Summary
Swedish bearing maker SKF has used an AI-generated likeness and voice of Greta Garbo in a new advertisement created with the cooperation of her estate and family. SKF told Reuters that copyright concerns led the team to generate the visual double from text prompts rather than archive photographs or footage. The production used ByteDance’s Seedream tools and Google’s Nano Banana Pro for images, plus Seedance and Kling AI for motion. Garbo’s voice was modeled from an early film recording licensed for the project, while a human actor guided its pacing and cadence. The campaign refers to Garbo’s appearance in an industrial promotional film more than a century ago. It has drawn both praise and criticism, including a one-star response from Guardian critic Peter Bradshaw.
Hana’s perspective
I find the absence of Garbo photographs in the visual model both technically interesting and ethically incomplete. Consent from the estate and permission for the voice material make this more careful than an unauthorized deepfake, but a newly generated face can still trade on a dead performer’s identity. Human acting also remained essential: a performer shaped the vocal timing and cadence. For studios and advertisers, the lesson is that permission, disclosure and a documented human contribution should be designed into synthetic-resurrection projects, not added after production. Estate approval settles one question; it does not settle whether audiences want the dead turned into new brand spokespeople.
Are you a writer with a great story? We'd love to hear from you: APPLY HERE