An AI-made “RushTok” character managed to slip into one of the internet’s loudest college rituals, and plenty of viewers kept watching even after they suspected she was fake. The experiment turned a glossy sorority recruitment trend into a live test of how people react when a polished online personality is actually generated by software.
“Okay, hi everyone. I’m Janie,” says a curly-haired, bespectacled redhead, clasping her hands in front of her as she shyly addresses a camera from her dorm room.
“It’s day one of rush at the University of Alabama,” Janie continues. Then, she lists off where she got each item of clothing she’s wearing: a pleated skirt from Abercrombie, a knit top from J. Crew, socks from Target and sneakers from Nike.
The video looked like the kind of thing that floods TikTok every summer when sorority recruitment takes over the feed. Janie’s “outfit of the day” post pulled in big views fast, and the comments were packed with people calling her adorable and cheering her on like she was a real freshman trying to survive rush.
That was the hook. Janie wasn’t a student, wasn’t rushing, and wasn’t even human. Every part of the account, from the dorm backdrop to the face on screen, was built with artificial intelligence.
Olivia Moore, an investment partner at Andreessen Horowitz, created Janie as a little experiment with a big question attached to it: what happens when audiences realize content is AI-generated? With a relatively small budget in AI credits, she produced a batch of videos and watched people argue over whether the character was real, fake, or somewhere in between.
Moore said the goal was less about making a perfect fake and more about watching the reaction. Some viewers caught on quickly, while others kept treating Janie like a sweet, harmless newcomer they wanted to support.
She said the broader issue is getting more important by the day because AI image, video, and audio tools are moving fast. The line between obviously synthetic and convincingly human keeps shrinking, and that puts more pressure on people to question what they are seeing before they share it, trust it, or get attached to it.
To build Janie, Moore leaned on tools like ChatGPT and MiniMax. A real dorm room image was used as the base, then AI helped generate Janie’s frizzy red hair, glasses, retro vibe, and the kind of personality that could fit neatly into the RushTok universe.
MiniMax was used to animate the still images into video, while ChatGPT helped shape the prompts and scripts that gave Janie movement and voice. The result was polished enough to pass for a social media newbie at first glance, but strange enough that younger viewers were quicker to sniff out the glitch.
RushTok turned out to be the perfect stage because it already runs on a flood of short, highly similar clips from real people. Moore said the trend gave her a high-traffic environment where Janie could blend in without selling a product or trying to mislead anyone about something serious.
As the posts rolled out, the audience split into two camps. Some users immediately flagged Janie as AI, while others shrugged and kept enjoying the performance, saying it did not matter as long as she was entertaining and harmless.
That response says a lot about where online attention lives now. People are getting better at spotting artificial content, but that does not always kill the vibe, especially when the account is charming, funny, or just weird enough to be fun.
Moore also said the platforms themselves are still playing catch-up. Of the videos she posted, only some were flagged right away, and the warnings increased later, which hinted at how messy large-scale detection can be when the content is constantly changing.
That gap matters because AI content is no longer some far-off concept. It is already sliding into the same feeds as real people, real stories, and real emotions, and the audience is being forced to decide what counts as a dealbreaker and what counts as just another scroll.
When Moore eventually revealed that she had created Janie, the reaction surprised her again. Instead of backlash alone, the post drew a wave of positive comments from people who seemed more curious than angry, especially when the experiment was framed as transparent and playful rather than deceptive.
That may be the real twist. A fake influencer can still win attention if the audience feels invited in, and that leaves social media culture in a strange spot where authenticity matters, but entertainment often wins the argument in the moment.
