Stallone pitched an AI de aged 'Rambo' prequel | what happened

Sylvester Stallone told press that he pitched a Rambo prequel in which he would play an eighteen year old John Rambo using modern de-aging and AI techniques. Stallone proposed capturing a new performance from himself at his current age, then mapping that performance onto a younger face using AI face replacement tools calibrated to how he looked at eighteen.

Multiple trade publications confirmed the pitch in September 2025. The project is moving forward without Stallone in the lead, with a different cast for the young John Rambo role. Director Jalmari Helander, the Finnish filmmaker behind Sisu, was reported as attached to the new prequel and publicly questioned whether the de-aging technology was ready to carry an entire feature.

What Stallone Was Proposing

The pitch described using current de-aging and face replacement techniques to let Stallone perform the role himself. The approach would involve capturing his present performance on set, then applying AI tools in post to map the facial geometry and skin texture to a reconstructed version of his younger appearance.

This is different from archival AI restoration, which works with existing footage to improve quality. Stallone's concept starts from a new live performance. The de-aging would apply to the face and potentially the voice, with body doubling and stunt performers handling physical action that the performance technology cannot address.

The technical components are established in principle. Face replacement, voice synthesis, and de-aging for contained scenes have appeared in released productions including The Irishman, Indiana Jones and the Dial of Destiny, and several Star Wars productions. The question Helander raised is whether current methods can sustain quality across a full feature rather than specific, carefully controlled scenes.

What the Director Said

Jalmari Helander told press he was "not sure the technology is there yet to pull that off." The statement distinguished between what AI de-aging can do in controlled conditions and what it would require to carry a feature with a protagonist who appears in almost every scene.

A few controlled de-aging shots in a largely live-action film benefit from the full production machinery surrounding them, including careful lighting, limited camera distances, editorial pacing that does not linger on the digital face, and a surrounding performance by the real actor that provides context. None of that applies when the de-aged performance is the primary story vehicle.

Helander's assessment reflects practical VFX experience rather than a philosophical objection to the technology. The director of a studio action film has a production budget and schedule that depends on the technology delivering consistent quality. A bet on AI de-aging that underdelivers partway through production is a budget crisis.

The Technical Challenge Explained

De-aging an actor for a full feature performance requires solving several interlocking problems simultaneously. The AI face replacement must be convincing in close-up for sustained dialogue scenes, not just in quick cuts or medium shots where softening is less visible.

It also needs to stay consistent across changes in lighting, camera angle, and facial expression over dozens of separate shooting days. Drift, where the replacement face subtly changes its proportions or texture across a production, is a known failure mode that can take months of post production to fix.

The voice adds a separate layer. Stallone at his current age does not sound like Stallone at eighteen. A fully coherent performance would require either synthesizing a younger voice from training data or accepting the mismatch between a visually youthful face and an older voice. Neither solution is straightforward at feature quality.

Rights, Consent, and Ethics

The moment you use an actor's younger likeness, you step into a different legal and ethical category than a live performance. Lock explicit consent in writing for face and for voice as two separate items, so there is no ambiguity about archival audio or synthetic speech. These are distinct capabilities, and the rights to each should be negotiated distinctly.

Spell out scope of use, term, geography, and downstream applications including trailers, games, merchandise likenesses, and promotional tie-ins. If you plan to train or fine tune any system on the actor's past work, that requires permission as a distinct activity, not as an implied step in the broader agreement.

Keep a paper trail with prompts, seeds, model versions, and vendor logs so you can reproduce a shot if a note comes in near delivery and so you can answer questions later about what systems were used and how. If estates are involved, route all of the above through the correct legal representatives and keep distribution limits documented explicitly.

What SAG-AFTRA Had to Say

Stallone's Rambo pitch landed during active SAG-AFTRA negotiations over AI and digital likeness provisions. Those negotiations centered on consent requirements, minimum compensation for AI use of a performer's likeness, and what studios could do with a digital representation without separate approval.

The interim and later formal agreement provisions require explicit consent for synthetic performance and likeness use and set compensation minimums that are separate from the base performance rate. They also include provisions about the scope of likeness use and how far an approved use can extend without triggering a new consent requirement.

Stallone's willingness to pitch himself as the subject of the de-aging was unusual in the context of talent pushback against AI likeness use. Most documented cases involved talent or their representatives opposing studio AI use. His pitch represented the opposite. He was proposing AI tools as the path to a role he could not play otherwise.

De-Aging in Current Productions

As of 2025, de-aging has appeared in several major studio productions but consistently in specific, bounded use cases. De Niro, Pesci, and Pacino in The Irishman appeared younger in flashback sequences that ran for defined periods rather than across the full runtime. The Harrison Ford de-aging in Indiana Jones opened the film but gave way to the live performance within minutes.

No full-length feature has cast a de-aged actor as the primary lead for the entire runtime at the quality level required for mainstream theatrical release. The closest precedents involve digital characters with established visual styles that viewers understand as stylized, not photorealistic human faces.

Photorealistic de-aging of a human face for sustained close-up dialogue performance across two hours remains an unproven proposition at studio quality. The Stallone pitch was proposing something that would be a first at its intended scale. Helander's skepticism reflects that absence of precedent.

What Filmmakers Should Do

Treat de-aging as a contained production tool, not a casting strategy. Use it for flashback sequences, opening scenes, and specific moments where the story specifically needs the same actor at a different age. Keep the de-aged segments to a defined portion of the runtime and build the surrounding production to support those sequences with careful lighting and editorial pacing.

For budget and schedule planning, add a full VFX character build to your estimates whenever de-aging is a primary story element. The work includes scans, training, tracking passes, vendor integration, quality control reviews, and likely a second wave of ADR or voice synthesis work. None of this is a filter applied in post.

Add a rights and safety checkpoint before any marketing handoff. Trailer clearance for AI de-aged faces can be a separate negotiation from the production agreement, particularly when the footage will be seen before the production is finished. If you plan public materials with de-aged content, establish the clearance scope explicitly in the original agreement.

De-Aging Methods: AI Era vs. Earlier Approaches

Industrial Light and Magic's work on The Irishman in 2019 used a combination of reference footage scanning, tracked facial geometry, and frame by frame compositing supervised by human artists throughout the process. The workflow required dedicated monitor operators watching live output during production and weeks of manual cleanup in post. It cost tens of millions of dollars.

The AI era equivalent compresses parts of that workflow. Neural face replacement models can apply a tracked younger appearance to a recorded performance in post without the monitor operators and some of the manual frame by frame work. The generation happens computationally rather than through frame by frame human compositing. This reduces cost and time but introduces different failure modes, such as temporal flickering across frames or inconsistent skin texture in varying lighting conditions.

Early pre-AI de-aging, such as practical prosthetics and makeup, had no flickering problem but a different set of constraints. A prosthetic face is physically present in the scene and responds correctly to light because it is literally being lit. The AI replacement face is generated from learned models that may not generalize perfectly to every lighting condition in the shot.

Each approach has conditions where it excels. Practical prosthetics and makeup work best for controlled single camera setups with consistent lighting. Digital frame by frame compositing works for contained sequences where the budget allows thorough human supervision. AI face replacement is fastest for high-volume work where absolute perfection in every frame is less critical than overall plausibility. The Stallone scenario would require the most demanding version of AI de-aging. Sustained, high quality, feature length performance where every close-up will receive full audience attention.

Likeness Rights and the Estate Problem

One specific challenge in the Stallone pitch is the absence of photographic reference for Stallone at eighteen. De-aging that targets an actor's known younger appearance, such as Harrison Ford at forty or Robert De Niro at thirty-five, can be calibrated against archival footage and stills. Stallone at eighteen predates his public profile, which means the training data for that specific appearance would need to be assembled from limited sources.

When the training reference is sparse, the AI system fills in the gaps with averaged features from training data. The result is a face that resembles Stallone at his current age, de-aged, rather than an accurate reconstruction of what he actually looked like at eighteen. Whether that distinction matters to audiences depends on how precisely the story requires a historically accurate young Stallone versus a plausible young version of the character.

For productions working with living actors who can provide reference documentation, the training data problem is solvable through structured photo sessions and video reference that cover the full range of lighting and expression conditions needed for the production. For historical figures or earlier periods of an actor's life with limited documentation, the reconstruction will always involve a degree of inference that audiences may or may not accept depending on the story context.

Why Stallone Proposed This

The core logic of the Stallone pitch is that the Rambo franchise is a specific asset with a specific identity that has high awareness and commercial value. A new young actor playing John Rambo, however talented, starts with zero audience connection to the franchise. The IP recognition is to Stallone as Rambo, not to the character abstracted from the actor.

This is a genuine problem that de-aging AI addresses in a way no other technology does. Casting, prosthetics, and CGI replacement all break the identity continuity in different ways. De-aging lets Stallone himself perform the role, with the AI handling the visual translation to a younger face while his performance, voice, and physical choices remain his own.

From a commercial standpoint, the pitch makes sense if the goal is to extract value from the identity of Stallone as Rambo. The question is whether the AI technology can deliver what the commercial logic requires. The film would need a convincing, sustained young Stallone across a feature length production at a cost the economics justify.

Helander's skepticism reflects a practical judgment that the technology is not yet at the point where that commercial logic and the technical reality align. A film that promises a de-aged Stallone performance and delivers something that looks uncanny or inconsistent has a worse outcome than one that casts a new actor without the de-aging promise.

The situation illustrates the broader dynamic playing out across multiple AI de-aging projects in 2025. The creative and commercial arguments for the technology are clear, and the technology is advancing, but the gap between controlled demos and full-feature delivery has not been closed in a publicly released production at the scale Stallone was proposing.

Where This Leaves the Rambo Franchise

The prequel is proceeding with Noah Centineo cast as the young John Rambo. This is the conventional solution to the problem Stallone's AI pitch was trying to solve. Find a young actor who can credibly play the character at that age, and build the film around their performance. Helander's comments suggest he believed this was the more reliable production strategy given where the technology stands.

Centineo's casting does not make the Stallone pitch irrelevant as a case study. The franchise creative team evaluated the AI approach and chose against it based on current technology readiness. That is a documented decision point that other franchises and studios will reference as they evaluate similar proposals.

The Rambo prequel will be the first major project directed by Helander outside his native Finland after Sisu's international success. His willingness to publicly comment on the technology limitations while the franchise IP holder had proposed using it is unusual. Most directors attached to major franchise projects avoid public disagreement with producers over creative or technical choices before production is complete.

What Helander's comment does is establish a public record of director-level skepticism about full-feature AI de-aging at current technology levels. That record will be referenced against future productions that attempt similar approaches, both as precedent for the skepticism and as context for evaluating whether the technology has advanced past the threshold he identified in September 2025.

The outcome of the Rambo prequel, including its box office performance and how audiences respond to the Centineo casting versus what a Stallone de-aged performance might have been, will become part of the evidence base that studios and directors use to evaluate whether AI de-aging is worth pursuing for their own franchise sequels, prequels, and reboots in subsequent years.

Filmmakers developing their own approach to AI character work can generate footage and test concepts through the AI FILMS Studio video workspace.

The Rambo prequel story landed alongside broader conversations about what AI makes possible and practical in franchise storytelling. As those tools evolve, the Stallone pitch will look either prescient or premature, depending on how quickly the technology catches up with the creative ambition it inspired.

---

Sources

Variety | Deadline | The Hollywood Reporter | Polygon