[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.]
Rohingya fleeing Myanmar’s northern Rakhine State pay $300 to $500 for documents to get past Arakan Army checkpoints—to smugglers, to village administrators, or to the armed group itself (para. 26). But the costs of being Rohingya in Rakhine are not limited to leaving. Villagers say a one-day permit to visit a neighboring village will run them up to 5,000 kyat (about $2.40) and require two signatures. Staying home is not much of a savings opportunity: in parts of the north, homes, shops, boats, fishing nets, cattle, and rice paddies are all taxed. The UN High Commissioner for Human Rights labels the practices “arbitrary taxation and systematic movement restrictions exclusively targeting Rohingya” (para. 22).
The permits, at least, are not new. In the summer of 2017, weeks before the military’s clearance operations began driving more than 700,000 Rohingya into Bangladesh, nearly every Rohingya I spoke with in Rakhine offered the same complaint: visiting a neighboring village—let alone another township—meant layers of travel authorization, expensive and hard to come by. But today, one of the signatures on those permits belongs to the Arakan Army (AA), an ethnic Rakhine armed group that captured most of the state in 2024 and has added to the old controls forced labor quotas and the seizure of Rohingya land.
Most of those who have documented the group’s conduct call it war crimes, when they classify it at all. But that label fails to account for the more mundane daily realities that have come to define life in northern Rakhine in the wake of the fighting. Among the weapons turned on Rohingya today are tax schedules, permit counters, labor rosters, and land allocations. Together, these seemingly benign bureaucratic layers form a system of rule that by design falls hardest on Rohingya simply because they are Rohingya.
Crimes against humanity is the category the law built for exactly such a scenario. It deals with the question of what an authority does to the people in its power, and it also happens to be the frame in which the International Criminal Court (ICC) can already act.
In such cases, the authority in question need not be a State; indeed, the ICC has already convicted members of armed groups from the Lord’s Resistance Army to the Seleka of crimes against humanity. Against such groups, the key element is proof of an “organizational policy” to attack civilians. And here, the AA’s own case for being treated as a government supplies much of that proof.
What the Record Now Shows
In May, Human Rights Watch documented the AA’s 2024 massacre of Rohingya civilians at Hoyyar Siri. AA fighters opened fire on fleeing villagers, some of them waving white flags. More than 170 people were killed or remain missing, at least 90 of them children—and a survivor who slipped back months later found “heaps of skeletons and skulls scattered everywhere.” The AA denies responsibility.
What followed was not combat. In February 2025, the group ordered survivors into a nearby camp it controls at Nassawr Para. No one leaves without permission, and residents are routinely warned against exposing the massacre. They are put to construction work without pay, and each week some are taken in turns to tend the looted livestock.
The 2026 annual report of the Independent Investigative Mechanism for Myanmar, released in August, moves from violence against people to the remaking of territory. It has collected evidence of alleged rapes of Rohingya women, summary executions, and the forced recruitment of children. It goes further, finding that “[r]eminiscent of the 2017 clearance operations, entire Rohingya areas have been cleared out by the Arakan Army.” The cleared land does not stay empty: its investigations “show that the Arakan Army is destroying mosques and repurposing such land for non-Rohingya resettlement,” activity that has “increased substantially since 2024” (para. 37).
The High Commissioner’s September report confirms the pattern and demonstrates how it works. Rather than let Rohingya uprooted by the fighting go home, it finds, the AA has been “settling non-Rohingya communities on Rohingya land” (para. 20). Satellite imagery shows large-scale changes in land use, with an estimated 2,000 acres reportedly seized. Some of the newcomers get new settlements built on that land; others are moved into 11 former Rohingya villages, some of them emptied in 2017. The High Commissioner also raises “serious concerns about a reported process of renaming Rohingya villages,” and interviewees told the UN that Rohingya were forced to build not only their own camps but also the newcomers’ homes and buildings for the AA.
Across northern Rakhine, the labor is organized village by village, “with each village compelled to provide labourers to the Arakan Army for a certain number of days each week or month” (para. 21). In January 2026 alone, local sources recorded 2,353 such instances, including 75 involving children. Witnesses described workers as young as 6 clearing jungle, putting up buildings, and even doing menial work for Rakhine families.
None of this diminishes the case against the Myanmar military. Myanmar remains the respondent at the International Court of Justice (ICJ). Some Rohingya organizations have called the AA’s conduct genocide as well, but on this record the charge does not fit. As the ICJ held in Croatia v. Serbia, genocidal intent must be the only reasonable inference from the evidence (para. 148), and here the evidence points at least as readily to an intent to drive Rohingya out—or to subjugate those who stay—as to an intent to destroy them. In substance, those are deportation and persecution, the two crimes against humanity in the ICC Prosecutor’s arrest warrant application against Min Aung Hlaing, the former military chief now installed as president, on which the judges have yet to rule publicly.
Why War Crimes Falls Short
Reporting on the AA, by contrast, has largely stopped at war crimes, though not because rights groups shy away from larger findings. When the Myanmar State confined Rohingya behind checkpoints, permits, and movement restrictions over a period of years (and in some cases decades), Amnesty International and Human Rights Watch each found it amounted to the crime against humanity of apartheid. So part of the issue here is timing: roughly two years into AA rule, the reporting proceeds incident by incident, yielding incident-level crimes.
Yet the deeper problem is doctrinal. The Rome Statute’s war crimes can reach the order that moved survivors into a camp or the seizure of Rohingya land and livestock, but not the discrimination that ties them together. Article 8 also offers no general crime of enslavement or any crime of forced labor; it can offer no name for a village quota that sends people out to clear jungle and put up buildings for an armed group.
Applying the laws of war, Fortify Rights has cited the prohibition on outrages upon personal dignity in warning that the AA’s forced labor “may constitute war crimes.” For the labor itself, that is about as far as the Statute’s war crimes reach—and charged that way, perhaps the most systematic conduct in the record would be judged by how degrading or harsh the work was, not by the fact that the laborers were forced to work at all. The Office of the Prosecutor conceded the gap in its 2024 policy on slavery crimes (“[f]orced labour is not an express crime under the ICC Statute”) and pointed to the provision that does reach it, noting that when forced labor involves the exercise of powers of ownership over a person, it is enslavement—a crime against humanity under Article 7(1)(c). Some of the AA’s forced labor may meet that threshold—the confined, unpaid workers at Nassawr Para come closest—but much of it may not.
The ICC exposed a second weakness inherent to a war crimes framing on September 23, when it convicted Mahamat Said Abdel Kani, a Seleka commander in the Central African Republic, of crimes against humanity—torture among them. Yet it acquitted him of torture as a war crime—alongside cruel treatment and outrages upon personal dignity—because the majority could not find an armed conflict was still underway once the Seleka had seized power, and even if one was, all three judges found he was unaware of it.
In Rakhine, by contrast, fighting between the AA and the Myanmar military continues today. Thus for now the laws of war do reach every township the AA controls “whether or not actual combat takes place there,” as the ICTY held in Kunarac (para. 57). But should the junta’s offer of peace talks ever produce a settlement that ends hostilities, efforts to charge the continuing AA abuses against Rohingya as war crimes would presumably meet the same fate as the counts that failed in Said. By avoiding the need to prove an armed conflict and a link to one, then, a crimes against humanity framing offers a more durable basis for prosecution. What it requires instead is a widespread or systematic attack on civilians pursuant to a State or organizational policy—elements which, if the AA’s actions to date are any guide, would surely outlast the fighting.
The Policy Element: The Arakan Army’s Own Evidence
Against a non-State actor, proving a policy is usually the hard part. The other contextual elements—a widespread or systematic attack on civilians—are not seriously in doubt. The attack is made up of acts Article 7 lists: the killings at Hoyyar Siri, the confinement of survivors at Nassawr Para, rape, forcible transfer and deportation, and the persecution examined in Part II. Conduct spanning six townships for two years is widespread; conduct run on quotas, schedules, and permits is systematic; and those on the receiving end are civilians. What remains is Article 7(2)(a)’s “organizational policy,” which the ICC treats as two requirements that are easily conflated: an organization and a policy—one that, in the words of the Elements of Crimes, “actively promote[s] or encourage[s]” an attack against a civilian population (Introduction to art. 7, para. 3).
The organization question is not close. Running quotas and permits on this scale requires an administration, which the AA has in the Arakan People’s Revolutionary Government—a body that oversees eight districts through more than 10 departments, runs a three-tier judiciary, and even taxes households by income band. The ICC asks for far less: structures, “whatever those may be,” that are “sufficiently efficient to ensure the coordination necessary to carry out an attack,” as it repeated this September in Said (para. 606). The AA would pass even the narrow test Judge Hans-Peter Kaul urged in his 2010 dissent on Kenya, which would require an organization to “partake of some characteristics of a State” (para. 51).
On that half of the test, the AA has supplied the evidence itself. The other half (a policy to attack civilians) is harder to prove, and it is where Part II picks up, together with the question of which crime best fits the record and how the Court might reach it.
