[Ashley S. Kinseth is an international human rights lawyer working on atrocity crimes and statelessness and a member of the Genocide Experts Working Group convened by Genocide Watch. She has conducted field research in Myanmar’s Rakhine State and in Rohingya refugee camps in Bangladesh and India.]
In Part I, I described the elaborate system of permits, arbitrary taxes, village labor quotas, and land seizures and resettlements that the Arakan Army (AA) has imposed on Rohingya in northern Rakhine since 2024. A group able to run such a system comfortably qualifies as an “organization” under Article 7. But the more challenging question remains whether that organization has a policy to attack civilians. Indeed, in Gbagbo, the case collapsed at the no-case-to-answer stage largely because the prosecution could not prove one—even against a defendant with an actual State at his disposal.
Because organizations rarely announce such a policy, it is usually proven from what they do. In Said, the ICC noted that, in principle, an organization committing a systematic attack “will satisfy the policy requirement” (para. 608). Among the factors chambers weigh are a recurrent pattern of violence, the organization’s use of its own resources and structures to carry it out, and the motive underneath. Nor need the policy predate the conduct: it “may also crystallise and develop only as actions are undertaken by the perpetrators” (Al Hassan, para. 1110).
Here, the pattern has recurred in the Rohingya-majority townships of Buthidaung and Maungdaw since the AA took them in 2024, if not always in the same form: in Buthidaung, displaced Rohingya have been largely barred from going home while forced Rohingya labor builds new villages for non-Rohingya settlers on their land, whereas in Maungdaw the settlers have simply been moved into 11 former Rohingya villages. Taxation and extortion reach further still, into central Rakhine’s Mrauk-U, Kyauktaw, Pauktaw, and Minbya. The AA sustains the pattern with its own resources (the farmland it confiscates and hands to settlers) and carries it out through its own structures, which levy labor village by village. As for motive, the Office of the High Commissioner for Human Rights has said the group’s abuses “may reflect an intention of the AA to motivate further displacement and cross border movements to Bangladesh” (p. 7). The High Commissioner’s September report adds that AA officials have even told Rohingya as much at village meetings, warning that any family unable to pay taxes or meet recruitment quotas must leave Rakhine State (para. 25).
The AA’s security concerns are admittedly not wholly invented. Rohingya armed groups allied with the junta in 2024, and the military recruited thousands of Rohingya men—most of them coerced or induced to serve—into its militias. Yet in Said, the chamber accepted that one reason the Seleka raided Bangui neighborhoods really was to find weapons, and still it refused to treat the operations as legitimate “given the manner in which the operations were carried out, the population targeted and the pattern of crimes” (para. 864). Here, a satellite study attributes some 2,400 structures burned in April and May 2024 to the junta and its Rohingya auxiliaries—and another 8,500 to the AA as it retaliated. Whatever the AA’s security worries, they hardly explain the scale of destruction, let alone fees charged only to Rohingya or settlements built by forced labor for Rakhine families.
A rogue-units defense also fails on its face, as conduct administered through meticulous village government for two years is the opposite of rogue behavior. That leaves the argument the AA makes for itself—that it governs for everyone. By way of proof, a sympathetic outlet reports that the group’s ranks have included Muslims “in para-military and police forces, administrative, judicial, and security roles” and cites reports of more than 5,000 Muslim employees. But the High Commissioner finds the AA’s abuses continue “[c]ontrary to its stated policy of inclusive governance and equal treatment” (para. 19). And in any event, even-handedness is no defense. In Al Hassan, the Islamic court in militant-held Timbuktu “treated all defendants who came before it equally” (para. 623), and yet the chamber found that repression through institutions like it was “an inherent and explicit part of the project,” and as such, proof of the groups’ organizational policy (para. 1290).
Why Persecution Fits
In late 2022, more than 400 Rohingya survivors in Cox’s Bazar were asked what justice would mean to them, and fully 86 percent volunteered citizenship. Summarizing her team’s findings, Rebecca Hamilton concluded that “citizenship in Myanmar is a kind of justice that no international court can deliver.”
No court can. But a court can judge the authority that now governs northern Rakhine’s Rohingya. And of the acts Article 7 lists, what that authority does fits one above all: persecution, under Article 7(1)(h). The crime requires a severe deprivation of fundamental rights contrary to international law by reason of the victims’ identity, in connection with another listed act—here, deportation and forcible transfer. Deportation could be charged separately, but persecution is the better fit because it encompasses measures that are not crimes on their own. In Al Hassan, the ICC held that “discriminatory orders, policies, decisions or other regulations” qualify when they “infringe on basic rights and reach the necessary level of gravity” (para. 1202), and it found persecution in the rules militants imposed on daily life in Timbuktu. (Apartheid may come to fit as well, but the Court’s Myanmar investigation has so far rested on deportation and persecution.)
On the identity requirement, the High Commissioner’s report is unequivocal, describing measures “exclusively targeting Rohingya” (para. 22). While residents of AA territory generally need a series of recommendation letters to leave the state, Rohingya need a paid permit just to reach the next village. The land seizures and mosque demolitions go further, as does the reported renaming of villages, which would wipe Rohingya names off the map. And the people fare no better than their villages: the High Commissioner finds the AA “mirroring the military policy of denying Rohingya identity, referring to them as ‘Bengali’” (para. 19).
The group is the Rohingya, targeted on ethnic and religious grounds, as ICC judges accepted in 2019 (para. 110). The fact that the AA employs some Muslims defeats neither ground, because nothing in the law requires that persecution touch every member of a faith (Al Hassan, para. 1206), and the AA singles out Rohingya, a name it will not even use. Of course, Hindu and Mro villagers have also been put to forced labor, and were the AA to admit to the practice at all, it could cite them to argue that its abuses are indiscriminate. But persecution does not require that only one group suffer. Forced labor may be shared, but arbitrary taxes and village-to-village permits fall on Rohingya alone, while it is Rohingya land doled out to others—among them Mro and Khumi settlers.
The Rohingya thus suffer by virtue of their very identity, stripped of the freedom to move, to hold land, to worship, to keep what one earns, and to own one’s labor. In isolation any one of these measures might pass for hardship, but stacked against one group they amount to the severe deprivation of fundamental rights that persecution requires, with the connected crimes already documented.
Where the Case Goes
The choice of category governs what, if anything, the ICC can do about the AA’s abuses. All these happen inside Myanmar, which is not a party to the Rome Statute, and the ICC has not acted on the opposition National Unity Government’s 2021 declaration accepting its jurisdiction. The Court’s only working route into Myanmar runs through Bangladesh and was premised on the crime against humanity of deportation. In 2018, ICC judges held that deportation “initiated in a State not Party to the Statute … and completed in a State Party” falls within the Court’s jurisdiction, and that the same reasoning may reach other crimes with an element on State Party soil, naming persecution as their first example (paras. 73–76). The Prosecutor’s application against Min Aung Hlaing relies on that reasoning, seeking a warrant for persecution “committed in Myanmar, and in part in Bangladesh.”
On the facts here, however, that route is closed to war crimes, because none of their elements take place on Bangladeshi soil. The order that moved survivors into the camp at Nassawr Para, which a prosecutor could charge as ordering displacement (art. 8(2)(e)(viii)), was a complete crime the moment it was given, inside Myanmar, as would be an order sending Rohingya into Bangladesh—all before anyone reached the border.
Thus as Aakash Chandran noted in this blog’s recent Asia-Pacific symposium, even with the ICC’s limited jurisdiction over deportation and persecution, “an accountability gap remains” for crimes by “non-state armed groups like the Arakan Army.” That gap is real, but the Court already has a way across part of it.
The Independent Investigative Mechanism for Myanmar finds that AA conduct has produced “a new influx of Rohingya refugees into Bangladesh” (para. 37), and the High Commissioner counts nearly 150,000 Rohingya newly registered there since early 2024, driven in part by AA threats and forced recruitment (paras. 25–26). Deportation need not happen at gunpoint; the Elements of Crimes recognize that people can be “forcibly” deported by “taking advantage of a coercive environment” (art. 7(1)(d), n. 12), which the AA has cultivated. Nor does charging Rohingya hundreds of dollars to leave somehow make their leaving voluntary. Those deportations pass through AA-controlled checkpoints and end in Bangladesh, and thus, by the 2018 reasoning, fall within the Court’s jurisdiction in their own right. So too may persecution piggyback on that jurisdiction, all the way back to the mix of terror and red tape that drove people out of Myanmar. This is what the application against Min Aung Hlaing assumes, though the judges have yet to rule publicly.
Still, the existing investigation already leaves room for the AA. Authorizing the inquiry in 2019, ICC judges stated that the Prosecutor “is also not restricted to the persons or groups of persons identified in the Request” (para. 126), and that later crimes remain within scope “as long as such crimes are sufficiently linked to the situation” (para. 133). Rohingya once again driven into Bangladesh—this time by the AA—would seem to meet that test, as would the AA’s resettlement of villages emptied in 2017, which helps keep the 2017 deportees from returning. Indeed, the 2018 decision suggested that preventing return could itself be a crime against humanity, committed partly in Bangladesh (paras. 77–78).
In 2019, the AA did not govern northern Rakhine. Today it does, and the Office has not said publicly whether its investigation reaches the group. In the Court’s proposed 2027 budget, the Office describes itself as “nearing completion of the investigation stage in this situation, which is limited to crimes committed on the territory of Bangladesh” (para. 51). AA deportations would fit that description, and the Investigative Mechanism, which shares evidence with the Court (para. 49), is already investigating the group. Fortify Rights asked the Office in July 2025 and April 2026 to bring the group’s crimes within its scope. Before that stage closes, the Office should do so, or at the very least explain why not when it announces the investigation is complete.
The ICJ, for its part, cannot touch the group at all: for however historic the Court’s pending judgment on Myanmar’s responsibility for genocide may be, it will bind a State that no longer controls northern Rakhine, where the 2017 operations took place and to which any “safe and dignified return” (para. 54(b)) would have to occur, should the conditions ever allow.
The Arakan Army, which does control that territory, wants to be treated as the government of Rakhine, and its claim to that mantle rests, rather perversely, on the strength of the administration it has deployed against Rohingya. No doubt Bangladesh and the UN agencies seeking access to Rakhine have practical reasons to deal with it. Yet whatever they decide, the AA can hardly wield a government’s powers without also answering for what is done in its name. It has been signing that name all along, including on the one-day permits Rohingya must buy just to visit the next village.
