AI Avatars in Digital Blackface Promote Controversial Belt Buckle

Aliyah, a light-skinned Black woman in country-western attire, is trying to promote her handmade metal buckles on TikTok. In a March video, she tearfully appeals for views, expressing her belief that white women are more likely to support her business. However, Aliyah is not a real person, and her products are not handmade; she is one of many AI-generated influencers designed to market mass-produced items through dropshipping on platforms like TikTok, Facebook, and Instagram. Identical belt buckles, featuring a sunflower design and detachable knife inlay, can be found on fast-fashion sites for a fraction of the price.

There are several indicators that reveal the AI nature of her video. Aliyah’s voice sounds robotic and lacks emotion, contrasting sharply with her crying expression. In one scene, she is shown sewing a leather belt in an unrealistic manner, and when she wipes away a tear, the stream of liquid vanishes. Additionally, numerous similar videos featuring different AI characters circulate on TikTok, including one for an account named “Aliyahsbuckles,” which shares identical backgrounds and props.

Research has uncovered many TikTok accounts with similar narratives and a range of dropshipping products, from belt buckles to mugs shaped like cowboy boots. Some videos are labeled as AI-generated, and similar accounts are also present on Instagram and Facebook. Most elements of these accounts, from the characters to automated comment responses that sometimes mimic African American vernacular, are AI-generated, raising concerns about the increasing prevalence of such scams.

“It’s massive,” says an expert in AI-generated media, noting that many of these accounts operate independently, while some are coordinated. These AI avatars simulate crafting products, attending fairs, and responding to comments through automation, often linking to Shopify websites. The expert’s team identifies up to 100 accounts daily that use AI-generated avatars to sell products. Many of these accounts, created recently, feature videos about small businesses owned by marginalized individuals, often with only minor script variations. While characters of various ethnicities are present, the most popular AI-generated personas are Black women, with Aliyah’s account boasting 40,000 followers.

“This is empathy bait,” the expert explains, highlighting how these accounts exploit niche markets by creating relatable personas. The trend often capitalizes on random opportunities for profit, with platforms indifferent to the authenticity of the content. Aliyah’s most popular video has garnered 814,000 likes and 6.5 million views, with many comments expressing a desire to support her business, despite some recognizing the AI-generated nature of the content.

One commenter, a real Black business owner, expressed her intention to support Aliyah, feeling a sense of solidarity as she works to launch her own café. Despite the lack of research on the latest AI video models, the expert believes these videos are convincing enough to mislead most viewers, who are accustomed to mindlessly scrolling through content. Ironically, this habit may have prevented the commenter from making a purchase, as she couldn’t find a store link and quickly moved on.

However, many are falling for these scams. A recent example involved a celebrity who admitted to buying crocheted bags after being moved by an AI-generated video featuring a Black boy claiming he was bullied for crocheting. The trend has been described as a form of digital blackface, where non-Black individuals exploit Black cultural expressions for personal gain. This phenomenon is rooted in historical exploitation and perpetuates the idea that Blackness is something to be commodified.

Even without confirmation of the creators’ identities, experts assert that these videos exemplify digital blackface. The avatars often have names that suggest Black identity, but lack any authentic representation beyond that. This replication across racial identities flattens the complexity of those identities, leading to a homogenized portrayal.

The videos on Aliyah’s account are often mirrored across various accounts, with minor adjustments to fit different characters or products. In one instance, another Black character, Amaya, is shown dealing with a white woman who mockingly throws coffee on her products. The emotions displayed by these characters often appear robotic and inauthentic.

Experts note that these videos appeal to viewers’ desires for virtue signaling, prompting them to support causes they believe are just. However, this support is often superficial, as many viewers do not take the time to investigate who they are actually supporting. This trend reflects a broader issue with short-form video content, where solidarity is performed at a surface level rather than through a deeper understanding of the issues at hand.

The appeal of these videos extends beyond race, tapping into themes of working-class struggle and the challenges faced by small businesses. However, the narratives are ultimately used to sell mass-produced goods, relying on AI-generated templates that optimize consumer engagement. Algorithms, which can perpetuate biases, deliver this content based on inferred user interests, often targeting specific demographics.

Upon closer inspection, inconsistencies and signs of AI generation become apparent. The characters rarely appear on screen for more than a few seconds, and the lighting and visual details often reveal their artificial nature. Recognizing these signs requires a level of media literacy that many users lack, especially as technology advances and the production of realistic avatars becomes easier.

The structure of short-form video platforms encourages users to engage without critical thought, making it easier for AI-generated content to thrive. Those profiting from these videos are aware of this dynamic, and resources for creating scalable AI-generated ads are readily available online. Tutorials exist that teach users how to replicate viral videos using AI tools, allowing them to replace real influencers with AI characters.

While media literacy could help users avoid falling for these scams, social media platforms currently lack effective moderation and labeling for AI-generated content. Experts advocate for stronger detection and transparency measures to help users identify AI-generated material. However, reversing the ingrained habits of short-form video consumers may prove challenging without significant changes from the platforms themselves. Until then, AI scammers will likely continue to exploit unsuspecting users.