Of the roughly ninety video clips dumped onto GTAForums on 18 September 2022 by the user "teapotuberhacker", one short sequence achieved a circulation far outstripping the rest: a work-in-progress capture of the two protagonists, later named Jason Duval and Lucia Caminos, holding up a roadside diner in what appeared to be the Leonida Keys. Press coverage seized upon it not because it revealed the most about the unannounced game's setting — that role fell to broader montages of Vice City streets — but because it was the first piece of authentic, voice-acted mission content from Grand Theft Auto VI that anyone outside Rockstar had ever seen (MacDonald, 2022). The clip survived takedown waves on Twitter and YouTube long enough to be analysed, frame-counted and compared, and the broad consensus among journalists who watched it was striking: this looked less like a sequel to Grand Theft Auto V and more like a contemporary cousin of Red Dead Redemption 2's scripted hold-ups.
That observation, modest on its surface, has substantial implications for how one reads Rockstar's design intent for the project. The studio has been pulled in two directions for over a decade: towards larger, more reactive, more emergent simulations on one side, and towards tighter, more cinematic, more directorially controlled setpieces on the other. The diner clip — and its eventual public re-emergence, in modified form, in the second official trailer of May 2025 — is one of the clearest pieces of evidence that Rockstar has chosen the RDR2 path rather than the procedural one. This report sets out what the press actually observed in the leaked footage, why the comparison to RDR2 mission design was made so readily, what the setpiece-versus-procedural debate consists of, how Trailer 2 retroactively confirmed the scene as genuine production work, and what the whole episode tells us about Rockstar's continued commitment to bespoke, scripted, "feel-first" cinematic missions.
The body of this report deliberately limits itself to what mainstream press described in print at the time. It does not reproduce, transcribe or describe specifics of leaked frames beyond what reporters at The Guardian, Bloomberg, the BBC and similar outlets publicly committed to the record. Frame-by-frame analysis of stolen footage is both ethically dubious and, more practically, unverifiable; the surrounding press commentary, by contrast, is on the record and citable.
The first wave of coverage on 18 and 19 September 2022 had to do triage. Reporters watched dozens of clips, often through poor mirrors that had survived takedowns, and tried to determine which were authentic and which depicted what (MacDonald, 2022). The diner sequence stood out for several reasons that reporters explicitly noted.
First, it was one of the few clips that contained fully voiced dialogue. Most of the leaked material consisted of animation tests, traversal experiments, or untextured level layouts with debug overlays; the diner clip, by contrast, played out with what reporters described as performed lines from both protagonists and from NPC patrons (MacDonald, 2022; Wikipedia, 2026). The presence of voice work suggested that this particular sequence had progressed beyond block-out and into a recognisably "shippable" form of pre-production — even if textures and lighting were placeholder-grade.
Second, the clip depicted a clearly mission-shaped event, not free roam. There was an objective (the robbery), a posture shift (Lucia or Jason drawing a weapon at a specific cue), a hostage population (the diner's patrons and staff) and a clear before/during/after structure. The Guardian characterised the leaked footage broadly as containing "animation tests, level layouts and gameplay tests, including some fully voiced conversations between characters", and singled out the diner robbery and a strip-club scene as the two locations with discernible scripted activity (MacDonald, 2022).
Third, reporters noted behavioural responses from NPCs — patrons reacting, dropping items, raising hands, complying or attempting to flee. Bloomberg's Jason Schreier, who confirmed the footage's authenticity with sources inside Rockstar, framed the leak as showing genuine mid-development work rather than pre-rendered marketing material; the implication was that the NPC behaviour visible in the clip was the actual in-engine behaviour of the systems Rockstar was building, not a curated cinematic (Schreier, 2022, cited in MacDonald, 2022).
Fourth, the comparison to Red Dead Redemption 2 was made almost immediately and almost universally. Outlets running explanatory pieces on what the leak revealed about GTA VI's mission design — Dot Esports, The Guardian, Kotaku — converged on the same vocabulary: "scripted", "cinematic", "RDR2-style", "contextual" (MacDonald, 2022; Robertson, 2022). Reporters who had spent dozens of hours with Red Dead Redemption 2's robbery missions recognised the tells: the way the player character's stance changed when a weapon was drawn on a person at close range; the way NPCs spoke individually rather than as a generic crowd; the way the criminal pair appeared to coordinate dialogue across the room rather than reading scripted lines at a fixed point.
It is worth being precise about what reporters did not claim. No mainstream outlet asserted that they understood the mission's branching structure, its mechanical depth, or how it would integrate into the wider campaign. The clip was short; it was unfinished; and most journalists explicitly cautioned readers that work-in-progress builds are not representative of the final product (MacDonald, 2022). What they did claim was that the clip showed a mission archetype — the up-close, voice-acted, hostage-laden hold-up — that Rockstar had spent the previous half-decade refining in RDR2, now transplanted into a contemporary GTA setting.
To understand why the comparison to Red Dead Redemption 2 was so immediate and so resonant, one has to look at what RDR2's robbery missions actually do mechanically and tonally — and then at how those properties appeared, in embryonic form, in the leaked diner clip.
Red Dead Redemption 2 (2018) was built around a robbery vocabulary that Grand Theft Auto V (2013) only partially supported. In Arthur Morgan's hands, a stagecoach hold-up, a train robbery, a homestead invasion or a shop hold-up all shared a common interaction grammar. The player held a weapon at a person; that person had a defined "compliance state" that was visible in their animation and audible in their voice lines; the player could issue verbal demands (mapped to a contextual button) that could escalate or de-escalate the situation; bystanders had their own compliance states that propagated through the local NPC population; and the consequences — witnesses, law enforcement response, bounty — were determined by how the encounter was managed, not solely by whether shots were fired. The whole system was wrapped in animation work of a granularity that became one of the game's most-discussed features: hands raised differently depending on who held the gun; eyes tracked the player; dialogue branched at every decision point.
What press observers saw in the leaked diner clip was, by their own description, a recognisable cousin of this system. The reported elements — NPCs reacting individually, dialogue triggering off proximity and weapon state, the criminal pair speaking to each other and to victims rather than reading fixed scripts — are exactly the behaviours one would expect to see if Rockstar had ported the RDR2 robbery framework forward into a contemporary urban setting (MacDonald, 2022; Robertson, 2022). The tonal continuity is not a coincidence of style; it is a continuity of systems. The same animation team, the same dialogue tools, the same mission scripting backbone (RAGE's mission script layer, descended from work that began on GTA IV) all sit under both projects.
The press did not, of course, have access to source code or design documents to verify this. What they had was a few seconds of footage. But the inference was reasonable and was made repeatedly: the diner clip looked like RDR2's Valentine general-store robbery transplanted into a Florida-inspired greasy spoon, with the cowboys swapped out for a millennial criminal couple. The implication was that Rockstar had not abandoned the RDR2 robbery vocabulary on the way to GTA VI; they had brought it with them.
There is a further tonal point that several commentators raised. GTA V's heists were spectacular but largely impersonal: the player executed plans, drove getaway vehicles and fired guns, but rarely had to look a frightened person in the eye and negotiate their compliance. RDR2, by contrast, made that eye-to-eye negotiation a central act of the player's moral life with Arthur. The diner clip suggested that GTA VI was reaching for RDR2's intimacy of violence rather than GTA V's spectacle of violence — that the new game's robberies might feel closer to the Bonnie-and-Clyde framing that Schreier had reported years earlier (Schreier, 2022, in Bloomberg coverage cited via Wikipedia, 2026).
For roughly a decade prior to the leak, a recurring conversation in open-world game design has set "setpiece" design against "procedural" or "emergent" design. The shorthand is reductive but useful: a setpiece mission is hand-authored, scripted to play out within tight bounds, and optimised for a specific cinematic feel; a procedural mission is generated or assembled from systems, designed to surprise even its authors, and optimised for variety and replayability. Bethesda's Radiant Quest system, Ubisoft's outpost-and-event templates, and the entire Mount and Blade lineage all sit on the procedural end. Naughty Dog and pre-2018 Rockstar sit firmly on the setpiece end. Studios like Hello Games (with No Man's Sky) and FromSoftware (with the Elden Ring open world) explore hybrids.
By the late 2010s, the conventional industry wisdom — articulated in trade press, GDC talks and developer interviews — was that AAA budgets and player expectations were pushing studios away from purely hand-authored setpieces and towards procedural or systemic content. The arguments were economic (setpieces cost a fortune per minute of play), engagement-driven (modern audiences expect dozens or hundreds of hours of content, not the twenty-hour campaigns of the PS3 era), and creative (systems can generate situations that designers could never have hand-authored).
Against this drift, Rockstar's Red Dead Redemption 2 was a defiant counter-statement. The game's missions were almost entirely hand-authored, with branching dialogue, bespoke animation per character, and a tolerance for "failure on the smallest deviation" that drew significant criticism from players who preferred their open worlds looser (this critique was widely aired in reviews and post-launch discourse). Yet RDR2 sold over fifty million copies and is widely treated as one of the high-water marks of the form. Its success effectively reset the question: maybe AAA setpiece design was not dying after all; maybe it was simply expensive.
The leaked diner clip landed in the middle of this debate. Analysts and reporters reading the footage immediately recognised that Rockstar had chosen a side. The features visible in the clip — bespoke voice lines, specific NPC reactions, contextual animation triggers, hand-crafted level geometry — are expensive in ways that procedural generation tries to avoid. Industry commentary on the leak repeatedly framed the diner sequence as evidence that GTA VI would extend "the RDR2 model" rather than retreat from it (MacDonald, 2022; Robertson, 2022; Wikipedia, 2026, citing multiple analyst pieces).
This matters because of the budget rumours that have since attached themselves to the project. Various outlets have placed GTA VI's development cost in the one-to-two-billion-dollar range, which would make it the most expensive game ever made (Collins and Richardson, 2025; Wikipedia, 2026). A budget that size only makes sense for a setpiece-heavy game. If the game were primarily procedural, the budget would scale with system development rather than with hours of bespoke content; a billion-dollar budget on a procedural game would be incompetent allocation. A billion-dollar budget on a hand-authored, voice-acted, bespoke-animated setpiece machine is, by contrast, a coherent business decision. The diner clip is part of what makes that business decision legible.
The leaked clip remained a piece of contested evidence for two and a half years. Rockstar formally acknowledged the leak as a "network intrusion" on 19 September 2022 but declined to identify which clips depicted real content versus throwaway tests (Take-Two Interactive, 2022, cited in MacDonald, 2022). The studio's official position throughout the intervening period was that the leaked material was unrepresentative, not that it was inauthentic.
Trailer 2, released on 6 May 2025, retroactively settled the question. Among the new footage — which Rockstar described as "composed of cutscenes and gameplay recorded on the PlayStation 5" (Rockstar Games, 2025, cited in Wikipedia, 2026) — were sequences depicting a diner-set criminal encounter with the two protagonists. The geometry, framing and beats were recognisable to viewers who had seen the leaked material, even though the trailer's version was visually polished, lit, scored and obviously cut for marketing impact (Collins and Richardson, 2025).
The press reaction was telling. Outlets reviewing the trailer — the BBC, Eurogamer, Game Informer, GamesRadar — pointed to the diner footage as one of several visible callbacks to the 2022 leak (Collins and Richardson, 2025; Wikipedia, 2026). The implication, drawn by multiple commentators, was straightforward: the leaked clip had not been some throwaway prototype or alternative direction. It had been a genuine production setpiece, in the pipeline, that survived the intervening years of development and made it into the marketing campaign as a key scene. The opening line of Trailer 2, in which Jason describes himself as "just fixing some leaks", was widely read as a deliberate wink at the 2022 incident (Warren, 2025; Wilson, 2025; Wikipedia, 2026).
For the purposes of design analysis, this confirmation matters in two ways. First, it validates the press's 2022 reading: what reporters identified as a scripted, voice-acted, RDR2-style robbery setpiece really was that, and really did survive to ship. Second, it suggests that the production pipeline for these setpieces is long-running and stable enough that scenes drafted three or four years before release can persist through revision into final marketing. That stability is itself a signal about Rockstar's process — setpiece content is treated as a long-lived asset, not as something thrown together late in production.
Reading the diner clip, the press response and the Trailer 2 confirmation together, several inferences about Rockstar's intent for GTA VI become reasonable to draw.
The first is that Rockstar is committing to mission quality density rather than mission quantity. The vocabulary used to describe a single diner robbery — bespoke voice lines, individual NPC reactions, contextual animation per beat — describes a unit of content that takes a small team weeks or months to build. A game built primarily from such units will have fewer missions than a procedurally assembled one, but each will be richer. This is the same trade RDR2 made, and the diner clip is the clearest evidence yet that GTA VI will make it again.
The second inference is that Rockstar continues to bet on what might be called a feel-first mission philosophy. The phrase, used by Rockstar's own designers in past public talks, points to a process in which the desired emotional and tactile feel of a moment — the way it should make a player feel, the way it should make the protagonist look — is established first, and the systems, animations and scripting are then built to deliver that feel. This is the inverse of a systems-first process, in which generic mechanics are built and missions are then designed to exercise them. The hand-authored quality of the leaked diner sequence, with its specific lines and specific staging, is a signature of the feel-first method.
The third inference is that Rockstar is willing to absorb the cost. A feel-first, setpiece-dense game at GTA's scale and modern fidelity is exactly the kind of project that supports the rumoured billion-dollar-plus budget. The leak — and its eventual ratification by Trailer 2 — is consistent with a company that has chosen expensive, hand-authored craft as its competitive moat, on the theory that the resulting game will be uncopyable by any competitor unwilling to absorb equivalent costs.
The fourth inference, more speculative, is about how the protagonist pairing functions mechanically. The leaked clip showed both Jason and Lucia present and active. RDR2's robbery framework largely assumed a single player-controlled actor with NPC partners. GTA V's heists allowed protagonist-switching but rarely staged two of them in the same close-quarters scripted scene. The diner clip implies that GTA VI may be building a robbery system that genuinely treats the pair as a unit — with coordinated dialogue, role-specific positioning, and possibly shared compliance management of the room. This would be a meaningful extension of the RDR2 framework rather than a straight port. The Trailer 2 footage was consistent with this reading but not, on its own, conclusive.
The factual claims in this report — that the diner clip existed in the September 2022 leak; that press observers compared it to RDR2 robberies; that Rockstar acknowledged the leak as a network intrusion; that Trailer 2 in May 2025 contained recognisably related footage — are well sourced and uncontroversial (MacDonald, 2022; Collins and Richardson, 2025; Wikipedia, 2026). Confidence on these points is high.
The interpretive claims — that the diner sequence demonstrates Rockstar's continued commitment to setpiece-driven design over procedural design, and that it confirms an RDR2-derived robbery vocabulary as the spine of GTA VI's mission design — are inferential. They rest on press readings of unfinished footage and on Trailer 2's marketing-curated re-presentation of that footage. Confidence on these points is moderate-to-high: the inferences are consistent with everything Rockstar has shown, said and shipped over the past decade, but they remain inferences. The final game may reveal procedural systems beneath the hand-authored surface that are not yet legible from trailers and leaks.
The most speculative claim — that Rockstar has built a fundamentally new two-protagonist robbery framework that goes beyond RDR2's single-actor model — is genuinely uncertain. The clip and the trailer are consistent with such a framework, but neither demonstrates it. Confidence on this point is low-to-moderate, and it should be treated as a hypothesis to test against post-launch coverage rather than a settled fact.
Collins, R. and Richardson, T. (2025) 'What have we learned from Grand Theft Auto 6's second trailer?', BBC News, 6 May. Available at: https://www.bbc.com/news/articles/c4g2grmrx4po (Accessed: 9 May 2025).
MacDonald, K. (2022) 'Rockstar owner issues takedowns after Grand Theft Auto VI leak', The Guardian, 19 September. Available at: https://www.theguardian.com/games/2022/sep/19/rockstar-owner-issues-takedowns-after-grand-theft-auto-vi-leak (Accessed: 20 September 2022).
Robertson, S. (2022) 'GTA 6 leak reveals return to Vice City', Dot Esports, 18 September. Available at: https://dotesports.com/general/news/gta-6-leaks-reveal-return-to-vice-city (Accessed: 21 September 2022).
Warren, M. (2025) 'Everything we spotted in the GTA 6 Trailer 2', VG247, 6 May. Available at: https://www.vg247.com/ (Accessed: 7 May 2025).
Wikipedia (2026) 'Grand Theft Auto VI', Wikipedia. Available at: https://en.wikipedia.org/wiki/Grand_Theft_Auto_VI (Accessed: 14 May 2026).
Wilson, I. (2025) 'Every GTA 6 location revealed so far', GamesRadar+, 6 May. Available at: https://www.gamesradar.com/games/grand-theft-auto/gta-6-locations/ (Accessed: 7 May 2025).