$2.4 Billion Is Flowing to People Who Don't Exist

A while back, I came across an account.

2.3 million followers. Updates every single day. A comment section buzzing with conversation. Brand deals landing one after another — ads alone were bringing in a few thousand dollars a month.

I tapped into the profile, curious to see which creator was this prolific.

After scrolling through it, I realized something: this creator does not exist.

The face was drawn. The body was computed. Even the details of her persona — what she likes, what she fears — were generated.

God. My first reaction was: who, exactly, have all those people in the comments been talking to?

Only later did I learn that this is far from an isolated case. It's a gold rush in full swing — and its scale is probably much bigger than you think.

How This Craze Caught Fire

First, sixty seconds of history.

AI influencers have actually been around since 2016. That year, a startup called Brud created a virtual idol named Lil Miquela, who racked up millions of followers — Prada and Calvin Klein both came to her for collaborations.

Sounds lovely. But back then, making one of these meant supporting an entire team behind the scenes: 3D artists, scriptwriters, campaign planners — you couldn't do without a single one. Expensive. Slow. Nobody could produce a second one.

The real change came at the beginning of this year.

Generative AI tools matured to the point where one person, with one computer, can build a virtual persona with a consistent appearance, publish content every day, and run several accounts at the same time. The technical barrier? It collapsed.

Collapsed how far? In January, search volume for "how to make an AI influencer" jumped 340%. Tutorials on YouTube teaching people to build AI influencers rack up millions of views. Venture capital piled in. A whole wave of ordinary people took this up as a side hustle — one person operating four or five "people," like running an account farm.

Not curiosity. Profit.

So let's do the math on exactly how this money gets chased.

What's in It for Brands

Think about it: when a brand pays a human influencer for placements, what exactly is it paying for?

A mid-tier influencer with around 500,000 followers quotes $5,000 to $15,000 for a single sponsored post. What the brand is buying is attention.

But the brand is simultaneously paying for something else: uncertainty.

Human influencers get sick, flame out in scandals, demand raises, switch agencies, or simply stop posting. Every one of those is a risk for the brand.

AI influencers? They work around the clock, throw no tantrums, never demand a raise, and will never fire off a controversial post in the middle of the night. And research from Harvard Business Review gave an even more ruthless number: once the persona is built, the cost of producing the content that follows approaches zero.

Approaches zero. What do those two words mean for a brand?

Now compare another set of quotes. A human influencer with over a million followers can charge $250,000 or more for a single post. The company behind Lil Miquela openly quotes $9,000.

$250,000 on one side. $9,000 on the other.

Which do you think the brand picks?

Think Nobody's Watching? The Numbers Are Scary

At this point, someone is bound to say: they're fake people — who would follow them?

Plenty of people do. And at a level that surprised even me.

Harvard Business Review analyzed more than one million posts published between 2014 and 2020, from 551 human accounts and 13 virtual ones. The result: for the virtual accounts, sponsored posts had engagement rates 13.3% higher than their everyday content. For the human accounts, the opposite — sponsored posts underperformed their everyday content by 2.1%.

In fashion and beauty the gap is even starker: virtual accounts' sponsored posts beat their regular content by 16.3%, while humans' fell 2.3% short.

Why is that? My read: when fans see a human they follow take an ad, something in them mutters, "you've changed." When an account that was manufactured from the start takes an ad, there's no expectation gap to violate.

Then there's a data point that cuts even deeper. The same study noted in passing the racial makeup of those 551 human influencers: 68% were white. Virtual personas don't carry that baggage. British fashion house Ralph & Russo created a virtual model named Hauli — the name comes from the Swahili word for "strength" — and shot her at landmarks like the Taj Mahal and the Great Wall. That campaign pulled in 19.4 million impressions worldwide, with media value estimated at $65.1 million.

From there, the direction of the money becomes clear. Influencer Marketing Hub's report predicts that by the end of 2026, AI-generated influencer content will take 12% of the entire influencer marketing budget.

That works out to roughly $2.4 billion.

$2.4 billion, flowing to people who don't exist.

The platforms have switched the light to green, too. At the end of last year, Instagram quietly revised its creator rules: as long as the account is clearly labeled, AI personas are officially allowed. TikTok followed at the beginning of this year. With the gray zone gone, large agencies and big brands stopped watching from the sidelines. In just the first quarter of this year, startups building AI influencer infrastructure raised more than $300 million, and platforms from companies like Synthesia and D-ID can get a virtual persona that's ready to take brand deals live within an hour.

But the Bubble Has Started to Leak

Every boom has its flip side.

Around the middle of this year, the wind changed. Discussions on Reddit and X turned collectively skeptical: Is this deception? Will beauty standards get warped? Is the last shred of "real" in digital space gone too?

One scholar said something that has stayed with me. Safiya Noble, a professor at UCLA and author of Algorithms of Oppression, put it this way: AI influencers are the logical endpoint of a system that "always puts aesthetics ahead of humanity." "We've removed the human entirely, yet we still call it influence."

Harsh words. But there's a case that all but makes her argument for her.

An agency in Barcelona built a virtual model named Aitana. A journalist asked them point-blank: you've sculpted a near-perfect body, and some followers say "I really wish I had that body" — is that not a problem?

Their answer was disarmingly honest: early on, they had tried making models with more diverse body types, and clients didn't like them. "Now look at her next to the other influencers — she looks exactly like everyone else."

Stings, doesn't it? Even the opening for "diversity" has been sealed shut by client budgets.

And there are two more troublesome things.

The first is disclosure. Platform rules require AI accounts to label themselves, but enforcement is a shambles. The Oxford Internet Institute did a sampling study: 68% of AI influencer accounts do not clearly state in their bio that they are synthetic. In other words, the "person" you think you're interacting with is most likely a program.

The second is stolen content. Some people pair AI avatars with stolen scripts, reciting real creators' copy word for word to ride on other people's research and grow an account from scratch. Once these fake accounts have built up trust, they can be mobilized at any moment to whip up sentiment and steer public opinion. Think one layer deeper — this already goes beyond business.

The bill came due fast. Some AI influencers, once unmasked, lost tens of thousands of followers overnight. More broadly, as the accounts multiply, engagement rates across the entire category have started to slide. A market that looked boundless in January already felt crowded by June.

Trust, once shattered, stays shattered — even if you're a synthetic persona.

What Comes Next

My judgment comes down to one word: shakeout.

Right now, thousands of low-cost AI influencer accounts are fighting over attention, and most of them will die. The survivors will have talent agencies behind them, genuine creative control, and something to offer beyond the usual lifestyle content. In fact, top agencies like CAA and WME are already quietly signing AI influencers. These are not hobbyists working from their bedrooms — they're full-fledged digital talents, arriving with brand partnerships, merchandise, and media exposure.

Human-machine hybrid models are emerging too. Some creators pair their account with an AI partner and split the content work; others use AI to clone themselves, one real person splitting into several selves, speaking different languages on different platforms — without burning themselves out.

Regulation is on the way as well. The EU's AI Act will take effect in phases between 2026 and 2027, requiring synthetic content in commercial contexts to be clearly labeled. In the US, the Federal Trade Commission has this space in its sights, focused squarely on disclosure and false advertising.

But I think the biggest change will be cultural.

We used to call an account "real" and mean there was a living human behind it. Going forward, the definition may become: does it honestly tell you what it is?

I'm not just guessing. Research cited by Harvard Business Review shows that 52% of American social media users already follow at least one virtual influencer — and the share is even higher worldwide.

The question was never whether AI influencers will exist. The question is, in a market where attention is already crowded to the point of overflowing, who — or what — we're willing to give it to.

One Last Thing

Back to that account with 2.3 million followers.

As I write this, she's probably still posting. The comments are probably still buzzing. The brands are probably still lining up.

It's just that from now on, every time you tap open a polished-looking account, a flicker of hesitation may cross your mind — an extra half second:

Is there actually anyone on the other side of the screen?

That question is worth $2.4 billion.

And maybe far more than that.