On 23 September 2026, the International Criminal Court convicted Mahamat Said Abdel Kani of crimes against humanity for the Bangui OCRB detentions of 2013, and acquitted him of the war-crimes counts charged on the same facts. The facts of that verdict have been widely reported, including in this firm’s account of the conviction, the coming sentence and reparations. This article covers what comes next, and it starts from a document most summaries skip: the 14-page separate opinion of Judge Maria del Socorro Flores Liera, published with the judgment. The judge agrees with the convictions, and, crucially, she even agrees with the outcome on the war-crimes counts, while demolishing the majority’s reasoning on the existence of an armed conflict in the Central African Republic in 2013. With the Prosecution and the Defence each holding a 30-day right of appeal expiring around 23 October 2026, that disagreement is not academic commentary. It is the map of the appeal. This analysis reads the separate opinion line by line, explains who can appeal what and with what prospects, and sets out what the appeal phase changes in practice for victims, witnesses and suspects connected to France: the reparations track, the possibility of a French prosecution, and the procedure if France is asked to surrender someone to The Hague while the appeal is pending.
I. Can the Said Verdict Still Change Before 23 October 2026?
Yes, and the starting point is the Court’s own procedural notice. The questions-and-answers document issued with the verdict states: Both sides may appeal the trial judgment in whole or in part within 30 days of issuance, and any appeal goes to an Appeals Chamber of different judges ruling on the grounds raised. The press release of 23 September 2026 adds that the Defence and the Prosecution may appeal within 30 days, that the Chamber will then decide on the sentence after receiving submissions and holding a separate hearing, and that a reparations process will begin for the victims of the crimes of conviction. Mr Said remains in detention in the meantime. Three clocks therefore run at once: the appeal clock, the sentencing clock and the reparations clock. The appeal is the only one that can still alter the verdict itself, and the separate opinion shows exactly where it is vulnerable.
A. What Does the Dissenting Judge Say About the War That Allegedly Stopped?
The majority acquitted on the war-crimes counts of torture, cruel treatment and outrages upon personal dignity because, by majority, it was unable to establish beyond a reasonable doubt that an ongoing non-international armed conflict existed during the charged period, holding that Bozize’s ouster on 24 March 2013 represented a complete defeat of the Central African armed forces and ended the conflict that began in 2012. Judge Flores Liera’s separate opinion of 23 September 2026 rejects that finding root and branch. She agrees that a non-international armed conflict existed from at least December 2012 between the Central African armed forces and the Seleka, but parts company with the majority on whether that conflict survived the Seleka takeover of Bangui on 24 March 2013, finding the evidence sufficient to conclude that it continued. Her counter-thesis has three limbs, each documented across the opinion’s fourteen pages. On the parties, she holds that the Anti-Balaka were a splinter group of the Central African armed forces which, as a non-state armed group, carried on the fight against the Seleka after 24 March 2013, recalling that deserting soldiers carried away what weapons they could, hid the rest with the intention of recovering them, and regrouped at Gobere and Zongo while Bozize and his circle organised from exile. She concludes that the parties to the conflict before and after 24 March 2013 were continuous. On organisation, she concedes the Anti-Balaka started small and could not mount full confrontations between April and June 2013, but finds they always retained the capacity to mobilise, gathered, armed and trained in the bush, attacked provincial locations in September 2013 with a common military goal, and by December 2013 were capable of fighting that forced Michel Djotodia’s resignation, a capacity built, in her assessment, through concerted efforts begun immediately after Francois Bozize’s ouster rather than in a matter of weeks. Requiring regrouping within weeks, she writes, “imposed an unnecessarily high standard of assessment.” On intensity, she holds that the months without armed confrontations between the Seleka and the Anti-Balaka were a mere lull in hostilities rather than proof that the armed conflict had ceased, pointing to the April killing of two Seleka in Bangui, the August attempt to disrupt Djotodia’s inauguration, the Seleka’s high-calibre retaliation on Boy-Rabe as suppression of reorganisation rather than policing, and the resumption of confrontations in August and September 2013 in the provinces. Her bottom line: the evidence pointed to continued armed conflict between the Seleka and the pro-Bozize forces well into 2014, and she would have held that the conflict that began in December 2012 continued after 24 March 2013, with a reconstituted splinter of the Central African armed forces fighting on as the Anti-Balaka at undiminished intensity well into 2014.
Beyond the case, she warns of a protection gap: a finding that humanitarian law switched off during the chaotic months of 2013 and back on in December would open a gap in humanitarian-law protection at a moment when functioning state authority was absent, recalls that the classification of a conflict must not become a revolving door for humanitarian law, switching protection off and on, and states plainly that In her view, neither statute nor case law currently gives clear criteria for when a non-international armed conflict ends. For counsel, that passage is quotable in any future case about conflict classification, including before French courts applying the war-crimes provisions of the Criminal Code.
B. Who Can Appeal What, and With What Realistic Prospects?
Here the opinion holds a surprise that reverses most first readings. After fourteen pages dismantling the majority’s finding of no armed conflict, Judge Flores Liera agrees with the acquittal outcome. She writes: Even if such a conflict existed at the time, she accepts that Said lacked the required knowledge of the facts establishing it, given his place in the Seleka hierarchy under Michel Djotodia, so that his responsibility for the war-crimes counts of torture, cruel treatment and outrages upon personal dignity did not need to be assessed. In other words, even on her own analysis, the war-crimes counts fail against Said personally for want of the knowledge element. The practical consequence is sharp. A Prosecution appeal against the war-crimes acquittals would have to attack both the conflict finding and the knowledge finding, and even winning the first point would not revive counts 3 to 5 against this accused without overturning the second. The Prosecution may still appeal the legal characterisation to protect future cases, since a published finding that no armed conflict existed in Bangui in mid-2013 constrains later CAR prosecutions, but as a route to a heavier outcome for Said it is narrow. The Defence’s appeal looks structurally stronger: it can challenge the crimes-against-humanity convictions themselves, the contextual finding of a widespread and systematic attack pursuant to an organisational policy, the co-perpetration analysis built on Said’s control of the OCRB, and ultimately the sentence. The irony is that the separate opinion, written against the majority, helps the Defence on procedure and the Prosecution on history, while changing nothing for Said’s war-crimes acquittal, which all three judging voices support on one ground or another. Anyone calendaring the case should therefore watch two filings before late October 2026, read the Defence grounds for any attack on the organisational-policy finding, and read any Prosecution grounds for whether it seeks to overturn the no-armed-conflict precedent rather than Said’s personal acquittals. Meanwhile the sentencing phase advances on its own track: the Chamber will take submissions and hold a dedicated hearing, with a maximum of 30 years’ imprisonment, life imprisonment available in extreme cases, and possible fines and forfeiture, as the Court’s questions-and-answers document confirms.
II. What Does the Appeal Phase Change for Cases Connected to France?
France follows ICC cases closely because it may have to act on them: victims and witnesses live here, suspects transit through here, and French courts hold their own jurisdiction over the same categories of crimes. Readers tracking France’s duties toward the ICC, from cooperation and funding to domestic prosecution, should understand that an appeal in The Hague freezes none of the French tracks described below. It changes the finality of the verdict, not the existence of the jurisdiction.
A. Can Victims in France Still Claim, and Can Paris Still Prosecute During the Appeal?
On reparations, the appeal changes nothing for victims’ standing. The Court has confirmed that victims of the crimes of conviction may seek reparations whether or not they previously took part in the proceedings, with the procedure to be addressed in due course, and the 32 victims who participated through counsel Sarah Pellet show the mechanism works. A victim of the OCRB detentions living in France should therefore assemble now the file that any forum will demand: proof of identity, any record of detention at the OCRB between April and August 2013, medical certificates for physical and psychological after-effects, and names of fellow detainees or witnesses. An ICC appeal suspends neither the reparations procedure’s preparation nor the victim’s interest in being identified early. On French prosecution, the governing provision is article 689-11 of the Code of Criminal Procedure, which allows French courts to try persons suspected of genocide, other crimes against humanity and war crimes committed abroad: “Hors les cas prévus au sous-titre Ier du titre Ier du livre IV pour l’application de la convention portant statut de la Cour pénale internationale, ouverte à la signature à Rome le 18 juillet 1998, peut être poursuivie et jugée par les juridictions françaises toute personne soupçonnée d’avoir commis à l’étranger l’une des infractions suivantes : 1° Le crime de génocide défini au chapitre Ier du sous-titre Ier du titre Ier du livre II du code pénal ; 2° Les autres crimes contre l’humanité définis au chapitre II du même sous-titre Ier ; 3° Les crimes et les délits de guerre définis aux articles 461-1 à 461-31 du même code.” Three filters apply cumulatively. First, residence: “La personne soupçonnée doit avoir une résidence habituelle sur le territoire français, celle-ci se définissant par un lien de rattachement suffisant avec la France.” assessed on duration of presence, reasons for it, willingness to settle and family, social, material or professional ties. Second, prosecutorial monopoly: “La poursuite de ces crimes ne peut être exercée qu’à la requête du procureur de la République antiterroriste et si aucune juridiction internationale ou nationale ne demande la remise ou l’extradition de la personne.” Third, subsidiarity: prosecution requires this double lock, with the prosecutor verifying the absence of ICC proceedings and of any competing surrender or extradition request. An appeal pending in The Hague means the ICC remains seized, so Paris still defers on the same person and the same facts, but complaints concerning other perpetrators of the same detention system, or new evidence, can and should be filed without waiting for the appeal’s outcome. Time is not a bar: article 7 of the Code of Criminal Procedure provides that “L’action publique des crimes mentionnés aux articles 211-1 à 212-3 du code pénal est imprescriptible.” The general gateway is article 689 of the Code of Criminal Procedure: “Les auteurs ou complices d’infractions commises hors du territoire de la République peuvent être poursuivis et jugés par les juridictions françaises soit lorsque, conformément aux dispositions du livre Ier du code pénal ou d’un autre texte législatif, la loi française est applicable, soit lorsqu’une convention internationale ou un acte pris en application du traité instituant les Communautés européennes donne compétence aux juridictions françaises pour connaître de l’infraction.” The substantive definitions match the ICC counts: article 211-1 of the Criminal Code defines genocide as “le fait, en exécution d’un plan concerté tendant à la destruction totale ou partielle d’un groupe national, ethnique, racial ou religieux, ou d’un groupe déterminé à partir de tout autre critère arbitraire, de commettre ou de faire commettre, à l’encontre de membres de ce groupe, l’un des actes suivants :” and article 212-1 of the Criminal Code punishes with life imprisonment “l’un des actes ci-après commis en exécution d’un plan concerté à l’encontre d’un groupe de population civile dans le cadre d’une attaque généralisée ou systématique :” expressly including “L’emprisonnement ou toute autre forme de privation grave de liberté physique en violation des dispositions fondamentales du droit international” and “La torture.”
Two rulings of the Court of Cassation frame the advice. On double criminality, the Criminal Chamber held on 24 November 2021, appeal no. 21-81.344: “Les crimes contre l’humanité sont définis au chapitre II du sous-titre Ier du code pénal, et nécessairement commis en exécution d’un plan concerté à l’encontre d’un groupe de population civile dans le cadre d’une attaque généralisée ou systématique.” adding that “l’exigence posée par l’article 689-11 du code de procédure pénale, selon laquelle les faits doivent être punis par la législation de l’Etat où ils ont été commis, inclut nécessairement l’existence dans cette législation d’une infraction comportant un élément constitutif relatif à une attaque lancée contre une population civile en exécution d’un plan concerté.” The Court quashed the Paris ruling and declared the French courts without jurisdiction in that case, a reminder that every French complaint must map each offence onto the territorial law’s matching elements. On torture by non-state actors, the Plenary Assembly held on 12 May 2023, appeal no. 22-82.468, that universal jurisdiction under article 689-2 covers only a defined set of torture, “le cas de compétence universelle de l’article 689-2 précité est limité aux tortures imputées à un agent de la fonction publique ou une personne agissant à titre officiel ou à son instigation ou avec son consentement exprès ou tacite.” while extending the notion: “la notion de personne ayant agi à titre officiel, au sens du renvoi fait par le texte national à la Convention précitée, doit être comprise comme visant également une personne agissant pour le compte ou au nom d’une entité non gouvernementale, lorsque celle-ci occupe un territoire et exerce une autorité quasi gouvernementale sur ce territoire.” A Seleka commander running the OCRB in 2013 sits squarely inside that extension, which is why Paris complaints about CAR detention facts remain legally arguable even where the ICC route is congested.
B. What If France Is Asked to Surrender Someone to The Hague While the Appeal Runs?
Cooperation with the Court is a standing statutory duty, appeal or no appeal. Article 627 of the Code of Criminal Procedure provides: “Pour l’application du statut de la Cour pénale internationale signé le 18 juillet 1998, la France participe à la répression des infractions et coopère avec cette juridiction dans les conditions fixées par le présent titre.” for any person prosecuted before or convicted by the ICC for genocide, crimes against humanity or war crimes. Surrender to the ICC, called remise, follows a dedicated judicial track before the Paris courts, separate from classic extradition and from the European Arrest Warrant. Three practical points matter for anyone arrested in France on an ICC warrant during the Said appeal window. First, the Said judgment itself sets the standard the surrender judge will recognise: adversarial testing across 130 hearings, reasoned findings, acquittals where doubt persisted, a published separate opinion and a sentence pronounced only after dedicated submissions, with appeal before different judges. That record answers in advance most fair-trial objections to surrender. Second, French surrender litigation systematically argues proportionality, family and private life, and detention conditions, and the courts can attach conditions to surrender. The European Arrest Warrant case law illustrates the method: on 23 November 2022, appeal no. 22-86.162, the Criminal Chamber recalled that under article 597 of the EU-United Kingdom Trade and Cooperation Agreement “la coopération par le moyen d’un mandat d’arrêt doit être nécessaire et proportionnée compte tenu des droits de la personne recherchée et des intérêts des victimes et, eu égard à la gravité de l’acte, à la peine susceptible d’être infligée et à la possibilité de prendre des mesures moins coercitives que la remise de la personne recherchée, notamment en vue d’éviter des périodes inutilement longues de détention provisoire.” and that under article 604(b) “si la personne qui fait l’objet d’un mandat d’arrêt aux fins de poursuite est résidente de l’Etat d’exécution, sa remise peut être subordonnée à la condition que la personne, après avoir été entendue, soit renvoyée dans l’Etat d’exécution afin d’y purger la peine ou la mesure de sûreté privative de liberté prononcée à son encontre dans l’Etat d’émission.” The ICC regime organises sentence enforcement through designated States rather than automatic return, but the litigation reflex is identical: contest detention immediately, document family and professional roots in France, and coordinate the Paris defence with counsel in The Hague, because statements made in one forum travel to the other. Third, the pending appeal does not suspend the duty to cooperate: warrants remain enforceable, detention continues, and victims’ applications proceed. For families, the message is that the appeal phase is the worst moment to stay passive, whether the goal is to support a surrender, oppose it, or secure victim status before the sentence and reparations crystallise.
Conclusion
The separate opinion transforms the Said case from a closed verdict into an open legal debate. The majority says the Bangui war stopped in March 2013; the dissenting judge says it never stopped, only paused, and warns against a revolving door that would strip civilians of humanitarian-law protection exactly when chaos peaks. Yet she joins the majority on the outcome that matters most to the accused: the war-crimes counts fail for want of proven knowledge, whatever the conflict’s legal status. The appeal window closing around 23 October 2026 will decide whether that debate stays academic or reshapes the judgment, while sentencing and reparations move on their own tracks. For France, the position is settled regardless: cooperate with The Hague where the ICC is seized, prosecute in Paris where the residence, prosecutorial and subsidiarity filters of article 689-11 are met, and never let time bury the file, since these crimes are imprescriptible. Victims, witnesses and suspects connected to France should act within the appeal window, not after it.