Tilly Norwood never existed. That did not stop her from becoming the newest darling on TikTok and Instagram. She quipped from cafés, she pouted in parks, and she gathered followers by the hundred thousand.
On May 6 this year, Tilly made her Instagram debut as an "aspiring actress" with the kind of lifestyle content influencers love and their followers lap up. "Hey besties," she cooed. The captions mimicked real influencers, and many believed the act.
Then came the reveal: she was not a clever new influencer with a gift for snappy lines. She was a puppet of algorithms. Because AI image creation has reached levels of near-photographic realism, it was easy to pull the pixels over people’s eyes.
The collapse occurred in days, but not before Tilly had "landed" her first acting role: a short comedy sketch called AI Commissioner, lampooning the future of TV development. It seemed tongue-in-cheek, a wink at her own synthetic identity. A disclaimer at the start reads: "The video you’re about to view is 100% AI-generated.
So far, so AI.
Attention grew, brands began circling, and her creators, AI production company Particle6 and talent studio Xicoia, smelt the possibility of a gold mine.
The real turning point came last weekend, when producer Eline Van der Velden, the Dutch comedian and founder of Particle6, formally introduced Tilly at the Zurich Film Festival — as an AI creation.
She claimed talent agencies were interested in signing her. Tilly, she said, could become "the next Scarlett Johansson or Natalie Portman".
You can guess how that went down.
Within 24 hours, the backlash had erupted. Over the next few days, Melissa Barrera, Emily Blunt, Whoopi Goldberg and Mara Wilson lined up to denounce the project.
The Hollywood actors’ union, Sag-Aftra (Screen Actors Guild — American Federation of Television and Radio Artists), still scarred by its long strike over AI in 2023, weighed in: "Tilly Norwood is not an actor — she’s a character generated by a computer programme trained on the work of countless professional performers — without permission or compensation."
By Tuesday, Van der Velden was "walking back" her comments, as politicians say.
"She is not a replacement for a human being, but a creative work — a piece of art," she argued. "AI offers another way to imagine and build stories."
But the internet had already decided: Tilly was a fraud.
For businesses tempted to roll out their own AI faces, the lesson is blunt: popularity does not equal trust. Norwood’s rise and fall showed how fast the tide turns once authenticity is questioned.
But AI personas aren’t doomed; Japan has cultivated virtual idols for more than a decade. Hatsune Miku, a purely digital pop star, fills stadiums with fans who know exactly what she is: a hologram backed by software.
The audience buys into the spectacle with full knowledge of the artifice.
That is the crucial point for companies considering similar experiments. People do not necessarily reject artificiality. They reject being lied to. A chatbot that admits it is a chatbot will still get the job done if it solves a customer’s problem. A synthetic assistant that pretends to be human will alienate customers in record time.
South African companies should pay close attention. Our consumers are ruthless calculators. Discovery Vitality keeps members hooked because the points translate into cheaper gym sessions and flight discounts. Shoprite’s Xtra Savings works because it slices a few rand off the grocery bill. That is loyalty.
If either tried to slip in a fake face to front their programmes, the public would shred them in record time. Consumers in South Africa don’t cling to brands out of sentiment; they measure what they get back.
Many executives still believe consumers will accept AI for its own sake. They imagine that a digital assistant will thrill users just because it is powered by machine learning. It won’t
The question consumers ask is brutally simple: does this save me money, time or effort? If the answer is yes, they engage. If the answer is no, they move on.
The regulatory backdrop adds more weight. The Protection of Personal Information Act already puts businesses on a tight leash regarding the handling of personal data. Imagine the scandal if an AI "influencer" were exposed for harvesting user behaviour without consent?
The fine would hurt, but the reputational wreckage would take far longer to clean up.
• Goldstuck is CEO of World Wide Worx, editor-in-chief of Gadget.co.za , and author of ‘The Hitchhiker’s Guide to AI – The African Edge’
Would you like to comment on this article?
Sign up (it's quick and free) or sign in now.
Please read our Comment Policy before commenting.