); Customary International Humanitarian Law Rule 54).