Netflix’s latest Willy Wonka offshoot is arriving with a familiar kind of wonder—only this time, the magic trick is happening in the audio track.
A new teaser trailer for Wonka’s The Golden Ticket, which Netflix says will premiere on September 23rd, confirms that the show’s voiceover uses an AI-generated recreation of Gene Wilder’s voice. The production is built around real sets and practical-looking visuals, but the narration layer is synthetic, and the decision raises a question that’s becoming harder to ignore across entertainment: when a studio can convincingly recreate a performer’s sound, what does “consent” actually mean, and how should audiences interpret it?
According to reporting from Deadline, Netflix worked with ElevenLabs, an AI audio company, to generate the voice. Deadline also reports that the project was done with consent from Wilder’s family. That detail matters, because it distinguishes this effort from the most controversial uses of AI voice technology—those that rely on impersonation without permission, or those that blur into deceptive deepfake territory. Netflix has previously pursued similar recreations for other recognizable voices, including Michael Caine and Stan Lee, and this latest move appears to be part of a broader strategy: using AI audio tools to speed up or expand the ways studios can evoke iconic figures, while tying the work to rights holders and licensing agreements.
The teaser itself doesn’t look like a typical “AI content” experiment. The imagery shown in the trailer points to a physical production—factory-like spaces, candy-colored set pieces, and a sense of scale that suggests the show is being made in the usual way. In other words, Netflix isn’t leaning on fake-looking visuals to sell the concept. Instead, it’s using AI where it’s easiest to miss: in the voice. That choice is telling. It implies that the studio believes audiences will accept synthetic audio more readily than synthetic video, at least when the result feels seamless and when the legal and rights framework is in place.
To understand why this is significant, it helps to look at how AI voice tech has evolved. Early consumer versions of voice cloning often sounded uncanny—too smooth, too flat, or too obviously “generated.” But modern systems can capture cadence, tone, and even subtle speech patterns well enough that many viewers won’t notice unless they’re told. That’s exactly what makes these projects feel different from older forms of impersonation. A voice actor can imitate a performance, but AI can reproduce a specific vocal identity with startling fidelity, and it can do so repeatedly, quickly, and at scale.
In the case of Wilder, the stakes are higher than they would be for a generic narrator. Wilder’s voice is not just a sound; it’s part of the cultural memory of a character and a performance style. When Netflix uses an AI-generated version of that voice, it’s not merely adding narration—it’s borrowing a piece of authorship. Even if the family consented, the audience still experiences the voice as Wilder’s, because that’s how the system is designed to function. The ethical question becomes less about whether the voice is “real” and more about whether the audience is being asked to accept a new kind of creative substitution.
Netflix’s approach also fits into a larger pattern: the company has been expanding its Roald Dahl universe through partnerships and adaptations, and it has already shown willingness to treat IP as something that can be reconfigured across formats. The Golden Ticket is positioned as a reality competition, and that matters because reality competitions depend heavily on voice—announcers, judges, narrators, and the emotional pacing of confessionals. If you can generate a voice that carries authority and whimsy, you can shape the tone of the entire show without relying on a single human performance session. That’s not inherently sinister; it’s simply a new production tool. But it changes the economics of voice work and the creative workflow behind the scenes.
There’s another layer here: Netflix is releasing this show right after Squid Game: The Challenge, continuing the trend of turning fictional worlds into live competition formats. Those shows are built on spectacle and tension, and they often use narration to heighten stakes, guide attention, and keep momentum. In that context, an AI-generated Wilder voice functions like a branding asset. It’s a shortcut to a particular mood—playful, theatrical, slightly mischievous—without having to cast someone who can match Wilder’s specific delivery.
But the “unique take” in this story isn’t only that Netflix used AI. It’s that the teaser suggests the company is testing how far it can go while keeping the rest of the production grounded in conventional filmmaking. The visuals appear to be real sets rather than “Glasgow-style” AI fakes—an important distinction because it signals Netflix isn’t trying to normalize AI-generated imagery as the default. Instead, it’s normalizing AI-generated voice as a background layer, something that can be integrated into mainstream production without triggering the same level of skepticism that synthetic video often does.
That difference may reflect both technical and cultural realities. Synthetic video is still easier to detect when it’s wrong, and it tends to create immediate discomfort. Synthetic audio, by contrast, can be more forgiving. If the voice matches the expected rhythm and emotional emphasis, viewers may accept it as narration—even if they wouldn’t have accepted a fully AI-generated character on screen. This is why the voiceover is such a strategic choice. It’s the part of the experience that can be altered while leaving the rest of the production intact.
Deadline’s reporting also notes that Netflix’s Wilder voice recreation is separate from Charlie vs. the Chocolate …, implying that Netflix is managing multiple parallel projects within the Dahl ecosystem. That separation matters because it suggests Netflix isn’t treating AI voice as a one-off stunt. It’s building a repeatable capability—one that can be applied across different productions, potentially across different seasons, and potentially across different narrative needs.
So what does “consent from Wilder’s family” actually mean in practice? In the simplest terms, it means the rights holders agreed to the use of Wilder’s likeness in an AI-generated form. But consent in AI contexts is complicated. Voice cloning can be used for benign purposes—like dubbing, accessibility, or archival recreations—but it can also be used for impersonation. The difference between those outcomes often depends on licensing terms, usage restrictions, and oversight. When a family consents, it may include conditions about how the voice is used, what outputs are allowed, and whether the voice can be used beyond the specific project.
However, even with consent, there’s still a broader question: does the audience have a right to know when a voice is AI-generated? Some viewers will feel that disclosure is essential, not because they oppose AI in principle, but because transparency affects trust. Others may argue that if the voice is licensed and the result is indistinguishable, disclosure becomes less important. Netflix’s teaser doesn’t appear to foreground the AI element; it’s only confirmed through reporting. That means the first wave of audience awareness comes from journalists and industry watchers, not from the show itself.
This is where the conversation shifts from ethics to media literacy. As AI voice becomes more common, audiences will increasingly need cues to interpret what they’re hearing. Studios may eventually standardize disclosures—similar to how streaming services label certain content or how credits list voice actors. But we’re not there yet. For now, the burden falls on press coverage and on the occasional telltale sign that something is “off,” even if the viewer can’t immediately identify what.
There’s also a creative question: what happens to performance when AI can replicate a voice identity? In traditional production, a voice actor brings interpretation—choices about pacing, emphasis, and emotion. With AI voice cloning, the interpretation can be generated from text prompts or scripts, and the system can produce variations quickly. That can be efficient, but it can also flatten nuance if the model isn’t guided carefully. The best AI voice work tends to be the result of iterative direction—human producers shaping the output until it matches the intended performance. In other words, AI doesn’t remove creative labor; it changes where that labor happens. Instead of directing a performer in a booth, producers direct a system through prompts, settings, and post-processing.
If Netflix is using ElevenLabs, it likely involved a pipeline: capturing voice samples, training or configuring a model, generating lines, and then editing to fit the show’s timing and sound mix. That process can still require skilled audio engineers and creative direction. But it also introduces a new kind of risk: if the voice is used incorrectly, it can drift into uncanny territory. And if the voice is used correctly, it can become so convincing that audiences forget it’s synthetic—raising the trust issue again.
Another angle worth considering is how this affects the broader voice acting industry. Voice actors have long dealt with impersonation, but AI voice cloning changes the scale. A single licensed voice model could theoretically be used to generate thousands of lines without additional recording sessions. That threatens a portion of the market, especially for roles that are primarily narration or character-like delivery rather than complex acting. Even if Netflix’s use is licensed and consented, it sets a precedent: major platforms can integrate AI voice into high-profile productions, making it harder for smaller studios to compete without adopting similar tools.
At the same time, there’s a counterargument: AI voice licensing could create new opportunities for performers and their estates. If families can monetize voice models with clear terms, it might offer a new revenue stream. It could also allow performers to participate in projects posthumously in ways that were previously impossible. But that optimistic scenario depends on robust legal frameworks and on ethical standards that prevent misuse. Without those guardrails, the technology can easily outpace the protections.
Netflix’s decision to work with ElevenLabs also signals that the company is comfortable partnering with specialized AI vendors rather than building everything in-house. That’s consistent with how the industry has approached other AI tools—using third-party platforms for speed and capability. It also means the quality of the voice recreation depends partly on the vendor’s