On 16 September 2026, Trial Panel II of the Kosovo Specialist Chambers (KSC) convicted the former President of Kosovo, Hashim Thaçi, and three other former high-ranking members of the Kosovo Liberation Army (KLA). While the judgment triggered fierce political reactions and violent protests across Kosovo, its legal reasoning has largely remained in the background.

The judgment has doctrinal and procedural shortcomings. We argue that it nonetheless moves international criminal law forward in two respects. Factually, it promotes accountability. Of the KLA members tried before the ICTY, only two were convicted (Bala and Brahimaj), each for crimes against a limited number of detainees. The KSC have now held members of the KLA’s top leadership responsible for war crimes committed on a large scale. Given their later state power, this serves a core aim of international criminal law: ending impunity for the powerful. Doctrinally, the judgment consolidates the KSC’s recognition of arbitrary detention as a war crime in non-international armed conflicts. No other international(ised) tribunal has done so. This, too, moves the law forward – but at the expense of the principle of legality. Like the KSC’s reliance on joint criminal enterprise (JCE), it rests on a doubtful reading of customary international law. We also question the reasoning behind the acquittals on crimes against humanity and the response to sustained witness intimidation.

Our analysis is provisional: pending the public redacted version of the judgment, it rests on the non-authoritative summary and other public case documents.

What the Judgment Achieves

The Panel held former KLA leaders Hashim Thaçi, Kadri Veseli, Rexhep Selimi and Jakup Krasniqi responsible for war crimes: arbitrary detention, torture, murder and cruel treatment. According to the Panel, the four pursued a politically motivated common criminal purpose. The crimes targeted hundreds of victims in Kosovo and northern Albania between April 1998 and June 1999. The victims came from different ethnic backgrounds and included members of the Roma and Serb minorities. Most were Kosovo Albanians perceived as associated with the FRY/Serbian authorities or as political opponents of the KLA, in particular members of the Democratic League of Kosovo and the Armed Forces of the Republic of Kosovo. But rather than convicting across the board, the Panel assessed each count on its own merits. It acquitted the accused of the crimes against humanity charged.

The trial is all the more significant because the KLA leaders were precisely the powerful actors that international criminal law, developed primarily in response to state crimes, is meant to reach. Thaçi, the KLA’s former commander-in-chief, served as Prime Minister, Foreign Minister and President of Kosovo between 2008 and 2020. He resigned to face the charges and is one of the few former heads of state to stand trial for international crimes.

The Trial Panel strives to be conflict-sensitive within the limits of its mandate. The Court of Appeals Panel’s interpretation effectively confines the KSC’s jurisdiction to KLA members (para. 83), leaving the campaign of violence against the Kosovo Albanian population waged by FRY/Serbian forces beyond their reach. Mindful of the risk of one-sidedness, the judgment stresses that the case concerns neither the legitimacy of the KLA’s struggle for independence nor the crimes “Serbian forces and paramilitaries undoubtedly committed against Kosovo-Albanians” (para. 5). With victim-centred reparation proceedings and translations into Albanian and Serbian to follow, the Panel seeks to contribute to transitional justice in Kosovo.

Widespread and Systematic – Yet No Crime Against Humanity

The Panel found that KLA violence against civilians between April 1998 and August 1999 amounted to a widespread and systematic attack. However, in its view, the KLA members selected their victims “mainly and for some of them exclusively” on individualised grounds rather than for “their membership in a targeted civilian population”. It nonetheless found a plan to target “individuals perceived as [o]pponents” (para. 41), a group defined largely by (perceived) political affiliation and, for some of its members, by ethnicity (para. 78). As the ICC has held, such a group can constitute a targeted civilian population (para. 164). The Panel thus leaves unexplained why victims not selected exclusively on individualised grounds fall outside it.

The acquittal weighs heavily, as incidents commencing after the armed conflict ended around 20 June 1999 could not qualify as war crimes (paras 43 et seq.). In the following weeks, Serbs, Roma and Kosovo Albanians perceived as opponents fell victim to abductions, detentions and killings – within an attack the Panel found to have lasted until August 1999. By denying that this attack targeted a civilian population, the Panel leaves these post-conflict abuses unaccounted for: too late to constitute war crimes, yet not recognised as crimes against humanity.

Illegal or Arbitrary Arrest and Detention as a War Crime

With at least 385 victims, “illegal or arbitrary arrest and detention” is the war crime with the most victims in this case. As the KSC apply customary international law, the crucial question is whether customary law criminalised such detention as a war crime in non-international armed conflict at the time of the conduct. In 2021, the Court of Appeals Panel answered this question for arbitrary detention in the affirmative in this very case (paras 90 et seq.). Later appeal judgments in Mustafa (paras 428 et seq.) and Shala (paras 176 et seq.) merely refer to this decision. The legal basis is Article 14(1)(c) of the Law. It covers “serious violations” of Common Article 3 of the Geneva Conventions, which also binds non-state armed groups. The KSC infer from Common Article 3 that detention becomes arbitrary whenever the detaining party disregards the minimum procedural safeguards flowing from the principle of humane treatment.

This reasoning is methodologically deficient. Although the Court of Appeals Panel invoked the Tadić conditions, it did not examine whether every disregard of these procedural guarantees of humane treatment entails grave consequences for the victim. Instead, it presumed that any violation of Common Article 3 is serious enough to constitute a war crime (para. 97). The ICTY, by contrast, never recognised arbitrary detention in non-international armed conflicts as a war crime. We also doubt that the Panel’s sources support its conclusion. The Panel itself described state practice in 1998 as “rather limited”. To prove a “continuing development”, it turned to legislation adopted after 1998 (para. 106). Neither shows that arbitrary detention was already criminal at the time of the conduct, as the principle of legality requires. The Rome Statute, adopted mid-conflict in July 1998, does not list arbitrary detention among the war crimes in non-international armed conflict.

Even if one accepts this reasoning, it covers only part of the charge. The KSC give “illegal” no distinct meaning. They define arbitrary detention as deprivation of liberty without legal basis or without basic procedural safeguards. The Court of Appeals Panel left open whether international law even prohibited detention without legal basis in non-international armed conflict, let alone criminalised it. The Trial Panel nonetheless convicted on both aspects.

Stretching JCE

The Panel based the accused’s responsibility for all war crimes on their participation in a JCE. In its view, all four shared a common criminal purpose, contributed significantly to it and intended the associated crimes. The doctrine stems from the ICTY’s Tadić appeal judgment (paras 185 et seq.), which distinguishes basic (I), systemic (II) and extended (III) forms. Absent an express provision, the Pre-Trial Judge held that JCE I and III formed part of customary international law at the time of the crimes. The Court of Appeals Panel agreed and subsumed JCE under “committed” in Article 16(1)(a) of the Law (para. 138).

JCE requires a common purpose that “amounts to or involves the commission of a crime” (para. 227; emphasis added). A non-criminal aim suffices if the participants pursue it by criminal means. The Trial Panel left open whether the “dual objectives” of independence and political control over Kosovo were lawful (para. 75). Instead, the means rendered the accused’s acts criminal, namely “the systematic arrest, detention, mistreatment and killings” of perceived opponents (para. 80). Yet killings were part of the plan only “if necessary” (para. 83).

The Defence attributed the crimes to uncoordinated violence rather than to a common plan (para. 85). This reading is most plausible for the killings and torture, which neither control over Kosovo nor the detentions required. If so, only JCE’s extended form could be applied: JCE III covers crimes outside the plan that were a “natural and foreseeable consequence”, provided the accused “willingly took that risk” (paras 204, 228). For torture, even this mode is doubtful. The Pre-Trial Judge followed STL and SCSL rather than ICTY case law and excluded JCE III for special intent crimes such as torture (paras 207 et seq.). The Court of Appeals Panel set this ruling aside as premature without questioning its substance (paras 234 et seq.).

So far, the KSC have convicted only under JCE I, although the SPO pleaded JCE III in the alternative – for murder in Mustafa and Shala, and for all crimes in Thaçi et al. Here, too, the Trial Panel applied JCE I throughout. It found that the common purpose encompassed killings as a mere contingency. This effectively overextends JCE I: the common purpose no longer needs to include the crime itself, but only the possibility of its commission in unspecified circumstances. The more contingencies the purpose absorbs, the more the line between shared intent (JCE I) and mere foresight (JCE III) blurs. Precisely this distinction matters, as JCE III is the most heavily criticised form of JCE (pp. 159 et seq.): it makes the accused liable as principals for crimes they merely foresaw as possible. This amounts to objective liability at odds with the principle of culpability. A JCE I stretched to contingencies imports this problem into the basic form without labelling it.

Responding to a “Persistent Climate of Witness Intimidation”

Beyond the substantive issues, witness protection was the central challenge of the trial (para. 24).

The KSC responded by prosecuting the offences against the administration of justice rigorously. So far, all such cases have led to substantial sentences (heaviest: four years three months; lowest: two years). Currently, Thaçi and four co-accused are also on trial for crimes against the administration of justice, connected to influencing witnesses, with the prosecution seeking up to six years for Thaçi.

Additionally, the Panel heard a large share of witnesses (64 of 134) mainly or exclusively in private sessions (Rule 80(4)(b) RPE). Early on, the Panel itself took issue with this and acknowledged “a public interest in these proceedings taking place […] in a public fashion” (p. 4207). More worryingly, the Panel admitted the evidence of 132 witnesses through written testimony (Rules 153 and 155 RPE), according to the Veseli Defence without any counterbalancing acts (para. 43).

Protecting witnesses physically and psychologically from dangers inside and outside the courtroom serves the goal of establishing truth and has an expressive function, affirming the protection of their human rights. However, especially international courts trying core crimes must be aware of the conflicts with the heightened fairness standards in place. We doubt that the Panel adequately considered, inter alia, the Defence’s right to confrontation and the expectation of a public trial – possibly damaging the judgment’s legitimacy.

What the Appeal Judgment Must Deliver

The judgment advances international criminal law. It holds former KLA leaders who later wielded state power to account for systematic war crimes. Its findings on arbitrary detention, crimes against humanity and JCE, however, are doctrinally doubtful. Questions of legality extend to sentencing, since the 25-year sentences for Thaçi and Krasniqi exceed the maximum under the 1976 SFRY Criminal Code. While the Chambers treat such ranges as non-binding (para. 174), critics argue that the lex mitior principle in Article 33(4) of the Kosovo Constitution requires that maximum to be applied.

The full judgment will show to what extent these impressions hold. The Defence and the SPO may file notices of appeal by 27 November 2026. We expect them to target these weaknesses. The Court of Appeals Panel should seize this opportunity to address them. Holding the powerful to account strengthens international criminal law only if convictions and acquittals alike rest on a sound reading of the law protecting the fundamental values of the international community.

The authors are indebted to Arber Ahmeti (University of Vienna) for extensive discussions and critical comments on earlier drafts.