In 2018, Time magazine listed a pink-haired, freckled digital character among the 25 most influential people on the internet. She had never existed in physical space, never held a phone, never taken a selfie in the ordinary sense. Yet millions of people followed her, argued about her, and felt something when she posted. The rise of virtual influencers is often told as a marketing story, but that misses the more interesting part: audiences kept showing up for characters they knew were not real. This is a short history of how that happened, and what it reveals about connection.
The prehistory: synthetic stars before the feed
Virtual personas did not begin with Instagram. Japan built the template decades earlier. In the mid-1990s, a computer-generated pop idol named Kyoko Date was marketed as a digital celebrity, an experiment that arrived long before the technology or the audience was ready.
The idea took hold in 2007 with Hatsune Miku, a Vocaloid singing voice sold as software. Miku had no fixed personality and no single author. Fans wrote her songs, drew her art, and eventually filled concert halls to watch a projection perform. That participatory model, a character owned partly by her audience, quietly became the blueprint for everything that followed.
What Miku proved was subtle but important. A synthetic character did not need to convince anyone it was alive to command real devotion. The emotional investment came from participation, from fans building the world together, and from the steady rhythm of new songs and art. Later virtual influencers would rediscover this lesson repeatedly, usually after trying and failing to win audiences on realism alone.
2016: the two debuts that split the field
Two arrivals in 2016 set the direction for the next decade. In Japan, Kizuna AI began posting on YouTube and effectively launched the VTuber format: an anime-styled character, voiced and motion-tracked in real time, talking directly to viewers. In Los Angeles, an Instagram account named Lil Miquela started posting photorealistic images so convincing that followers spent months debating whether she was a real person.
The contrast mattered. Kizuna AI was openly a cartoon. Miquela looked almost human. One leaned into stylization; the other chased realism. The years since have been, in large part, a running experiment about which approach earns a more durable audience, and the answer surprised a lot of people.
Why the stylized characters kept winning
The intuitive bet was that the more human a virtual influencer looked, the more people would like it. The evidence pushed back. Research on synthetic characters keeps bumping into the uncanny valley: the point where near-perfect human likeness starts to feel unsettling rather than appealing.
Audiences who know a character is computer-generated often engage more, not less, than they do with hyper-realistic ones. In Block and Lovegrove’s 2021 analysis of Lil Miquela, the authors describe a deliberate “honest fakery,” where the character’s openly constructed, part-human, part-nonhuman quality is precisely what draws people in. Perceived authenticity, in other words, did not require perceived humanness.
That finding reframes the whole history. The characters who lasted were rarely the ones trying hardest to pass as real. They were the ones comfortable being characters.
From novelty to industry
By the early 2020s, the rise of virtual influencers had turned from curiosity into a working commercial category. Fashion houses, beauty brands, and tech companies began casting synthetic faces for campaigns, drawn by creative control and predictable schedules.
The behind-the-scenes reality also came into focus. A “single” top persona is usually run by a team: writers, 3D artists, photographers who shoot real background plates, and social managers who keep the character consistent across posts. The lone-creator myth rarely survived contact with the biggest accounts.
Brands then faced a practical question: build a proprietary character or partner with an existing one. A 2024 study of 33 virtual influencers on Instagram found that non-branded virtual influencers tended to generate more engagement than brand-owned ones, and suggested most companies are better off collaborating than building from scratch. The market, it turned out, could tell the difference between a character with its own life and a mascot with a logo attached.
The economics helped the category grow. A well-run persona can appear in campaigns worldwide at once, never ages out of a look, and carries no risk of an off-brand moment caught on camera. Those qualities made virtual influencers attractive to marketers well before anyone had settled the harder questions about disclosure, labeling, and how honest a synthetic character owes it to be. The commercial rise, in other words, ran ahead of the norms, and much of the last few years has been spent letting the rules catch up.
What the rise was actually about
Strip away the campaigns and a quieter pattern remains. People followed these characters for the same reasons they follow anyone: story, continuity, and the feeling of being spoken to. A VTuber going live for three hours is offering presence. A serialized persona with recurring friends and small dramas is offering a world to check in on.
That is the thread connecting Miku’s fan-made concerts to today’s best known virtual influencers. The technology changed; the underlying want did not. Some of that want is companionship, the low-stakes comfort of a familiar voice that shows up on schedule.
It is also where a related category sits. An AI conversational companion application is a different thing from a public persona you scroll past: it is a private space to talk, listen, and practice connection rather than a broadcast to watch. The history of virtual influencers helps explain why such tools resonate. People were never only reacting to pixels. They were responding to the sense of someone, or something, paying attention.
Where the story goes next
The next chapter is being written around interactivity. Early virtual influencers mostly broadcast: a post went out, followers reacted. Newer formats let audiences talk back in real time, and the line between a character you watch and a character you converse with keeps thinning.
If the first two decades were about proving synthetic personas could hold an audience at all, the coming years are about how responsive and reciprocal that relationship can feel. The rise, in that sense, is not finished. It is changing shape.
Frequently asked questions
Who was the first virtual influencer?
It depends on the definition. Japan’s Kyoko Date, marketed as a digital idol in the mid-1990s, is an early example. Hatsune Miku (2007) popularized the fan-driven virtual star, and Lil Miquela (2016) is often credited as the first modern Instagram virtual influencer.
When did virtual influencers become popular?
The category moved from niche to mainstream around 2016 to 2018, when Kizuna AI launched the VTuber format and Lil Miquela reached millions of followers and mainstream press coverage.
Why do people follow characters they know are not real?
Research suggests audiences respond to story, consistency, and the sense of being addressed, not to realism alone. Openly stylized characters often engage audiences as well as or better than photorealistic ones.
Are virtual influencers the same as AI companions?
No. A virtual influencer is a public persona you follow in a feed. An AI conversational companion application is a private chat experience built around listening and conversation. They share roots but serve different needs.