A crying “Black businesswoman” named Aliyah begs TikTok viewers to watch 13 seconds longer so she can save her handmade belt buckle business. She isn’t real. Neither are the buckles, according to a new investigation from The Verge AI, which traced her and dozens of similar characters to a fast-growing scam: AI-generated influencers built to sell mass-produced dropship junk you can buy on Shein for a quarter of the price.
This is one of the clearest signs yet that generative video has crossed a threshold. The tech is now good enough, and cheap enough, to manufacture fake human beings at scale and point them straight at your wallet.
What’s actually happening
The Verge AI found dozens of accounts across TikTok, Instagram, and Facebook running nearly identical scripts. A marginalized small-business owner struggles to make a sale, pleads for support, and links out to a Shopify or Shein storefront. The products rotate: belt buckles, cowboy-boot mugs, crochet bags, cardigans. The characters rotate too. Almost every layer is automated, including the replies to comments, some of which mimic African American vernacular.
The scale is the story. Jeremy Carrasco, who runs AI video detection outfit Riddance.ai, told The Verge his team finds up to 100 of these accounts every single day. “It’s massive,” he said. Most aren’t coordinated. A single operator can run one AI actor across a whole web of shops.
Aliyah’s account alone pulled 40,000 followers. Her top video racked up 6.5 million views and 814,000 likes.
Why empathy is the attack vector
Carrasco calls the tactic “empathy bait.” Find a dropship item that resonates with a specific community, then build a personality designed to trigger support. The most-viewed characters The Verge found were Black women, because solidarity drives clicks, comments, and shares, which is exactly what the algorithm rewards.
It works on people who should know better. Gizelle Bryant from The Real Housewives of Potomac admitted on her podcast that she bought two crocheted bags after watching an AI-generated Black boy claim he was bullied for crocheting. “How did I get tricked? Viola Davis was on there, too,” she said.
University of Pennsylvania researcher Cienna Davis frames this as a new form of digital blackface: non-Black operators using AI to mimic Black cultural expression for economic gain. The harm runs deeper than a wasted $15. It launders real solidarity into someone else’s revenue and poisons trust in actual marginalized founders.
India Cater-Campbell, a real Black business owner opening a café in Seattle, commented on Aliyah’s video hoping to help. “I was trying to be supportive to an independent Black businesswoman,” she said. What saved her from buying was mundane: she couldn’t find the store link, got bored, and scrolled away.
How to spot the fakes
The tells are still there if you slow down. The Verge AI flagged a few:
- Voice and face don’t match. Aliyah’s voice is robotic and flat while her face is crying.
- Physics glitches. A wiped tear leaves a stream of liquid that vanishes. Someone “sews” a belt where no sewing belongs.
- Clone armies. Dozens of near-identical videos share the same background, tabletop, and spool of twine under different names.
Why it matters now
The platforms aren’t stopping it. Carrasco’s blunt read: “the platforms don’t really care and people don’t notice.” Detection labeling is inconsistent, only some of these videos carry an AI tag, and enforcement isn’t keeping pace with 100 new accounts a day.
For anyone building or marketing online, this is a warning shot. The same tools spinning up fake founders can clone your brand, your testimonials, and your face. Three practical moves:
- Watermark and verify. If you sell with video, build proof-of-human signals into your storefront, real footage, named suppliers, verifiable reviews.
- Train your audience to check links. Scams die when buyers pause before tapping out. Ironically, friction protects people.
- Watch the regulatory gap. Disclosure rules for AI-generated commercial content are coming. Get ahead of them rather than scrambling later.
The deeper shift here is trust collapsing as a default. When a tearful face on your feed is more likely synthetic than sincere, skepticism stops being cynical and starts being basic hygiene. The full investigation, with side-by-side examples, is worth reading at The Verge AI.