The late-September 2025 presentation of the virtual “actress” Tilly Norwood at the Zurich Summit sent ripples through the film industry. Created by Dutch actress-producer Eline Van der Velden through her studio Xicoia, a spin-off of Particle6, Tilly Norwood is promoted as the first fully synthetic “talent” potentially eligible to sign with a Hollywood talent agency.
The character already has a significant online presence — her Instagram account, launched in May 2025, had about 36,000 followers by the end of September — and was first introduced to the public in a short sketch entitled AI Commissioner, produced using around ten AI tools and partly scripted with ChatGPT.
The unveiling occurred against a tense industrial backdrop. Following the 2023-2024 labor dispute, the U.S. performers’ union SAG-AFTRA had secured contractual protections against uncontrolled use of “synthetic performers.” News that Tilly might be represented by an agency provoked an immediate backlash: the union denounced the project as “non-human-centered” and accused it of relying on the unconsented use of human performers’ work.
Prominent actors — including Emily Blunt, Whoopi Goldberg, Natasha Lyonne, Melissa Barrera, and Mara Wilson — publicly voiced concerns that such avatars could undermine the acting profession and diminish the value of genuine human emotion on screen. Some agencies, such as Gersh Agency, stated explicitly that they would refuse to represent an AI “actor.”
The controversy echoed a similar development in the music industry: only days earlier, the virtual singer Xania Monet had reportedly signed a major recording contract, signaling that purely synthetic performers are starting to enter mainstream commercial representation.
The first set of concerns arises from the right of publicity, recognized in many U.S. states, notably California and New York. This right protects an individual’s name, image, likeness, voice, and other personal attributes against unauthorized commercial exploitation. Reports that Tilly’s facial model was generated from composite datasets raise the question whether identifiable individuals’ likenesses may have been appropriated without consent.
The second issue concerns copyright law. Training an AI system on images and performances of human actors invites scrutiny under the doctrines of fair use versus unlawful reproduction. Current U.S. jurisprudence and guidance from the U.S. Copyright Office hold that works created without substantial human authorship are not protected by copyright; exploitation of an avatar such as Tilly would therefore rely on contractual rights rather than on autonomous copyright.
Performers’ neighboring rights are also implicated: the reuse of elements derived from an actor’s recorded expressions, gestures, or voice could amount to an unauthorized exploitation of their performance.
Finally, the SAG-AFTRA collective bargaining agreements require advance notice and negotiation whenever synthetic performers are to be used in a production. Non-compliance may expose producers to union grievances and to boycotts. In addition, the Federal Trade Commission’s endorsement and advertising rules may come into play if the public is misled into believing it is dealing with a human performer.
French law does not recognize a stand-alone right of publicity, but it affords strong protection to the right to one’s image and to privacy (Article 9 of the Civil Code), as well as to the neighboring rights of performers (French Intellectual Property Code). The use of recognizable human features, even as part of a composite digital likeness, can give rise to civil liability or be challenged as parasitic commercial conduct.
Training AI models on audiovisual or photographic works is subject to the European Union’s regime on text and data mining under Directive 2019/790: rightholders may opt out of such uses; ignoring an opt-out may result in a claim for copyright infringement.
From 2025 onward, the EU AI Act imposes transparency obligations, including explicit labeling of synthetic and deepfake content. A film producer using a virtual avatar such as Tilly will be required to inform viewers that the “actress” is not a human performer.
In addition, the EU General Data Protection Regulation (GDPR) applies to the use of personal data — including images and voices that may constitute biometric data. Such processing requires an appropriate legal basis and, in many cases, a data-protection impact assessment (DPIA), along with compliance with rules governing international data transfers.
The emergence of Tilly Norwood marks a decisive shift for the creative industries: the rise of synthetic talent must be reconciled with the protection of human performers.
In France and, more broadly, across the European Union, the right to one’s image, performers’ neighboring rights, and the new AI regulatory framework together provide strong tools for ensuring transparency and accountability.
The spread of avatars such as Tilly Norwood and virtual singers like Xania Monet therefore calls for solid contractual and regulatory governance to preserve the value of human creativity, safeguard fair competition, and maintain public trust.
Disclaimer: This analysis is provided for general informational purposes. Any decision or strategy involving the application of U.S. law should be preceded by consultation with a qualified U.S. attorney.
