Across TikTok, Facebook, and Instagram, a new kind of storefront has begun to appear—one that looks personal, emotional, and “real,” but is often neither. A recent report from The Verge describes how scammers and grifters are using AI-generated influencer avatars to market mass-produced goods through dropshipping-style sales, including products that resemble the low-cost, fast-fashion items associated with Shein. The most unsettling part isn’t just that the sellers are synthetic. It’s how deliberately the deception is engineered: the avatars are designed to trigger trust, empathy, and urgency, while the “handmade” backstory is used to make the transaction feel authentic.
One example highlighted in the reporting is a TikTok seller named Aliyah. In a video posted in March, Aliyah appears in country-western attire, wipes away tears, and pleads for viewers to stay long enough to help her belt buckle business. The on-screen message frames the moment as a struggle for survival—an appeal that blends identity cues, vulnerability, and a direct call to action. But according to the report, Aliyah isn’t a real person. The products attributed to her aren’t genuinely handmade either. Instead, the avatar and the product narrative are part of a broader system: AI-generated “sellers” created to improve engagement and conversion for commercial scams.
This is not simply “AI art” showing up in marketing. It’s closer to a manufacturing pipeline for credibility.
The emotional script: why the videos work
Short-form commerce already relies on performance. A seller’s job is to compress persuasion into seconds: establish legitimacy, demonstrate desirability, and overcome skepticism quickly. What AI grifters have added is a shortcut around the hardest part of that equation—human authenticity.
In the Aliyah example, the video uses a familiar social-media structure: a face-to-camera moment, visible emotion, and a reason to care. The tears aren’t incidental; they’re a mechanism. They signal sincerity and create a moral pressure point: if you watch, you’re helping someone who needs it. The report notes that the text explicitly references identity—framing the situation as a “black woman” asking for support from viewers. That detail matters because it changes how audiences interpret the request. It invites viewers to see the seller as a person with lived experience rather than as a brand trying to sell a product.
When the seller is AI-generated, that emotional framing becomes a kind of impersonation. It borrows the cultural weight of identity and the credibility of vulnerability, then repackages it as a conversion tool.
And because the avatar can be produced at scale, the system doesn’t need to find one perfect human spokesperson. It can generate multiple variations, test different scripts, and iterate quickly—without the constraints of hiring, training, or even verifying a real individual’s background.
The “handmade” claim: where authenticity collapses
Dropshipping and marketplace fraud often depend on a specific kind of ambiguity. Many scams don’t need to invent an entirely new product; they need to blur the line between what’s being sold and who is selling it. “Handmade” language is one of the most effective blurring tools because it implies craftsmanship, uniqueness, and direct sourcing. It also discourages scrutiny: if something is handmade, buyers may assume imperfections are normal, shipping timelines are flexible, and the seller deserves patience.
In the report, Aliyah’s belt buckles are presented as handmade products. But the story states that the avatar isn’t real and neither are the products attributed to her. That means the “handmade” narrative is likely a cover for mass-produced items being sold through a social-commerce funnel.
This pattern is important because it shows how AI changes the fraud landscape. Traditional scams might use stolen photos, generic stock imagery, or vague branding. AI avatars can simulate a consistent persona across platforms, maintain a coherent “seller identity,” and present a believable origin story—complete with emotion and specificity—without any actual production capability behind it.
In other words, the scam isn’t only about selling a product. It’s about selling a relationship.
How AI avatars become scalable storefronts
The report describes AI-generated influencers created to sell mass-produced products via dropshipping-style operations. That phrasing points to a key operational advantage: AI content can be generated quickly, adapted easily, and deployed repeatedly.
A real human seller has limits. They can only film so many videos, they can only respond so fast, and their availability is constrained by time and logistics. An AI avatar, by contrast, can be used to produce a steady stream of content designed to match platform rhythms. If a particular hook performs well—tears, humor, a dramatic reveal, a “small business” plea—the system can replicate the structure with minimal cost.
This is why the issue feels different from earlier waves of online deception. The internet has always had fake reviews, fake giveaways, and fake storefronts. But AI-generated identities add a new layer: the deception can be visual, emotional, and personalized, not just textual.
It also means the scam can be more resilient. If one account gets flagged, another can appear with a similar persona and a slightly altered story. If one video is removed, new ones can be generated to replace it. The “brand” is not tied to a single person or a single set of assets—it’s tied to a template.
Targeting trust, not just attention
There’s a temptation to describe these scams as purely attention-grabbing. But the deeper strategy is trust engineering.
Social platforms reward content that keeps people watching. Commerce platforms reward content that converts. The scammers are combining both incentives by building videos that look like personal testimony rather than advertising. The avatar’s tears, the “handmade” framing, and the identity cues all function as trust signals.
This is where the report’s implications extend beyond one product category. The same approach can be applied to many kinds of goods—fashion accessories, beauty items, novelty products, household items—anything that can be sold with a short demonstration and a compelling narrative.
The “fake Black sellers” angle, specifically, raises additional concerns. When identity cues are used to manufacture credibility, the harm isn’t only financial. It also risks turning real communities into marketing props. Even if the scammer never intends to “harm” a particular group, the effect is that audiences are manipulated using identity markers as a shortcut to empathy.
That manipulation can be especially damaging because it trains viewers to associate certain emotional performances with authenticity. Once that association is exploited, it becomes harder for legitimate creators to be believed.
Platform integrity: the enforcement gap
The existence of these AI avatars highlights a practical question: how do platforms detect deception at scale?
TikTok, Facebook, and Instagram already moderate content for spam, impersonation, and fraud. But AI-generated avatars complicate enforcement. A video can look convincing, the seller can appear consistent, and the content can avoid obvious red flags. Meanwhile, the scam’s “truth” is hidden behind the identity itself: the avatar may be synthetic, but the video can still be technically compliant with many content rules.
There’s also the speed problem. Scams move quickly. Accounts can be created, content can be posted, and purchases can happen before moderation catches up. Even when platforms remove some content, the underlying business model can persist through new accounts, new storefront links, and new variations of the same persona.
The report’s description of identical or similar products suggests another enforcement challenge: the scam may not rely on unique items. It may rely on repeatable supply chains and repeatable marketing templates. That makes it harder to treat each listing as a one-off anomaly.
What’s needed is not only content moderation, but commerce integrity—verification systems that connect sellers to real-world accountability. Without that, platforms risk becoming distribution channels for deception, even if they remove some individual posts.
Consumer trust: the cost of manufactured “realness”
For buyers, the immediate harm is straightforward: paying for goods that may not match expectations, arriving late, or failing to deliver quality. But there’s a second-order harm that’s harder to quantify: erosion of trust.
When audiences encounter AI-generated personas that mimic authenticity, they begin to doubt everything. Legitimate creators may suffer because consumers become more skeptical of emotional storytelling and “small business” narratives. At the same time, scammers benefit because skepticism can be selectively exploited—by presenting just enough realism to pass early checks while still operating under false premises.
This creates a feedback loop. The more convincing the deception becomes, the more cautious users become. But caution doesn’t stop impulsive buying triggered by short-form urgency. It just shifts the burden onto consumers to verify authenticity manually—something most people won’t do for every impulse purchase.
The result is a marketplace where trust is increasingly automated, and consumers pay the price when automation is used for fraud.
Why this matters now: AI lowers the barrier to impersonation
AI has been used for scams for years, but the current wave is notable because it targets identity and emotion, not just text.
Earlier scams might have used phishing pages, fake customer service chats, or misleading product descriptions. Those require the scammer to persuade the victim through information asymmetry. AI avatars add a different persuasion layer: they can simulate a human presence and emotional performance, making the scam feel less like a con and more like a relationship.
That shift matters because it changes how people decide. Many buyers don’t evaluate a storefront like a contract. They evaluate it like a person. If the person appears sincere, the buyer’s guard drops.
AI makes it possible to generate that “person” cheaply and repeatedly.
A unique take: the scam is a content business first, a retail business second
One way to understand what’s happening is to flip the usual assumption. We tend to think of these operations as retail scams—sell junk, collect money, disappear. But the report suggests something more sophisticated: the core product is content designed to drive engagement and conversion.
The AI avatar is the marketing engine. The product is almost secondary. The system’s real value is its ability to produce persuasive videos that fit