Hey! Ankur here, and this is the 21st edition of Lazy AI — 5 minutes to stay ahead of the AI curve.
Today’s piece is about something you’ve almost certainly been scrolling past without noticing. Or may be you have.
But once you see the pattern, you won’t be able to unsee it.
Let’s play a small game
Check out these 3 Instagram profiles, and tell me what’s common between all of them.
Here’s what’s thing: None of these people are real. They’re all AI Influencers, created by humans.
Take Naina, for example. She’s a 22-year-old from Jhansi who moved to Mumbai to become an actor. She has over 350,000 Instagram followers and brand deals with Nykaa, Puma, and Pepsi. Her Instagram handle is naina_avtr — go look, I’ll wait.
She’s not real. She’s a digital avatar built by a Mumbai company called Avtr Meta Labs, run by co-founder and CEO Abhishek Razdan. Razdan speaks at conferences about Naina. Naina “speaks” through the team he built.
Another one — Kyra (@kyraonig) has 260,000 followers. Created by George Tharian and Himanshu Goel at FUTR Studios. She’s walked a virtual MET Gala and been featured on the cover of Travel+Leisure India. Appeared on Shark Tank.
These aren’t rare experiments. According to Market.us, 58% of internet users now follow at least one virtual influencer (I’m part of this 58% btw), and Ogilvy projected that CMOs would allocate 30% of their influencer budgets to virtual creators by 2026 — a roughly $10 billion reallocation from human creators.
You don’t need to be an engineer to talk AI. Subscribe to this newsletter — every other day, you’ll understand AI well enough to be the smartest person in the room at work.
Each issue is just 5 minutes — less than the time you spend doomscrolling before bed. Except, this actually moves your career forward. Join 8,000+ subscribers now!
Here’s where it gets interesting.
There isn’t one type of AI influencer. There are 3, and they work very differently.
The brand-backed avatar. Naina, Kyra, Miquela. These are (mostly) publicly disclosed — as “AI accounts'“, “virtual influencers” or something similar. They’re essentially honest marketing vehicles built by agencies that sell them to brands as “influencers who never cancel, never age, never go off-brand.” A Gartner survey found replacing human influencers with virtual ones can cut campaign costs by roughly 30% — no travel, no talent fees, no contract renegotiation every six months. For a CMO, that’s a very easy line on a budget sheet.
The niche persona. Fitness coaches, travel bloggers, tech reviewers, “study with me” accounts. They’re often undisclosed. Run by small operators who found that generating a consistent face and voice is cheaper than hiring one. You’ll see these a lot in affiliate-heavy niches where trust is built through repetition rather than personality.
The synthetic creator. This is where it gets cynical. Operators running AI personas primarily for subscription revenue — often on platforms like Fanvue, which now makes 15% of its $100M annual revenue from AI creators. Almost always undisclosed. Almost always a man behind the account. Almost always a “woman” in the front.In other cases, they’re just Instagram accounts that run paid subscriptions. Some of them have thousands of subscribers paying them 399/month. You do the math.
All three types share one workflow:
Generate the character
write a backstory
post 3-5 times a week
drive traffic and monetise
What differs is the monetisation — brand deals, affiliate links, or subscriptions.
What’s common is the deception gradient: from “disclosed novelty” at one end to “outright fraud” at the other.
Should you be worried about AI influencers?
Well, people have two extreme opinions about AI influencers — that they’re either futuristic, or they’re dystopian. But I don’t think that opinion should live at two ends of the spectrum.
See, I’m not saying AI influencers are bad. Kyra doing a boAt campaign isn’t the problem. The problem is everything downstream — the undisclosed fake profiles. The synthetic creators selling fake intimacy.
Most people think AI will replace human creators. I think it won’t. I think it’s building a parallel economy where the operator makes the money — not the creator, not the AI. Same thing that happened with dropshipping in 2017 and content farms in 2020.
So how do you identify a fake AI profile?
There are signs. They’re not perfect, but you’ll be fairly sure.
Skin that’s too even. Real skin has pores, shine, asymmetry. AI skin is too perfect by default — even in “candid” shots. Even if it has pores, they seem “designed”
Static or mismatched backgrounds. Outdoor shots where the light on her face doesn’t match the light in the background. Same “café” that subtly changes between posts
Hands, ears, teeth. Generative models still struggle with these. Count fingers. Check if earrings match. Look at how teeth sit.
Video is rare or cut short. This is the biggest tell — Most AI influencers generally post only photos. Very few videos. And the motion and micro-expressions in the video can be identified if you look closely.
Comments feel off. Either no replies from the account at all, or generic one-liners. Real creators have idiosyncratic voices.
Reverse image search returns nothing. A real person with 300k followers has a trail — a tagged photo, a YouTube guest spot, a wedding album from 2019. AI influencers are born on Instagram.
The more of these you see, the more convinced you should be that it’s AI.
You don’t need to be an engineer to talk AI. Subscribe to this newsletter — every other day, you’ll understand AI well enough to be the smartest person in the room at work.
Each issue is just 5 minutes — less than the time you spend doomscrolling before bed. Except, this actually moves your career forward. Join 8,000+ subscribers now!
So what should you take from this?
Two things.
One — the question is no longer if your feed is part-synthetic. It is.
The question is what percentage, and whether you can tell which accounts are which. That’s a skill now, the same way spotting Photoshopped magazine covers became a skill in the 2000s.
Two, and this is the bigger lesson: Every time a new tech makes content cheap to produce, the winners aren’t the creators or the engineers. They’re the operators — the ones who understood the distribution funnel first. Naina is not winning because AI is clever. She’s winning because Abhishek Razdan built a company around the funnel before most brands knew the category existed.
Next time you see a shiny new “AI is going to replace ABC” headline, ask a better question:
Who’s already running the new version of ABC, and what’s their funnel?
That’s usually where the game is.
If you found this useful, please share it with a friend so that they don’t get fooled (or worse, scammed) by an AI.
I’ll see you next time..
Cheers,
Ankur