The “influencer who doesn’t exist”: Social media platforms struggle to detect AI profiles :: WRAL.com

Valeriia Litvinova Cunningham was pursuing her master’s degree at North Carolina State University as the war escalated in Ukraine. He helped his parents and sister Anya start a new life in North Carolina. “They basically came here with just the clothes on their back,” Cunningham said. Once reunited, his family went to the gym. It quickly became an escape route. “We started strength training together as a way to distract ourselves from what was going on and also do something together as a family,” Cunningham said. He started posting the family gym sessions on social media. Anya, who has Down syndrome, quickly became a source of inspiration to her growing community of online followers. “We saw a lot of progress in Anya’s confidence and she started to feel a sense of belonging at the gym.” “We’ve been told our whole lives that Anya couldn’t accomplish anything, that she couldn’t even walk. Now she’s strength training; she’s in the gym with other bodybuilders, with powerlifters and just training them,” Cunningham said. The sisters’ online community grew into exercise classes for people with Down syndrome through GiGi’s Playhouse, a nonprofit organization that provides resources to people of all ages with Down syndrome. They call it Strength Without Labels. “We have truly demonstrated, and we hope to continue to demonstrate, every day that her diagnosis does not define her,” Cunningham said. “Our participants can see that they belong in a gym with other people. They don’t need to be separated into a separate group. They can train together with everyone else, and that’s what’s important.” As his followers grew, so did his skeptics. “People are accusing Anya [of] not be real, because nowadays, apparently, we have to fight with artificial intelligence, and that’s exactly why people struggle to distinguish between what’s real and what’s not.” WRAL Investigates met Anya in person during an exercise class. However, we also found an account, @MiaTriso21, on both TikTok and Instagram that some people online believe modeled her face after Anya to put on other creators’ bodies. “To be honest, I can’t even believe she’s having this conversation right now, especially because… we share our journey on social media and fight all these stereotypes. Now I’m fighting against AI-generated content or fabricated content,” Cunningham told WRAL Investigates. “These accounts borrow the body, voice and audience trust of a real creator, put an AI face on top and then monetize that trust,” said Laeeq Khan, director of the social media analytics research lab at Ohio University. Khan says social media platforms tend to rely on victim reporting. “If someone hasn’t reported it, they don’t catch it,” he said. Sometimes reporting it “Sometimes I get a response saying something AI-generated isn’t, so the reports claim ‘we didn’t find that this content was AI-generated,’ and then I have to defend myself. “I can’t believe I’m spending so much time trying to fight a robot,” Cunningham said. Both TikTok and Meta require AI-generated content to be tagged and have policies for removing videos that violate their respective community guidelines. However, WRAL Investigates found several videos still posted without proper tags. We asked TikTok and Meta what they’re doing to find [AI-generated] profiles/content that may go unnoticed. Meta did not respond to our repeated inquiries. TikTok said it uses C2PA Content Credentials technology, which tracks the origin and editing history of digital content. The company also said it is testing detection systems to target accounts dedicated to posting AI-generated spam. Since WRAL Investigates began asking questions about @MiaTriso21, the account has been removed from both TikTok and Instagram. However, other versions of the same account already exist. If something seems perfect, Khan warns, be skeptical. And if it seems too good to be true, it probably is.