Olivia Moore, an investment partner at Andreessen Horowitz, created an AI-generated influencer named Janie who participated in the University of Alabama sorority recruitment process, known as RushTok. The experiment tested how consumers react to synthetic media by producing a week's worth of TikTok videos for a character who did not exist. Moore spent $100 in credits to generate the content, which included Janie's dorm room, outfits, and voice. The first video, posted on Aug. 8, gained over 120,000 views and received comments describing the freshman as adorable. Janie's account, @jani.b086, presented a curly-haired, bespectacled redhead listing clothing brands like Abercrombie and J. Crew in a standard outfit-of-the-day format. However, Janie does not attend the University of Alabama, did not rush a sorority, and is entirely a product of artificial intelligence. Moore used ChatGPT and the MiniMax H3 video model to animate the character from a still image of a real dorm room. She chose RushTok because it has high traffic and a low barrier to entry for new creators. Moore noted that Janie would not be selling products or causing direct harm to anyone. Reactions from viewers were split based on age. Moore observed that viewers under 25 quickly identified Janie as AI, while older members of the audience were fooled and sent well-wishes. By the second day of the experiment, the audience had largely detected the deception. Users on Reddit and TikTok created commentary videos and corrected others in the comment section. Despite knowing Janie was fake, some viewers continued to watch, with comments stating they did not care because the character seemed sweet and they wanted her to succeed. Moore adjusted her strategy by dropping hints in the videos, which digital sleuths then identified in the comments. She concluded that humans are currently better at detecting AI than AI is at becoming indistinguishable from humans. Moore stated that the curve for human detection is rising faster than the curve for AI quality. TikTok initially flagged only 8 of Moore's 20 uploaded videos as containing AI-generated media. As of publication, TikTok displays an AI warning on 16 of the 20 videos. Moore did not label the videos as AI when uploading them. She ended the experiment with a video explaining her identity and motives to the audience. Moore was surprised by the positive response to her disclosure, suggesting that consumer sentiment toward AI can shift if users feel they are not being tricked. She believes that honest disclosure and entertainment value can lead to positive engagement with AI content. Moore predicts that more consumers will use AI tools to create human-passing content, but it must compete with human-made media for emotional investment. The Alabama Panhellenic Association declined to comment on the experiment.