Dark-skinned Black women are going viral for their trending TikTok and Instagram dances. But a Glamour investigation has found that many of these accounts aren’t actually real women. What’s worse? In many cases, they’re linked to paid-for subscriber content services, directing users to legally dubious, AI-generated, sexualised content.
These AI influencers are built by scraping real Black women creators’ content without consent, replicating their faces, bodies and movements for views and profit. Their personas are often exaggerated and hyper-sexualised with darker, unrealistic skin tones, fetishised features, and content that appears increasingly tailored toward adult audiences. Experts describe this as a form of ‘synthetic doppelgänging’, where AI creates near-identical digital doubles that mimic real people while sidestepping image rights protections, marking a new, more insidious evolution of digital blackface.
Glamour investigates the rise of this disturbing phenomenon, exploring how lagging digital safety policies are still failing Black women online.
Since late December 2025, Instagram and TikTok accounts featuring hyper-realistic, AI-generated Black women have been amassing millions of followers. The videos typically depict Black women dancing, styled to appear hyper-exotic and, at times, hyper-sexualised, often with exaggerated features such as extremely dark skin tones or vitiligo. But do these women even exist?
In early January 2026, 21-year-old Josephine, a TikTok content creator based in Sierra Leone, was scrolling through her feed when she stumbled upon a video that looked eerily familiar. The dance, transitions and camera angles looked exactly like content she had posted nearly four years ago. Neither the video nor the account belonged to her. The clip had already garnered a substantial number of views, according to Josephine, far surpassing her original video. It featured a hyper-realistic, dark-skinned AI-generated woman.
“I thought someone was tagging me in their own version,” Josephine says. “But then I realised it wasn’t even a real person. It was an AI version of a Black woman with a dark skin tone that didn’t look real, recreating my every move. From my network of creators, I had heard that there were dark-skinned AI-generated Black women going viral on TikTok. I never gave anyone permission to use my video. I felt like someone had created a sloppy caricature of me for views and likes.”
The rise of synthetic doppelgänging
These AI deepfakes are a form of synthetic doppelgänging, as Sarah Armstrong-Smith, cybersecurity expert and keynote speaker with Champions Speakers Agency, explains. “There’s enough similarity to mimic the look, traits and personality of a real human, but giving it a synthetic and AI look, so as to try to bypass any potential claims on identity theft, or copyright infringement. These AI-generated female personas can pose significant risks to real women and girls who are disproportionately targeted by non-consensual image manipulation, including sexual exploitation and reputational damage.”
The now-deleted AI video in question, which Josephine says was littered with repetitive generic comments on how the AI model resembled an exotic goddess, wasn't the only one stolen from her account. Another one of her dancing clips had also been replicated on the same TikTok profile.
When Glamour reviewed the profile, there was no sign of the AI-generated clips that Josephine alleged had been stolen from her account. However, an archived side-by-side comparison of the two videos remains on a TikTok account dedicated to documenting stolen AI-generated content.
Josephine’s encounter is just one of many emerging across various social media platforms.
Many of these AI personas appear to be built from existing influencer content without consent. One of the most prominent examples was an account operating under the name Nia Noir, which has since been removed from TikTok. Before its deletion, the account had one million followers, and one video had garnered more than 160 million views. Glamour has seen evidence of multiple accounts using the same name, ‘Nia Noir’, across TikTok.
One of Nia’s dancing clips mirrors a video originally posted by a famous TikTok influencer, Tatiana Kaer. Another upload appears to have been lifted from a video from the creator, Janice Nichole. In some cases, the AI-generated version of Nia appears to be more overtly sexualised than the original footage. Nia and @blackcoffeeamara are just a few of the AI-generated Black women’s accounts that have been popping up on people’s algorithmic feeds.
TikTok content
TikTok content
Many of these personas have been seemingly tailored towards fetish audiences. One now-deleted TikTok video was captioned, “proud to be a Slavic doll.” Another now-deleted video encouraged viewers to “bounce to manifest a white boyfriend”, while another had the caption, “there’s a white man behind you,” in which the AI-generated figure drops to her knees in a suggestive pose, a gesture that can be interpreted as reinforcing racialised submission stereotypes. Another video lists “3 reasons to date me if you’re a white guy”, which include “free n-word pass” and “It’s actually pink,” referencing a heavily pornified image of Black women’s genitalia.
According to TikTok, multiple accounts and pieces of content were removed for violating the Community Guidelines. TikTok prohibits and has removed content that facilitates access to sexual services, as well as the posting of AI-generated content of private individuals without their consent. Spam and impersonation accounts are also prohibited. A spokesperson for TikTok further notes that over 1.3 billion videos have been labelled as AI-generated on the platform to date, helping users spot, shape, and understand their experiences with this content. Users are encouraged to report potential privacy violations, including non-consensual AI-generated content here.
But it’s not just TikTok; over on Instagram, there are multiple accounts with some variation of the name Nia Noir, one of which includes a bio describing her as “Just a girl with a dark side”. One account has over 600,000 followers at the time of writing. When Glamour clicked her Instagram bio, it took us to a Linkby account where Nia is described as ‘Your Midnight Obsession”. There was also the option to click through to ‘Exclusive content 🥵’, which led to Nia’s Telegram featuring more images geared towards adult content. At the time of publication, the group had 13k subscribers. Despite her rapid growth, Nia Noir is unlikely to be a real person. She is probably a combination of generative AI software, face-swapping tools and face-generation software. Nia also appears to be visually inspired by a real-life dark-skinned model, Lola Chuli, a South Sudanese model and internet sensation who went viral for her looks in 2016. Glamour was unable to make contact with Chuli for a comment.
These accounts are also being integrated into adult content ecosystems. In at least one instance, a well-known adult performer, Johnny Sins, appeared in a now-deleted reel posted to two of the AI-generated accounts on Instagram. While it remains unclear who operates these profiles, the hypersexualised content and the OnlyFans links that appear on nearly all of the accounts suggest they were created primarily for monetisation. There is nothing to suggest this was done with Johnny Sin's knowledge.
Glamour also investigated a now-deleted Instagram account purportedly run by an influencer named Sade. At the time of our investigation, the account had six images of Black women, all with slightly different faces, skin tones, and facial features. All had straight, black hair and slim bodies. A link in Sade’s bio directed us to “Exclusive content” on Fanvue, a creator-focused subscription platform. Sade’s profile picture was of another Black woman, lying on a bed wearing white stockings and a pink bikini. Once a subscription fee of £3.55 was paid (and age verification completed), Glamour was able to see partially-nude, sexualised photos of Sade. The account did not respond to our message asking if Sade was a real person.
A spokesperson for Meta said, “We have removed the accounts for violating our policies.”
A spokesperson for Fanvue said, “Fanvue is a creative home for creators across lots of different verticals, including music, fashion, sport and lifestyle. The platform does allow AI Influencers, and all AI-generated content is clearly labelled as AI. AI Influencers are growing in popularity, showing up in mainstream cultural moments like Coachella, and building engaged fanbases on social media. It's an exciting industry, that's helping more creators join the booming creator economy.”
How do we know who is and isn’t a real person?
Cyber security expert Armstrong-Smith says, “These accounts are created using advanced generative models and deep learning techniques that can mimic human appearance, voice, and behaviour with astonishing accuracy,” Sarah explains. “The synthetic personas are built by scraping real videos and images, then using AI to generate new content that appears authentic.” New tools have made it easy to create AI influencers and clone already existing content.
“Once the model is trained, it can produce endless variations, post content around the clock, and interact with followers in real time, making it easy for these accounts to gain traction. However, much of this engagement is not genuine; many likes and followers are generated by other AI or bots, creating a false feedback loop that further boosts the account’s visibility,” Armstrong-Smith adds.
As AI-generated content becomes more interactive, more people are finding it difficult to distinguish between real and synthetic accounts — most people are still not aware that Nia and Sade, for example, are likely to be AI-generated. However, James Bore, a Chartered Security Professional who works in the online safety and impersonation space, tells Glamour that there are some very telling signs that the content in these specific accounts may be AI-generated.
“The content tends to be quite generic and repetitive. You'll see the same poses, the same little dances, the same sort of thing coming up again and again because the people behind it will just be reusing prompts,” he says. “There’s also the lack of anyone else in the videos; there are no collaborations, there's nothing interactive. If you look at the different videos, there's really nothing human about them. They are all perfectly polished, perfectly presented, and done in a very specific way with very fixed expressions. They [ the videos] also tend to be short. Those aren't definitive that something is AI-generated, but they're a very strong hint.”
The rise of digital blackface
All of this is happening at a time when Black women creators continuously report ongoing shadowbanning, harassment and limited monetisation on the very same platforms. A 2023 peer-reviewed study on TikTok found that Black creators consistently report bullying from viewers, difficulties with monetisation and algorithmic suppression of their content. According to Tati Kapaya, a content creator and founder of Creators on Creators, the rise of AI Black influencers feels deeply ironic: “The whole issue of Black women’s content not being pushed in the same way as white women’s content has been ongoing. So to now see an AI account make it overtly clear that there is space and opportunity for us, but platforms still won’t give that space to real Black women feels a lot more deep-rooted.”
Tati points to the long-standing disparities in pay and exposure. “Black women are paid less. We get less exposure. There’s less respect across the board. This is just adding to that. It’s not surprising, but it's extremely disappointing.” Black women influencers are repeatedly undervalued across social media platforms. A 2024 Influencer Pricing Report by Sevensix Agency, which collects data from hundreds of influencers in the U.K. and beyond, revealed that Black influencers are paid 34% less than white influencers.
Unfortunately, what we are witnessing now is not entirely new. It is a resurgence and escalation of digital blackface, the use of Black imagery, aesthetics or personas by non-Black creators for entertainment, profit or spectacle. Historically, the term, likely conceived by Joshua Lumpkin Green in a 2006 academic paper, was most commonly associated with memes and reaction GIFs, where images of Black people were disproportionately used to convey exaggerated emotion. But in recent years, the practice has evolved.
In 2017, journalist Lauren Michele Jackson examined the dominance of Black faces in reaction GIF culture in her essay, “We Need to Talk About Digital Blackface in Reaction GIFs.” She argued that Black expression had become shorthand for all sorts of emotions online, from anger to attitude to drama, which was reinforcing long-standing stereotypes of Black animatedness. Since then, digital Blackface has evolved much faster. In 2025, there was a wave of AI-generated videos featuring Black women using exaggerated African American Vernacular English (AAVE) and yelling about not being able to use their EBT cards. Then came the ‘Bigfoot baddies’ AI videos featuring Black women as primates, complete with derogatory stereotypes.
“AI has become more accessible, and it’s strange to think that the first people who are being targeted are women, more so Black women,” Josephine says. “It’s a very sinister form of digital Blackface because whoever is using our image to push stereotypes about Black women, create fetishistic content for white men, and monetise exaggerated versions of Black femininity is probably not even Black.”
Benjy Kusi, a UK creator who comments on media literacy and pop culture, once drew harassment for a reel explaining the harms of digital blackface; some commenters accused him of overreacting. In a follow-up video, he warned: “Our failure to interrogate seemingly small behaviours like meme usage can lead to a wider failure to challenge systems that repeat these harmful dynamics at scale.”
Now, he tells Glamour UK that the risks have intensified.
“Generative AI is continuing a long pattern of Black people being dehumanised, treated as a resource, and extracted for financial gain with little responsibility taken. These ‘creators’ are influencing how Black women are perceived and treated online and offline. Calling it harmless ignores this tangible impact and the power dynamics that allow these patterns to be repeated.”
Benjy adds that there is a stark contradiction in how platforms and audiences value Blackness versus Black people. Systems consistently elevate aspects of Blackness they can control and monetise, without recognising responsibility toward the people behind that culture. AI-generated Black women, he argues, are a striking example of how online spaces are fuelled by Black creativity, yet credit, safety and profit are rarely distributed back to Black communities.
The new offence – which includes AI-generated images – carries a maximum sentence of six months’ imprisonment.
Where do we go from here?
Current safeguards are still struggling to keep pace with the rapid evolution of AI-generated content. While some social media companies have introduced labelling and watermarking for synthetic media, enforcement remains inconsistent and legal frameworks are still catching up, making it easier for AI-generated content to misuse real identities without any meaningful recourse.
Social media experts like 28-year-old Sabrina Fearon-Melville, a freelance culture journalist and social media executive for BBC Three, remain sceptical about current AI regulation standards in the U.K. “At this point, the ‘regulation’ often feels like more AI. If you’re unsure about a video, you’re told to run it through an AI checker to see if it’s synthetic. But if you don’t want to use AI in the first place, having to use AI to detect AI feels like a catch-22.”
Using synthetic identities could also pose a challenge to law enforcement, which is still playing catch-up with technological change, especially considering the fact that it might be difficult to prosecute a persona that simply does not exist.
Professor Clare McGlynn, a law professor and expert in violence against women and girls, explains that current UK legislation only applies in relatively narrow circumstances. When asked whether the images on Sade’s Fanvue account would be considered illegal, she explains that laws on sexually explicit deepfakes generally apply only where the image is realistic. “You have to be able to look at it and think it could be a real photo, so where the material is obviously AI and ‘not real’, the law won’t apply. The law applies to intimate states, but just underwear may not be included. The offence is also based on it being an image or altered image of a specific person, so there has to be an identifiable person who is the person who has not consented.”
This creates a grey area for many of the AI-generated accounts Glamour identified. Where content is clearly synthetic, or where images are composites built from multiple sources rather than one recognisable person, McGlynn says legal protections may not apply. “The closer it is to ripping off the content of one specific person, the more likely the law will protect them.” Copyright law may offer some recourse where original content is directly copied, but this is complex when AI systems generate new material from multiple sources.
Bore adds that there are very limited protections to your likeness in the UK. “Your copyright is protected, but your copyright doesn't mean your likeness. Regulation is theoretically simple but practically, it's a lot more difficult. On top of that, the people behind the accounts aren't in a regime which allows us to regulate them. It would be better to have regulations that could then pressure the platforms to take down content. The Online Safety Act, for example, takes some steps towards that, but it leaves a huge gap.”
Despite this, Armstrong-Smith tells Glamour that these platforms still have an ethical responsibility to identify, label and restrict AI-generated personas, particularly those designed to impersonate real people: “We need transparent labelling, robust moderation and clear accountability measures to prevent misuse and protect users from deception. Platforms should also provide accessible tools for reporting and removing synthetic content that violates consent or likeness rights, while users need resources to recognise and respond to AI-generated media.”
Sarah suggests that perhaps the only way forward is continuous ongoing adaptation to regulate this space, as well as a long-term collaborative approach between all the players that combines regulation, platform governance and industry accountability.
But adaptation alone might not confront the deeper issue behind this trend. If platforms have proven there is profit in the image of Black women, can they ever show the same urgency in protecting the real Black women behind it?
Josephine remains convinced that adaptation alone won’t fix what she sees as a much older problem, especially considering that appropriation of Black women’s voices, bodies and culture long predates AI.
“This didn’t start with AI,” she says. “Black women have long been copied, exaggerated and profited from for years. AI just makes it easier now.”
For Josephine, this still remains fundamentally a race issue. And until platforms are willing to confront that reality too, no amount of regulation, online safety laws or technical adaptation for Josephine is likely to address the harm at its root.
With additional reporting by Lucy Morgan.
‘Even if the rest of the school eventually moved on, I never did. It hung over me constantly.'
