Published June 24, 2024 · By CampaignsLive · Case Study
In June 2024, Toys R Us premiered a roughly sixty-six-second brand short titled “The Origin Story of Toys ‘R’ Us” at the Cannes Lions festival. The film was produced by the brand’s parent company WHP Global, in partnership with the agency Native Foreign, using OpenAI’s Sora as the primary production tool. The film tells the founding story of Charles Lazarus — the brand’s late founder — and his original 1948 vision of a children’s furniture store that became Toys R Us.
The reception was mixed. The critical response in design and film communities was unfavorable: the work was variously called soulless, uncanny, or insufficient. The general-audience response was more positive but less engaged. The trade press read the case as either a milestone or a misstep, depending on which conclusion the editor had decided to underline.
The reception, in retrospect, was not the most interesting part of the case. The strategic decisions that produced the film — by both Toys R Us and OpenAI — are the part worth studying, because they set the pattern for the brand-Sora collaborations that followed in late 2024 and into 2025.
What the production actually was
The film was made over a period of weeks by a small team, working inside Sora’s then-private preview. The text-to-video tool produced the visual material; voiceover, scoring, and editorial assembly were done traditionally. The team’s working process — reported in detail by the agency and the brand afterward — involved generating large volumes of clip material, selecting the strongest pieces, and assembling them into the narrative sequence.
The production economics were genuinely different from a traditional brand film. A live-action production of similar narrative scope, depicting 1948 New York and a young Charles Lazarus across multiple scenes, would have required period set construction, period costume, casting, and a multi-day shoot. The Sora version did not require any of those things. The film cost, by reported accounts, a small fraction of what an equivalent live-action production would have cost.
The trade-off was creative latitude. The film could not depict things the model could not produce reliably. Specific period architectural detail, specific Lazarus’s likeness from photographic reference, specific narrative beats requiring precise blocking — these were either avoided in the script or produced in compromise versions that the team accepted as the price of using the tool.
What it did well
Two things, by my read.
The first was tone. The film holds a consistent emotional register throughout its sixty seconds. The pacing is patient. The voiceover sits inside the visual rather than narrating across it. The musical score does its job. The aggregate effect — for a viewer not specifically attending to the visual seams — is a brand film that works as a brand film.
The second was the strategic move it represented. Toys R Us, as a brand, has been working through a long restructuring after its 2017 bankruptcy. The brand needs to remind audiences that it exists, that it has heritage, and that it is investing in its future. A film that says “we are doing new things with new tools, in service of telling our story” achieves the strategic objective regardless of whether the film itself is critically successful. The film’s purpose was, in part, the announcement of the film, and the announcement worked.
What it did less well
The seams the trade press identified were real. The faces drifted. The hands were imprecise. The period detail was approximate where it needed to be specific. The motion was the soft, almost-still motion that the 2024 generation of generative video tools produced as their default register, and that the cultural conversation had begun to identify as a marker of AI-derived film.
The cumulative effect, for a viewer attending closely, was that the film looked like AI. This was not catastrophic — the brand’s strategic objective was achieved either way — but it was a real limitation that the production team did not, or could not, fully address.
The deeper limitation was structural. The film could not have been a different film. Within the model’s capability envelope, the narrative arc the team produced was approximately the best one available. A different brand wanting to tell a different story would have run into the same envelope from a different direction. The tool was not a neutral production instrument; it was a capability with shape, and the shape constrained the creative outcome.
What this set up
Three patterns visible from the Toys R Us case persisted into the brand-Sora collaborations that followed.
Narrative-light, atmosphere-heavy. The most successful brand films produced with Sora in late 2024 and 2025 leaned into the tool’s atmospheric strengths and avoided narratives that required precise period, place, or talent detail. The films were more poem than story.
Voiceover-anchored. The films relied on voiceover and music to carry the narrative weight that the visuals could not. The visual material was the texture; the audio was the spine.
Disclosed prominently. The brands that fared best with the format were the ones that made the AI production a stated feature of the film, in the way Heinz had made the AI process a stated feature of “Draw Ketchup” in 2022. The films that tried to land as standard brand work — without acknowledging the AI provenance — fared worse than the ones that named the choice.
Toys R Us did all three. The films that followed did some or all of the three, with results that scaled to the degree they followed the pattern.
The longer question
What the Toys R Us case did not settle was the longer question: when does generative video stop being a stated production choice and start being a transparent production tool? The film could not have hidden its provenance even if it had tried; the visible markers were too strong. By late 2025 the question was beginning to surface, with the next generation of video tools producing work that was approaching but had not quite reached the point of indistinguishability.
When that line is crossed, the brand calculation shifts. The film no longer announces itself as an AI film. The strategic move becomes harder to read, the production cost case becomes more consequential, and the disclosure question becomes structurally similar to the one image generation has been working through since 2023.
For the broader history of where video tooling sat in the period before Sora, see Why Generative Video Was Almost-There for Two Years. For the related question of AI brand films that hide their provenance, see AI Creative vs. AI Slop.