Digital Blackface Avatars: AI's Sales Pitch

A compelling question emerges from the digital marketplace: can appeals to race, guilt, and empathy persuade consumers to purchase a $9 belt buckle for $40 on platforms like TikTok?

Consider the case of Aliyah, an avatar portraying a light-skinned Black woman in country-western attire, seemingly struggling to sell her supposedly handmade metal buckles on TikTok. In a March video, she appears to weep while appealing for visibility, with on-screen text declaring, “Even as a black woman, I have more faith that white women will stay 13 seconds [on this video] to save my belt buckle business.” She is depicted wiping a tear from her face.

However, Aliyah is not a real person, nor are her products genuinely handmade. She represents one of many AI-generated influencers engineered to dropship mass-produced goods across TikTok, Facebook, and Instagram. Identical belt buckles, complete with a sunflower design and detachable knife inlay, are readily available on the fast-fashion website Shein for a mere quarter of the price.

Several indicators reveal the AI-generated nature of these videos. Aliyah’s voice, for instance, is notably robotic and devoid of emotion, a stark contrast to her crying on screen. In one particular clip, she is shown sewing a leather belt in an area where stitching would not typically occur. Furthermore, when she wipes a tear, the visible liquid stream beneath her hand inexplicably vanishes. Compounding this, numerous strikingly similar videos featuring different AI-generated characters circulate on TikTok. An account named “Aliyahsbuckles,” for example, displays an identical background, tabletop, and spool of twine.

The Verge identified dozens of accounts on TikTok employing similar narratives to dropship a diverse range of products, including belt buckles, cowboy boot-shaped mugs, crochet bags, and cardigans. While some of these videos are explicitly labeled as AI-generated, many are not. Similar deceptive accounts are also active on Instagram and Facebook. Nearly every element of these accounts, from the on-screen persona to automated comment responses—some attempting to mimic African American vernacular—appears to be AI-generated, leading experts to warn of the daily proliferation of such scams.

“It’s massive,” stated Jeremy Carrasco, a researcher specializing in AI-generated media and director of Riddance.ai, an organization dedicated to AI video detection, in an interview with The Verge regarding AI-generated videos linked to e-commerce stores. He elaborated, “Most of them aren’t coordinated. Some of them are coordinated. A lot of the time they’ll run a single [AI-generated] actor, or a couple actors will run all sorts of shops.” These AI-generated avatars feign the creation of items, attend virtual fairs to display products, and automatically “respond” to comments. “What we’re seeing right now are these retail scams where they’ll link to the Shopify websites.”

Carrasco estimates that his research team uncovers up to 100 accounts daily that leverage AI-generated avatars to sell products. Most accounts discovered by The Verge were established within the past two months and feature videos depicting small businesses owned by marginalized individuals struggling to make sales. These videos exhibit remarkable similarity, with only minor script variations. While characters representing Native American, Hispanic, and white women were also found, the most viewed and engaged-with AI-generated characters identified by The Verge are Black women. Aliyah’s account alone boasts 40,000 followers.

“What we’re seeing here is empathy bait,” Carrasco explained. He added, “If there is a popular dropship item that could be sold to some sort of niche community, they will find it and they will try [to use] some personality to do it.” Carrasco clarified that these trends often represent opportunistic ventures to generate income. “It’s just an arbitrary opportunity, which is what a lot of AI content out there is — the platforms don’t really care and people don’t notice.”

Aliyah’s most popular video, the one highlighted at the beginning of this article, has garnered an astounding 814,000 likes, 6.5 million views, and nearly 30,000 comments. While some comments correctly identify the content as AI-generated, many express a genuine desire to support Aliyah’s business, commenting to boost her visibility on the platform.

India Cater-Campbell, an actual Black business owner in Seattle preparing to open a café, was one such commenter, intending to purchase Aliyah’s belt buckles. Cater-Campbell explained, “I was trying to be supportive to an independent Black businesswoman,” adding, “[I felt] solidarity as I am trying to start a business myself.”

Despite a scarcity of studies on the latest generative AI video models, Carrasco believes these videos are “realistic enough to trick” most individuals. Users of short-form video platforms have been conditioned to scroll mindlessly, often neglecting to scrutinize the content they consume. Ironically, this ingrained behavior may have inadvertently saved Cater-Campbell from making a purchase: unable to immediately locate a store link, she scrolled past and soon forgot about Aliyah.

“I was trying to be supportive to an independent Black businesswoman.”

Nevertheless, people are succumbing to these scams, and their prevalence is increasing. Just two weeks prior, Gizelle Bryant from The Real Housewives of Potomac confessed to buying two crocheted bags after encountering a video in which an AI-generated Black boy claimed he was being bullied by white boys for crocheting. “I was like, I want to help this little Black boy make his goal,” Bryant shared on her podcast, Reasonably Shady, noting that other celebrities had also appeared in the comment section. She expressed her surprise, saying, “How did I get tricked? Viola Davis was on there, too.”

This trend constitutes a form of digital blackface, according to Cienna Davis, a communications researcher at the University of Pennsylvania. Davis defined digital blackface during a video call as “a phenomenon where non-Black individuals are able to use the internet and digital technologies to mimic Black cultural expression for personal, economic, or political gain.” Davis, who has previously written on the subject, points to instances such as the use of GIFs featuring Black people and the impersonation of Black individuals for political motives. She clarified, “It’s rooted in blackface minstrelsy, which is tied to the legacy of slavery.”

Based on the premise that Blackness is “inherently exploitable” and “up for grabs,” digital blackface is employed to “extract value from Black bodies in whatever way [non-Black people] see fit,” Davis asserted. In this specific context, the videos simulate “a recognizable idea of Black struggle,” she added.

Even without definitive confirmation that the individuals behind these videos are not Black, Tempest M. Henning, an assistant professor of philosophy at Fisk University, confirms their classification as digital blackface. Henning explains, “Blackface is any kind of caricature-like portrayal of Black people, which can include Black people dressing up in a caricature-like fashion of Black people.” She cites historical examples, such as Black individuals occasionally coerced into performing in minstrel shows, and more recent instances like Zoe Saldaña darkening her skin tone and using prosthetics to portray Nina Simone in a biopic.

An inherent artificiality pervades the Blackness these avatars claim. As Henning elaborated, “The names of the avatar are coded Black, like Aliyah or Amaya, but there’s nothing else [that signals authentic Blackness] other than the avatar itself.” This inauthenticity is further accentuated by the replication of content across various racial identities, which, Henning notes, leads to a flattening of those identities.

The Verge attempted to contact Aliyahsbuckles and other vendors selling similar products but received no response.

The videos posted on Aliyah’s account are systematically replicated scene by scene across numerous similar accounts offering dropshipped items, with minor adjustments to suit the character’s identity or the product being sold. In one video featuring another Black woman character named Amaya, a white woman mockingly spills coffee on belt buckles displayed at a fair. A mildly frustrated Amaya — whose emotions appear somewhat robotic and never quite convincing — sighs and resumes her “honest work” of crafting buckles. A similar scenario unfolds in an account called ChubbyKnots; here, the avatar is a Black girl, and the product is a crocheted butterfly cardigan.

Henning highlighted the appeal to virtue signaling embedded in these videos, which encourages viewers to demonstrate kindness or even racial or class solidarity. “It’s virtue signaling in the sense of ‘Oh, that white lady did that, but I am not that kind of white lady, so where do I buy this belt buckle?’” Henning explained. “It’s like they’re saying, ‘I don’t stand with these people.’”

However, Henning pointed out the superficial nature of such race- or class-based support when users fail to pause and investigate whom they are truly supporting.

This phenomenon arguably reflects a broader social consequence of short-form video content and the consumption of "slop": solidarity is often performed at a “surface level,” she explained, rather than through a coherent and multifaceted political lens. Henning elaborated, “I might be very invested in a Black-owned business, but that doesn’t necessarily mean that all Black businesses align with my political views and what I stand for.”

This appeal to virtue signaling extends beyond racial categories, also leveraging markers of the working class and simulating individuals striving to earn a living in a struggling economy. Davis noted, “[It’s not just] Black struggle because on the other videos, we can observe working class struggle, small-business struggle.” She continued, “But it’s obviously using that [narrative] to sell mass-produced goods. It’s not creative at all. It’s really just AI-generated templates and creation of characters that get optimized for consumer identification and investment in the products. It’s broader than ‘white guilt’ altogether, because it can be used in whatever way feels necessary or feels applicable to the product.”

Despite, or perhaps because of, this flattening of racial identity, algorithms—known to perpetuate conscious and unconscious racial biases from their creators—are delivering AI-generated content tailored to each user’s inferred interests, according to Carrasco. These interests may include content from specific demographics that AI personas are designed to mimic. “It targets the demographic that [the algorithm] can see,” he said. “Usually a Black woman will see AI-generated Black people on their feed, an Asian person will get AI-generated Asians.”

Upon closer inspection, inconsistencies and tell-tale signs of AI generation become apparent. Characters are rarely shown on screen for more than a few seconds, due to the limitations of AI applications like Seedance 2.0, Midjourney, and OpenAI’s recently discontinued Sora 2, which cannot generate clips longer than 15 seconds. When products are showcased, the lighting is often brighter and less sepia-toned, and the hands holding the buckles often appear white. Furthermore, the emotional expressions on the characters’ faces frequently fail to align with the tone of their voices.

However, discerning these cues requires a level of AI media literacy that is not widely prevalent, especially given the sheer volume of these videos being created and the rapid advancements in automation and avatar realism. Davis cautioned, “People who aren’t trained in media literacy or critical media literacy are just going to take it in unquestionably.”

The very design of short-form video platforms, which serve as significant channels for misinformation, actively discourages user discernment. Users have been conditioned to prioritize scrolling to the next video over critically evaluating what they are watching. Carrasco explained, “[These platforms] are being exploited by AI content partially because in order to figure out something’s AI, it usually takes a second or two. By that time, you’ve already registered some engagement.”

Those who profit from these videos are aware of these behavioral patterns, and the tools necessary to create scalable, AI-generated short-form commercials are widely accessible online. During its reporting, The Verge discovered YouTube channels and forums offering tutorials on how to produce such ads without requiring product samples, influencer fees, or original scripts.

These tutorials instruct users on how to replicate viral videos using various AI tools, replacing real influencers with AI-generated characters. ChatGPT and Gemini are utilized to extract and copy scripts from authentic influencer videos, with the option to generate multiple script variations. These tools also facilitate the creation of fake individuals and backgrounds based on real influencers and their environments, which are then imported into applications like Kling 2.0 and Maxfusion. Some AI video generator models even allow for the full import of real influencer videos, enabling users to substitute the actual person with a custom AI character, effectively appropriating the original video’s script and background.

While increased media literacy might help individual users avoid AI-generated video scams, social media platforms are currently failing to adequately moderate and, at minimum, label this content. Davis emphasized, “Platforms should have stronger AI detection and labeling structures.” She added, “Bias checks on these platforms would also be welcome, as well as clear reporting pathways for reporting AI misuse on platforms.”

In addition to moderation, Carrasco proposed that account transparency would empower users to identify AI-generated content, advocating for clear, easily visible labels that do not require additional steps to access.

However, short-form video consumers have years of ingrained habits involving superficial engagement with their feeds. Reversing this cognitive conditioning on a mass scale appears almost insurmountable without platforms themselves taking proactive measures. With no significant incentive for short-form video platforms to act, and substantial profits at stake, such intervention seems improbable. Until then, AI scammers are likely to continue exploiting those who fall for their bait.