Facts
The prosecution appealed a ‘no case to answer’ ruling for ARU, AOC, and BHL, charged with murder and attempted murder after a fatal knife fight in a car park. The deceased was from their own group, and the appeal focused on the proper application of the joint liability principles in Gnango. (Members from the opposing group, including the actual killer, were tried separately on a prosecution case described by the court as ‘straightforward’ (at [22])).
The incident began when ARU, AOC, and BHL's group, some carrying large knives (as seen in videos), encountered another group (Athif's group) near a park. An initial friendly exchange quickly developed into a fistfight between Athif and BHL. AOC took a video of Athif with a bloody mouth and sent it to ARU and BHL via Snapchat. BHL saved the video with the caption ‘next time ur getting Poked up,’ which was understood as slang for being stabbed. There was no evidence of pre-existing hostility between the groups or gang associations. Over the next three hours, AOC sent messages to individuals named ‘Ibti’ and ‘Mohammed Raman’ containing threats such as ‘I'll acc hurt u’ and ‘got ur own boy smoked’. AOC also sent a message to Raman referring to a ‘zk’ (zombie knife). AOC made a Snapchat video of blood (likely from Athif) on the pavement and tried to call Athif. Messages between Ibti and AOC indicated both groups were looking for each other.
In the evening, both groups converged at a pub car park. A violent clash erupted: BHL, armed with a knife, chased an opponent, while Athif, also armed, stabbed ARU. Tragically, Athif then fatally stabbed H, an unarmed member of ARU's group, as H tried to escape.
It was the prosecution's case that ARU, AOC, and BHL had gone to the car park for a pre-arranged knife fight, with the intent to kill or cause serious harm. The prosecution alleged that it was part of the agreement to fight that they would be stabbed by the other side, wherefore when someone (like H) with whom they had gone to the car park was stabbed to death by the other side with murderous intent, they were jointly liable for that person's murder. (ARU was not charged with his own attempted murder because that charge did not satisfy the public interest test). Alternatively, the defendants had participated in a spontaneous joint fight with weapons. In the absence of significant eyewitnesses, the prosecution relied entirely on circumstantial evidence, including CCTV footage, mobile phone records, and text messages.
Submissions were made on behalf of the defendants that there was no case to answer as the evidence neither supported a pre-arranged, nor the alternative of a spontaneous, joint fight. In his ruling on these submissions, the trial judge concluded that, on the evidence, no reasonable jury could find that the defendants were party to a reciprocal agreement to stab and be stabbed at as required by Gnango [2011] UKSC 59; [2012] 1 AC 827 and Seed and others [2024] EWCA Crim 650. He noted the absence of gang associations and pre-existing hostilities, and the limited communication between the defendants and their opponents. The mere carrying of knives was not probative without more of an agreement to engage in a knife fight with others. Whilst the messages sent by AOC could be interpreted as provocative threats, they did not amount to evidence that the defendants were aware that Athif and/or others were armed. He also concluded that there was insufficient evidence to support a spontaneous agreement to stab and be stabbed at, noting that the defendants had run away when confronted with Athif's knife and that BHL's chase of an opponent was short-lived.
The prosecution applied for leave to appeal against this ruling. They argued that the judge's analysis was incomplete or flawed and that a reasonable jury could have inferred there was a reciprocal agreement between Athif and the defendants to stab and be stabbed. In acceding to the submissions of no case to answer, the judge had gone beyond his remit, as he had in effect reached a judgment on the likelihood of the prosecution case being accepted.
Held, granting leave but dismissing the appeal, that the test for granting leave to appeal pursuant to Section 58 of the Criminal Justice Act 2003 was that a seriously arguable case that the judge was wrong had to be established (at [4]). In a circumstantial case where there was substantial agreement as to the underlying facts, the issue was whether a reasonable jury properly directed could draw inferences sufficient to convict the defendants. Where that was the issue, it would be necessary for the prosecution to establish a seriously arguable case that the judge's evaluative exercise was wrong (at [5]).
The way in which the prosecution had put their case was unusual. The court had had to consider the ambit of the decision in Gnango [concerning a shoot-out between two men that had resulted in the death of a passerby]. It was in the interests of justice to consider the appeal on its substantive merits, irrespective of whether the appeal was arguable (at [5]).
The court endorsed a passage from Gnango where Lord Dyson had held that it was important to distinguish between combat which is analogous to a duel and a mere fight, and that an essential element of the former was an agreement by the combatants to fight each other (at [37]). The court noted that the ‘other members of the court who agreed with the decision [in Gnango] had also referred to the requirement of an agreement between the participants to shoot at each other and to the parallel to be drawn with the concept of a duel’ (at [37]). It further emphasised two aspects from Holroyde LJ's judgment in Seed and others: ‘that knowledge of the other party's possession of weapons and their intention to use them had to be “a virtual certainty” for the necessary agreement to be inferred from the circumstances’ and that ‘reciprocity’ had to be demonstrated (at [39]).
The court suggested that reported examples (Gnango; Riley and Robinson [2018] EWCA Crim 1000; Seed and others) indicated that an unusual factual background would be required (at [48]). Gnango concerned two men involved in a continuing feud who agreed to shoot at each other in a residential area. In Riley and Robinson, the circumstances were akin to a duel. In Seed and others, members of a gang did a ‘ride out’ into the territory of an opposing gang, carrying loaded firearms with the intent to fire them to cause at least really serious harm and appreciating that it was virtually certain that the opposing gang was similarly armed and would return fire with the relevant intent. The court concluded that the violence which had erupted [in the present case] sadly was not an unusual event (at [42]; [49]).
In determining whether [the required] agreement, be it made in advance or reached spontaneously, reasonably could be inferred from what those in the car park did and said, the judge had been entitled to consider the absence of evidence of prior enmity on the part of either group or individuals within each group, the fact that no knives were used on either side in the [first] fight, the absence of any evidence that either group knew that the other was in possession of knives until events began to unfold, the initial brandishing of a knife by BHL not being met with an armed response by the other side, and that the response by the defendants’ group to Athif approaching with a knife was to run away. Those matters militated against a finding that the necessary agreement could be inferred by the jury (at [45]-[46]).
The judge had made no error of law or principle. For the prosecution to have succeeded in their appeal against the judge's ruling, they would have had to demonstrate that the ruling was not reasonably open to him on the evidence called by the prosecution. The court had no doubt that it was reasonably open to the judge to conclude that the defendants had no case to answer. On the facts, the contrary conclusion would not have been reasonable (at [47]). Any attempt to extend the principles in Gnango to incidents of the kind which occurred [in ARU] should be avoided. As Lord Dyson had said, a mere fight was to be distinguished from the unusual facts giving rise to the criminal liability which arose in Gnango (at [49]).
Commentary
The present case is notable for two reasons: it tries to clarify the boundaries of Gnango-style complicity and clearly warns against applying its principles to typical knife crime situations, especially when there is no strong evidence of a shared purpose or agreement to ‘stab and be stabbed’. While the court deserves credit for reining in overly creative prosecutions, it falls short by not adequately addressing the principles of assistance and encouragement. Instead, it overemphasizes agreement and reciprocity, which are not essential for complicity after Jogee [2016] UKSC 8.
The court views Gnango as an unusual case involving a combat-type scenario akin to a duel which makes it distinguishable from the more common fact pattern exemplified by ARU. In Gnango, two armed men (Gnango and ‘Bandana Man’) engaged in a shoot-out, resulting in the death of an innocent passer-by. Gnango was convicted of murder, even though Bandana Man had fired the fatal shot. The (pre-Jogee) case had been left to the jury on parasitic accessory liability principles but, as the court notes, was ultimately upheld on the different (and presumably Jogee-compatible) basis that Gnango and Bandana Man had encouraged each other to engage in a shoot-out whilst sharing a common intention to shoot and to be shot at. In the present case, the Court of Appeal draws on Lord Dyson's opinion in the Supreme Court which identified ‘an agreement by the combatants to fight each other’ (Gnango at [100]) as the key feature that made Gnango complicit in Bandana Man's actions and justified restoring Gnango's murder conviction after it had been quashed by the Court of Appeal.
The court relies heavily on the idea of a ‘duel-like agreement’ from Gnango (as endorsed by subsequent cases, most recently Seed and others). However, it finds insufficient evidence of such an agreement on the facts, emphasising the high threshold for inferring it: ‘virtual certainty’ that the other party was armed and intended to use weapons, along with demonstrated reciprocity. This seems less about legal principle and more about policy. In a climate of rampant knife crime, it is (unfortunately) almost guaranteed that individuals within groups seeking confrontation are armed and intent on using their weapons (if it becomes necessary or opportune). As the court acknowledges: ‘Groups of young males in towns and cities have a tendency to drift around the area where they live. Some will carry knives. When two such groups meet, violence can break out for no consequential reason’ (at [49]). Against this backdrop, it would not be implausible to consider the element of reciprocity fulfilled, especially where, as in ARU, provocative text messages were exchanged ahead of the confrontation by people associated with either group.
Yet, the court strongly rejects any inference of a reciprocal agreement for a knife fight, stating that while provocative messages were exchanged, they did not evidence an agreement to ‘stab and be stabbed’. Crucially, the court stresses that neither group knew the other was armed until the violence began. This contrasts with previous decisions that have downplayed the importance of weapon knowledge, focusing instead on the intent to cause grievous bodily harm. As the court said in Tas [2018] EWCA Crim 2603 at [37], the focus on intention in Jogee reduces ‘the significance of knowledge of the weapon so that it impacts as evidence (albeit very important if not potentially irresistible) going to proof of intention, rather than being a pre-requisite of liability for murder’ (see also Harper [2019] EWCA Crim 343 at [30]). The requisite intent can form spontaneously, tacitly, and conditionally, as Jogee established. The analysis in ARU is thus not entirely convincing.
Nevertheless, the court is clearly and rightly concerned that applying Gnango principles to common cases of group violence would unfairly broaden the scope of murder. The court went to great lengths to distinguish ARU from Gnango, which itself was likely policy-driven to hold someone accountable for an innocent death when the actual shooter was unknown (an issue that did not arise in the present case as the killer was convicted in a separate trial).
The court insists that Gnango-style liability requires ‘unusual factual backgrounds,’ often involving guns rather than knives. It states that ‘[i]t would require very clear evidence of an agreement to stab and be stabbed for the participants in the violence to be fixed with liability for a fatal outcome… Any attempt to extend the principles in Gnango to incidents of the kind which occurred [in ARU] should be avoided’ (at [49]). While this statement clashes with orthodox legal principles (as a reciprocal agreement is not strictly necessary for complicity), it is desirable to curb Gnango's increasing use by prosecutors. Routinely applying Gnango to (spontaneous) group violence would undermine Jogee's efforts to narrow accessory liability for murder, while it is doubtful that Gnango-style complicity could serve as a deterrent against knife crime. Its principles are neither well-known nor easily understood.
However, the focus on ‘agreement’ should not be allowed to give the idea of joint enterprise liability a second lease of life. Jogee clarifies that while an agreement can indicate encouragement or assistance, it is encouragement or assistance that must be proven, not agreement. This distinction is vital, especially given the flexibility in defining ‘agreement’. When relying on agreement as evidence of complicity, the analysis should explicitly connect to the principles of assistance and encouragement (or, as the case may be, joint perpetration). That these are (largely) absent from the discussion in ARU is a regrettable omission.
