Hollywood is currently having a justifiable meltdown over the emergence of Tilly Norwood, an AI-generated “actress” whose creator, Dutch actor and comedian Eline Van der Velden, recently claimed that talent agents are showing interest in signing the digital persona. The Screen Actors Guild (SAG-AFTRA) fired off a blistering response: creativity must stay human-centered, they insist, and they reject the idea of replacing human performers with synthetic stand-ins.

“Let’s be clear,” SAG-AFTRA’s statement reads, “Tilly Norwood is not an actor — she’s a character generated by a computer program trained on the work of countless professional performers — without consent or compensation. She has no life experience, no emotion, and, so far as we can tell, audiences don’t want to watch pure computer-generated content disconnected from the human experience. This doesn’t solve any problem — it creates the problem of stolen performances pushing human actors out of work, devaluing the craft.”

They also warned that producers who sign on to use synthetic performers must comply with existing contracts — that is, unions still demand notice and bargaining when synthetics are involved (the fact that there is a term for this now is a tell for how far gone the situation has already become).

Van der Velden’s defense is a familiar one: Tilly is “a creative work, not a replacement for a human.” She invokes the lineage of animation, puppetry, CGI — tools that expanded storytelling without displacing actors. “I’m an actor myself,” she says, “and nothing — certainly no AI character — can take the craft or joy from human performance.” However, Van der Velden has also reportedly said she wanted Norwood to “become the next Scarlett Johansson”. Obviously, both things cannot be true at once.

The pushback from real actors has been emphatic. Emily Blunt, when shown an image of Tilly, reacted with unease: “That is really, really scary… Come on, agencies — don’t do that. Please stop taking away our human connection.” Whoopi Goldberg chimed in: “We move differently, our faces move differently, our bodies move differently — you can always tell them from us.”

---

Why This Isn’t New — And Why It Almost Always Fails

Parallel to the film industry, the music and marketing worlds also experimented with virtual personalities.

• Kyoko Date (1996): Created by the Japanese talent agency Horipro, Kyoko Date appears to have been first virtual idol. She was presented as a 19-year-old pop star, releasing a music single and having a detailed backstory. She was managed and marketed as a real human performer would have been to her target audience, but Horipro was always open about the fact that she was an artificial construct. She lasted only until generally available technology surpassed what they were doing with Kyoko.
• Ananova (1999): Developed by the Press Association in the UK, Ananova was a “virtual newsreader.” She was designed to read news headlines on the internet and mobile devices. Nowadays she wouldn’t pass the uncanny valley test, and she too was never presented as anything but a computer generated publicity stunt, really. She had her subscribers, and a few news services used her, but as technology advanced, her look aged like milk.
• Lil Miquela (2016): Miquela Sousa, or “Lil Miquela,” represents a significant milestone. Created by the company Brud, she was launched on Instagram and initially presented as a real person. Her creators kept the illusion going for years, having her engage in “feuds” with other virtual influencers and collaborate with real-world brands and musicians, successfully blurring the line between a real and a scripted personality for a massive audience. She traded on her novelty, finally started to wane in signficance and influence around 2022 when people figured out she wasn’t worth paying attention to because she wasn’t a real person.

In every case where “synthetic celebrities” tried to pass as humans, they don’t last more than a few years. Like the computers they’re generated on, they tend to become obsolete. The emergent pattern is: people accept stylized artifice; they don’t accept AI masquerading as flesh.

Tilly Norwood is more brazen: she’s not billed as “digital star,” she’s being pitched to agents as a real actor. And that shift into “realism by deception” is probably what has triggered such fierce resistance from SAG-AFTRA and the acting community at large.

---

A Future Where AI Augments, Not Replaces

To me, the way forward isn’t to pretend that AI can fully be a human. Use synthetic actors in strictly delineated roles — as avatars, stand-ins, stunt doubles, visual effects elements — but not as replacements for human lead performances. Let AI handle the mechanical, the background, the crowd, the interpolations. Productions have been doing this for decades already. The long shots of crowds you see filling the stands at stadiums, those are usually fake now. So are the parking lots packed with cars in the helicopter shots. I know this because it’s actually illegal to fly over a stadium or its parking lot when there’s an event, so capturing this shot for real is impossible.

Keep the emotional core human.

But to attempt full deception — a phantom celebrity signed to agencies, bidding for real roles — is to walk into a trap. Not because the technology isn’t astonishing, but because the moral, legal, creative, and audience resistance is rooted in our belief (rightly so) that stories are made to be lived, not simulated. What makes a performance engaging is what the flesh and blood actor or actress brings to it. An intentional work of fiction is one thing. That’s what entertainment in all its forms are about—but when the public learns that the thing they were led to believe was human actually has no soul behind its eyes, the social contract is broken. This, above all else, is what will doom the avatar known as Tilly Norwood.