I remember when Lil Miquela first showed up on Instagram. She’d pop up in my feed every so often, that distinctive blunt fringe, the freckles, the slightly uncanny vibe. Let’s be honest, she was obviously fake. Everyone knew it. The collaborations got announced with fanfare, there were press releases, it was a novelty act.
That was the early days of virtual influencers, and honestly, they were easy to dismiss.
But something changed. Recently I caught myself scrolling past what I thought was just another travel influencer posting from some beautiful restaurant, and it took me a second to realize I was looking at a completely synthetic person. The technology has gotten that good. Characters like Emily Pellegrini and Aitana Lopez don’t look like obvious CGI anymore. They look like that well-traveled friend from college you lost touch with, the one who’s always posting from nice places.
This is where we’ve arrived with AI-generated content, and it bothers me more than the obvious slop does.
The Numbers Don’t Add Up, And That’s The Point
Here’s what’s crazy. Platforms won’t tell you how many of the accounts in your feed are fake. They’ll publish transparency reports about everything except this. Virtual Humans database tracks a few hundred popular AI avatars, but that’s just the tip of the needle. Below them is an ocean of accounts flying completely under the radar.
And here’s the thing: it’s never been easier to make them. What used to require studios, money, and serious coordination now just needs a subscription to the right tools. Google and OpenAI have their mainstream products, but companies like Higgsfield, HeyGen, and ElevenLabs have made the barrier to entry so low that almost anyone can spin up a fake person. A fake person who posts regularly, engages with comments, and builds what looks like a real following.
Market research types are estimating the virtual influencer market could be worth over sixty billion dollars by 2030. We’re going from around twelve billion this year to sixty billion in a few years. That’s not a niche anymore. That’s an industry.
The business side of this has gotten weird too. There are AI influencer beauty pageants now. Awards. Talent agencies representing synthetic creators. And the real money seems to be in selling the shovels: courses teaching people how to build their own AI influencers, promising faceless passive income. It has that faint pyramidal smell to it, you know?
Why Platforms Don’t Care (And Why That Matters)
I’ve been thinking about why platforms tolerate this, and I think the answer is uncomfortable.
Engagement is engagement. A fake account posting content keeps users scrolling just as well as a real one does. The metrics look the same. The ad dollars flow the same direction. So there’s no real Internal link to the problem from a business perspective.
Sure, all the major platforms now have policies about synthetic media. YouTube, TikTok, Instagram, they all have rules requiring labels for AI-generated content. But here’s what those rules miss: they’re focused on individual posts, not the accounts behind them. An AI person who follows all the labeling requirements while building a fake following that looks real? They can exist in this grey area forever.
The policies were mostly built for deepfakes and obviously manipulated media, not for synthetic personas designed to blend in. When an AI influencer posts something with the right disclosure attached, what exactly should happen? It’s not impersonating anyone specific. It’s not necessarily a scam. It’s just pretending to be a person when it isn’t one.
The Technical Mess We’re In
From a developer standpoint, the tools keep getting better and that’s both exciting and terrifying.
Still images are already at the point where distinguishing them from real photos at a glance is genuinely hard, especially when they’re mixed in with heavily filtered, professionally shot content from actual influencers. The gap that used to exist, that obvious uncanny valley, has closed considerably.
Video and audio are catching up fast. We’re not quite at the point where a moving, speaking AI person is indistinguishable from a real human in conversation, but we’re getting closer. And once that gap closes, the problem becomes something else entirely.
We’ve built tools that can generate infinite content, and we’ve dropped them into platforms designed around human social interaction. That’s a fundamental mismatch, and I don’t think we’ve thought through the implications.
What Happens When The Pool Dries Up
Here’s the part that keeps me up at night. If so many AI influencers exist just to extract money from real human users, what happens when those humans get fed up and leave?
Social media requires actual people to function. That’s the obvious thing we’re all ignoring. If networks collapse under the weight of synthetic accounts because humans bail, the whole ecosystem falls apart. It becomes bots talking to bots, draining resources from server farms while real users go find somewhere else to be.
We’re already seeing early signs of this. People want AI-free spaces now. The backlash against deepfakes and impersonation is building. Regulators in Europe are starting to pay attention with their AI transparency rules, but even those focus on content disclosure rather than whether an account represents a real person.
The burden has fallen on users to spot and report suspicious profiles, which is a terrible solution for something specifically designed to pass as genuine.
I think the reckoning is coming, it’s just a question of whether it comes from platforms waking up or from users building their own spaces away from this mess. Maybe both. Either way, the version of social media we’ve known is probably numbered.