As deepfakes make their way into broadcast-quality productions, filmmakers and artists are facing new questions like how to cast the right actors for a deepfake
In 2019, two multimedia artists, Francesca Panetta and Halsey Burgund, set about to pursue a provocative idea. Tweets: @_karenhao Tweets: Karen Hao / @_karenhao : Now that deepfakes are being used in professional productions, directors and actors are facing new questions: how do you cast the right actor to deepfake? And how do you *act* when your face will never be seen?? I had loads of fun reporting this one. https://www.technologyreview.com/ ...
Context & Ripple Effects
The technology behind this story got cheap fast: one author documented building a convincing deepfake in two weeks with a $552 Faceswap setup, and hobbyists now publish YouTube tutorials teaching others the same techniques. What began as meme material has been climbing the production ladder ever since.
Francesca Panetta and Halsey Burgund's project marks the next rung: deepfakes reaching broadcast quality, where Karen Hao reports the craft questions flip — you no longer cast a face, you cast a voice and body to be overwritten. That puts artists alongside corporate adopters like Synthesia, whose multilingual training-video business already treats synthetic faces as a product line.
First-order effects
- Directors working on deepfake productions must recast their criteria: screen presence gives way to vocal performance and physicality, since the actor's face never appears — a workflow shift Panetta and Burgund are navigating right now.
- Actors hired for these roles are being paid to perform characters they will never visually embody, changing what a performance contract even means.
Second-order effects
- A market for licensed likenesses is consolidating around this demand: Hour One is already paying people for their faces to build AI-voiced characters, and broadcast-grade work pushes more talent toward negotiating face-rights as a distinct asset.
- Studios adopting the technique will need provenance and consent practices, because the same tools documented in DIY tutorials mean audiences can't assume any on-screen face is real.
Third-order effects
- If broadcast deepfakes normalize, likeness rights become the core IP fight — a dynamic already visible in adult content, where stigma leaves performers unable to protect their own images, per The Walrus's reporting.
- The industry drifts toward a trust infrastructure for synthetic media: verification layers and consent standards deciding which synthetic faces audiences are allowed to believe.
The trend: Synthetic faces are moving up the production stack — from $552 hobby projects to corporate video to broadcast film — turning likeness rights and consent into the defining labor question of AI-era filmmaking.