Viral & Trends

The AI Influencer Accounts That Almost Fooled Everyone

By Caroline MillerPublished: May 7, 2026

They had hundreds of thousands of followers. Brand deals. Fan communities. Comment sections full of people saying "she's so beautiful" and "I love her style." There was just one thing missing: they weren't real.

AI-generated influencers have been quietly infiltrating social media for years – and some of them got very, very far before anyone started asking questions. Here's a look at the accounts that blurred the line between human and generated, and what it says about where social media is headed.

The One That Started the Conversation: Lil Miquela

If there's a ground zero for AI influencers crossing into mainstream culture, it's Lil Miquela. The CGI character – created by LA-based startup Brud – first appeared on Instagram in 2016 presenting as a 19-year-old Brazilian-American model and musician living in Los Angeles. She posted selfies, attended fashion events, wore real designer clothes, and collaborated with brands including Prada, Calvin Klein, and Samsung.

What made Miquela different from an obvious cartoon mascot was the texture of her presence. Her captions read like a real young woman navigating life. She had opinions. She addressed social issues. She "broke up" with a fictional boyfriend in a post that went semi-viral. For a significant period, her followers weren't entirely sure what they were looking at – and Brud leaned into that ambiguity deliberately, neither confirming nor denying her artificial nature for months.

By 2018, Miquela had over a million Instagram followers and was being covered by Time magazine as one of the 25 most influential people on the internet. The acknowledgment of her CGI nature didn't tank her following – it arguably boosted it. She became a cultural conversation piece, which is its own kind of influence.

Shudu: The World's First Digital Supermodel

Where Lil Miquela leaned into the sci-fi aesthetic, Shudu Gram went in a different and more controversial direction. Created by British photographer Cameron-James Wilson in 2017, Shudu was presented as a strikingly beautiful Black woman – a digitally rendered model whose images were photorealistic enough that many viewers, and initially some media outlets, took her for a real person.

Shudu gained significant attention when Fenty Beauty reposted one of Wilson's images of her wearing their products. The post got thousands of likes and was initially interpreted by many as a brand campaign featuring a real model. The reveal that she was entirely computer-generated prompted a sharp debate: was this impressive digital artistry, or was it problematic that a white creator had built a digital Black woman and was profiting from her image? That conversation spread far beyond beauty industry circles and touched on questions of representation, authenticity, and who gets to create and benefit from digital likenesses.

The Shudu situation was significant not just for the debate it sparked but for the proof of concept it demonstrated: a photorealistic digital human could circulate in real media spaces and be accepted as genuine, at least initially. That had real implications for how the influencer industry would evolve.

Imma: Japan's Virtual Icon

While the Western conversation about AI influencers was heating up around Miquela and Shudu, Japan had quietly developed its own virtual star. Imma – created by Tokyo-based company ModelingCafe – has a distinctive aesthetic: photorealistic face, pink bob haircut, and a fashion-forward wardrobe. She's appeared in campaigns for IKEA Japan, Porsche, Amazon, and Valentino, among others.

What's interesting about Imma is how openly she operates as a virtual being. She doesn't attempt to pass as human – her uncanny-valley aesthetic is part of the appeal. Yet she engages with her audience in the first person, documents her "life" through posts, and has built a fanbase that treats her as a genuine personality. The distinction between "AI character" and "virtual influencer with a real following" is genuinely blurry in her case. She has over 400,000 Instagram followers and a verified account, and the brands she works with are treating her with the same commercial logic they'd apply to a human talent.

The Accounts That Were More Deceptive

Lil Miquela, Shudu, and Imma are all cases where, once people looked closely, the artificial nature was at least eventually discoverable. A more unsettling category involves accounts that were designed to deceive from the start.

In 2023 and into 2024, reports emerged of AI-generated influencer accounts operating on platforms including Instagram and TikTok that were selling subscriptions to "exclusive content" – including explicit material – featuring people who didn't exist. These accounts used AI image generation tools to create convincing personas, gave them backstories and personalities, and monetized them through platforms that don't require identity verification. Some accumulated significant followings before being identified and removed. Others are almost certainly still operating.

These cases moved the conversation from "interesting cultural experiment" to genuine concern. Followers who believed they were forming connections with real people – and in some cases paying for personal interactions – were being misled in a way that crossed from novelty into exploitation. The FTC and consumer advocates began raising questions about disclosure requirements and platform accountability that are still being worked through.

Why It Worked – and Why It Still Works

The success of AI influencers at fooling audiences (or at least capturing their genuine engagement) isn't really about the technology being perfect. If you look closely at Lil Miquela's early posts, the CGI is visible. Shudu's skin has a particular rendered quality that photographers recognize. The reason these accounts worked is more about how people consume social media than about the images themselves.

We scroll fast. We make snap judgments. We extend social trust to accounts that look like they fit the patterns we've come to expect – the right aesthetic, the right caption tone, the right follower count. Our brains are not running a reality check on every post we double-tap. The AI influencer accounts that succeeded exploited the exact same inattentive engagement that makes social media work for everyone else.

There's also the parasocial dynamic. Once you've followed an account for a while, left comments, watched stories, felt like you know the person – it becomes psychologically uncomfortable to revise that connection downward. Some of Lil Miquela's followers who found out she was CGI kept following anyway, because the relationship they'd formed didn't feel fake to them even if the subject of it was.

Where This Is Going

The technology has improved significantly since 2016. Modern AI image generation – and increasingly, AI video generation – produces outputs that are harder to detect than the early CGI influencers. The barriers to creating a convincing virtual persona have dropped dramatically: what required a well-funded startup like Brud to build in 2016 can now be approximated by an individual with access to widely available tools.

Several talent agencies have begun signing virtual influencers alongside human talent. Some human influencers have created AI versions of themselves to handle brand deals and engagement while they step back from the grind of constant content creation. The line between "a real person using AI tools" and "an AI character performing a person" is getting harder to define – which is exactly the kind of ambiguity that platforms, regulators, and audiences are going to have to get a lot more sophisticated about navigating.

The Federal Trade Commission has been moving toward clearer disclosure requirements for AI-generated content in advertising contexts. Some platforms have introduced labels for AI-generated images. Whether those tools keep pace with the technology is an open question.

Outro

The AI influencers that almost fooled everyone didn't succeed because people are gullible. They succeeded because they understood exactly how social media attention works – and they were built to fit perfectly inside it. That's a lesson worth sitting with, regardless of whether the account you're about to follow turns out to be human or not.