Brands are paying for faces that do not exist. See how people build an AI influencer, how it earns, and how you could launch one this month.
This in-depth guide covers everything you need to know about ai influencer: what it is, how it earns, and the rules (2026). Based on verified income data and real-world case studies from our database of 138 side hustle tactics.
An AI influencer is a computer-generated persona that posts on social media as if it were a person, while a real person or company writes its captions, makes its images and videos, and signs its brand deals. Some are disclosed, which means the audience is told the face is synthetic. Others are undisclosed, and those present a made-up character as a real human being. Whichever kind you are looking at, a human team decides everything it says and keeps the money it earns.
Maybe you saw one on your feed last night. A flawless face, a Paris café, a caption about a skincare brand, and thousands of comments from people who never asked whether she was real. Then you looked at your own phone, at a job that pays what it paid three years ago, and wondered who collects the money behind that face.
That question is fair. AI tools now make images, voices and video that used to need a studio, and the same tools are reaching ordinary work like copywriting, product photos and customer service. Some people will learn how this machinery works and get paid for knowing it. Others will keep scrolling past it while the jobs around them shift.
The easy early days are already fading. Platforms have rewritten their labelling rules, brands have grown pickier, and every month more accounts crowd the same feeds. Learning the rules and the real numbers now costs you a few evenings. Learning them after your account gets flagged costs a lot more.
So here is what is actually true, one piece at a time.
Below you will find the famous names with their real numbers, how these personas earn, how people build them, the rules that now apply, and an honest view of whether one is the best use of your evenings.
AI influencers at a glance
| |
|---|
| What it is | A synthetic persona that posts, sells and takes brand deals as if it were a person |
| Who runs it | A person, a small studio or a funded company. The persona has no will of its own |
| Two kinds | Disclosed (the audience is told it is AI) and undisclosed (presented as a real human) |
| Best-known example | Lil Miquela, on Instagram since April 2016 |
| How they earn | Brand deals, platform ad payouts, subscriptions, their own products, affiliate links |
| Published earnings | Rare. Most figures in the press are self-reported by the agency that owns the persona |
| Main rules | YouTube, Meta and TikTok AI labels, the US FTC Endorsement Guides and 2024 fake-reviews rule, the EU AI Act's deepfake disclosure duty |
| Worked example on this site | Elias Yoder, the AI "Amish farmer" with 366K YouTube subscribers |
What counts as an AI influencer
Picture a young woman posting gym selfies from a sunny café. She has a favourite protein powder, a running joke with her followers and a new outfit every week. Now imagine every detail of that post was chosen by a team, and the woman does not exist. That is an AI influencer.
The term covers a wider range than most articles admit, so it helps to split it two ways.
By how it is made. The older generation, often called virtual influencers, were built with 3D modelling and computer graphics, the same craft as an animated film. Lil Miquela and Imma started this way. The newer generation is made with generative AI: image models produce the photos, video models make the face move, and voice models read the script. Today most new personas are a mix of both, and the public uses "AI influencer" for all of them.
By how honest it is. This split matters far more to you.
- A disclosed persona tells you what it is. The bio says it is virtual, the posts carry the platform's AI label, and the operator talks openly to the press. The audience follows a character the way it follows a cartoon mascot, knowing what it is.
- An undisclosed persona presents itself as a real human being. It claims a hometown, a family and life experience, and nothing on the account says software made it. The audience trusts a person who is not there.
Same technology, very different businesses. The second one depends on the audience never finding out, and as you will see below, audiences do find out.
When you scroll past a face this week, would you know which kind you were looking at? Most people would not, which is the reason platforms and regulators now require labels.
Famous AI influencers, and what their numbers really say
Follower counts and earnings for AI influencers get rounded up as they travel. Below, every figure carries its source and date, and numbers that come only from the persona's owner are marked self-reported.
Lil Miquela (United States)
Miquela is the persona most people mean by "AI influencer", though she began as computer graphics. Her first Instagram post went up on 23 April 2016, and for two years her makers stayed in the background. In April 2018 a staged "hack" storyline ended with the Los Angeles startup Brud revealed as the company behind her (Wikipedia summary of the record).
The milestones are well documented:
- June 2018. Time named her one of the 25 Most Influential People on the Internet, on a list that also included Kylie Jenner.
- 2018 and 2019. Brand work included an Instagram takeover for Prada during Milan Fashion Week and a Calvin Klein advert with Bella Hadid that drew criticism (Wikipedia).
- 4 October 2021. Dapper Labs, the company behind NBA Top Shot, bought Brud and took on its 32-person team. TechCrunch reported that Brud had earned a $125 million valuation and that the price was not disclosed.
- May 2025. Miquela ran a sponsored storyline for NMDP, the US bone marrow donor registry, in which she was "diagnosed" with leukemia. People reported (16 October 2025) that many commenters called it disrespectful to real patients, and put her audience at about 5 million followers across Instagram and TikTok.
Her owners have never published an earnings figure. What the record does show is a funded company with a 32-person team behind one character.
Aitana López (Spain)
Aitana is the example most often quoted for money. She was created by The Clueless, a Barcelona modelling agency. Its founder, Rubén Cruz, told Euronews (published November 2023, updated December 2024) that the agency built her after brand projects kept being put on hold, so that it would "not be dependent on other people who have egos".
The money claims, all from the agency itself and checked by nobody else (self-reported):
- just over €1,000 per advert;
- around €3,000 a month on average;
- up to €10,000 a month at best.
The same Euronews piece says Aitana had more than 343,000 Instagram followers, became the face of the sports supplement brand Big, and posts lingerie photos on Fanvue, a subscription platform similar to OnlyFans. The agency talks openly about her being AI, which puts her in the disclosed camp in its press dealings.
Look at the gap between the headline and the detail. "Up to €10,000" is the number that travels. "Around €3,000 on average" is in the same article and gets far less attention. Which of those two numbers would you plan your rent around?
Imma (Japan)
Imma is a pink-bobbed virtual model from Tokyo, produced by Aww Inc., which describes itself on its own site as "Japan's 1st Virtual Human Company" and calls Imma "the most popular Virtual Influencer/Virtual Model in Asia" (self-described).
Her best-known job shows what brands buy from a virtual human. In 2020 IKEA Japan built a two-floor apartment inside its new Harajuku store and placed Imma "in" it on LED screens for a 72-hour livestream, where passers-by could watch her do yoga and tidy up (Creative Review; Domus, 1 October 2020). IKEA was buying a spectacle that a production studio could deliver, more than a follower count.
In early January 2025 Meta removed a set of AI character accounts it had launched as a test in 2023, after users resurfaced one called "Liv", described as a "Proud Black queer momma", and criticised the accounts as "creepy and unnecessary" (NBC News). Meta said the accounts were managed by humans and part of an early experiment. If Meta misjudged how people feel about synthetic personas claiming an identity, a solo operator can too.
Elias Yoder (the undisclosed case)
The fullest worked example on this site is our Elias Yoder case study, summarised in the next section.
The numbers side by side
| Persona | Audience (source, date) | Earnings (source) | Disclosed? |
|---|
| Lil Miquela | About 5M across Instagram and TikTok (People, Oct 2025) | None published | Owner public since 2018 |
| Aitana López | 343,000+ Instagram (Euronews, updated Dec 2024) | ~€3,000 average, up to €10,000 a month (self-reported) | Agency open in press |
| Imma | Not taken from a dated source | None published | Owner public |
| Elias Yoder | 366K YouTube subscribers, 18.9M views (channel, 9 Oct 2026) | "Over $100k" first month (operator, self-reported) | No disclosure on channel |
Elias Yoder: what an undisclosed AI persona looks like up close
Elias Yoder presents the YouTube channel @EliasYoderAmish as an Amish farmer from Lancaster County who keeps bees and teaches old household thrift. On 9 October 2026 the channel had 366K subscribers, 18,923,725 views and 189 videos, almost all published since 17 May 2026. The videos say "My name is Elias Yoder. I am Amish." Nothing on the channel or the shop mentions AI.
The public evidence says he is synthetic. The YouTuber Pleasant Green's video Don't Trust the AI Amish (24 September 2026) showed glitching mouth movements in generated talking clips of about five seconds, a "wife" whose face changes between appearances, and a shop whose own code labels its purchase pop-ups as illustrative. Our own check of the sales page found a code comment above the three customer reviews reading "PLACEHOLDER TESTIMONIALS, illustrative only. These are NOT real reader quotes."
The business is simple. Long videos build trust in a face, and every description links to a $47 e-book, "The Amish Home Savings Manual", sold through Gumroad with 214 ratings averaging 4.9. If every rating came from one full-price, unrefunded sale, that is a floor of $10,058 in gross sales. The operator, Ranga, has since claimed the channel and says it made "over $100k in his first month", a figure nobody has checked.
Then there is the name. A real man called Eli Yoder, an ex-Amish preacher and truck driver from Ohio, runs his own books and channel, and his official site now carries a disclaimer distancing him from an "AI-generated" lookalike channel. A borrowed culture and a borrowed surname pulled a real person into someone else's business.
The full case study walks through the evidence, the funnel step by step, who might be behind it and what an honest channel can copy. Read it before you build anything, because it shows both how fast the format can grow and exactly where it crosses lines. If you had been one of those 366K subscribers, how would you feel reading that code comment about the reviews? Your answer is a fair guide to how your own audience would react.
How AI influencers make money
Five income lines turn up again and again. Here is each one, with the published numbers where they exist.
1. Brand deals and sponsored posts
A brand pays for a post, a campaign or a face. Aitana's agency reports just over €1,000 per advert (self-reported, Euronews). Miquela's record includes Prada, Calvin Klein and a paid storyline for NMDP. Imma's IKEA job was a full installation.
These deals went to personas with hundreds of thousands or millions of followers and a professional team. And a sponsored post from an AI persona still has to say it is sponsored, as the rules section explains.
Ad revenue sharing pays a persona the same way it pays any creator. On YouTube, a channel can apply to the Partner Program with 1,000 subscribers plus either 4,000 watch hours in the last 12 months or 10 million Shorts views in the last 90 days (YouTube Help). Once in, YouTube pays 55% of net ad revenue on long-form videos and 45% of the Shorts revenue allocated to the creator (YouTube Help).
TikTok's Creator Rewards Program has its own gates: 10,000 followers, 100,000 views in the last 30 days and an age of 18 or over, as the site's TikTok Creativity Program guide explains, with a planning figure of $500 a month at the low end.
Platform payouts come with a catch for synthetic channels. YouTube's monetization policy was reworded on 15 July 2025 to target "inauthentic", mass-produced and repetitive videos (TechCrunch). An AI persona churning out near-identical videos from one template is exactly what that wording describes.
3. Subscriptions
Some personas sell paid access on subscription platforms. Aitana posts on Fanvue, according to Euronews. Most "AI model earns thousands" stories come from this corner, and it is also where impersonation, adult content and age rules bite hardest. Read the creator terms first.
4. Your own products
A persona can sell an e-book, a course, a preset pack or a planner. Elias Yoder's whole funnel points at one $47 PDF. The margin is high because nothing is shipped, What would your persona sell that a follower could not get free from its posts? If you have an answer, the site's guides to Etsy digital products and digital planners cover making and pricing it.
5. Affiliate links
A persona can earn a commission when followers buy through its links. An affiliate commission is a material connection, so the same disclosure duty applies. The site's TikTok Shop affiliate guide covers how that programme works.
The sixth line nobody headlines: selling the skill
The people with the most predictable income in this space are often the ones making AI video and persona content for businesses. The site's AI video production guide puts that work at $2,000 to $20,000 a month, with $0 to $1,000 to start and first payment in two to four weeks. Clients pay for the craft whether or not your own persona ever goes viral.
So how much do AI influencers make? Honestly, almost nobody publishes it. The only concrete monthly figures in the public record are self-reported by an agency selling its own persona, and the most-followed AI influencer of all has never released a number.
If your persona ever did land one brand post at the €1,000 Aitana's agency quotes, picture what that single payment would cover: the car insurance renewal that has been sitting on the fridge since spring, paid in one go, with nobody at home asking how. Keep the condition in view, though. That figure came from an agency with an established audience, and it is the reason the rest of this page is about building something that can survive scrutiny.
Notice which line on this list pays most reliably: the skill underneath the persona. Someone who can keep a face consistent, write a caption that sounds human and label it properly is useful to a café owner, a dental clinic or a small brand with no budget for a photo shoot. That client can pay this month, while a persona is still chasing its first thousand followers.
How people build an AI influencer
The operators behind Elias Yoder have never named their tools, and Aitana's agency has described its process only in general terms. What follows is the workflow you can see from the outside and from openly marketed tools. The tools named are widely used examples. This page endorses none of them, and prices change too often to quote here.
Step 1: Design the persona before you make a single image
Every working persona starts on paper. Write down:
- Name and look. Pick a name you have checked against real people, especially if it belongs to a real culture or community. The Elias Yoder story shows what happens when a name collides with a real man's.
- Niche. Fitness, cooking, gaming, home repair, finance. The niche decides who the brands are.
- Voice. How the persona talks: funny, calm, nerdy or blunt.
- Backstory, and how far it goes. A disclosed persona can have a fictional life the way a cartoon does. Write it so that nobody could mistake it for a claim about a real human.
- The disclosure line. Decide now, in one sentence, how the persona will tell people it is AI. Put that sentence in the bio.
Which niche could you make 50 posts about without running dry? That question matters more than which image model you use, and it is the same question the site's faceless YouTube guide asks of any channel.
Step 2: Generate the images
Image models make the stills: profile photos, outfits, scenes. Widely used options include Midjourney, open models in the Stable Diffusion family, and Flux from Black Forest Labs. Read each one's terms on commercial use and on who owns the outputs before you build a business on them. The site's AI art print on demand guide has a section on exactly that question.
Step 3: Keep the face consistent
Consistency is the hardest technical part. A generated face drifts from image to image: the jawline shifts, the eye colour changes, the "wife" looks different from one mention to the next, which is one of the tells Pleasant Green spotted in the Elias videos. Operators handle this with reference images, character features built into some generators, or by training a small custom model (often called a LoRA) on a set of approved images of the persona. Keep a reference sheet of approved images and check every new one against it.
Step 4: Make it move and speak
For video, there are two broad routes. Avatar platforms such as HeyGen and Synthesia turn a script into a talking presenter, and they are built for presenters that are openly synthetic. General video models, such as Higgsfield (affiliate link), animate a still image into a short clip. Voice tools such as ElevenLabs (affiliate link) produce a consistent voice from text. The site's AI video production guide covers how avatar tools like Synthesia and HeyGen are used for paid client work.
Never clone a real person's voice or face without their written permission.
Step 5: Write, edit and post on a schedule
Now the persona is a content business like any other. Someone writes, edits, and answers comments and brand emails, and for a solo operator that someone is you. A realistic weekly rhythm for a beginner is three to five posts, one longer video and a batch of short clips cut from it. The site's clip channel guide explains how to turn one long video into several short ones.
How many hours a week could you give this, honestly, on a week when work runs late and the house needs you? Write the number down. A persona that posts for three weeks and then goes quiet earns nothing.
Step 6: Label everything
Use the platform's AI disclosure every time a post shows a realistic person or scene. Some tools also embed C2PA "content credentials" metadata in their files, and YouTube says it may add a label automatically when it detects that metadata (YouTube Help). Treat that as a backstop. The disclosure is still your job.
The rules for AI influencers (as of 2026)
Running an AI persona is allowed on every major platform. Hiding what it is, and faking what it sells, is where the rules start. Here is each one with a link to the source.
YouTube: altered or synthetic content
YouTube requires creators to disclose when they "use AI to meaningfully alter or generate photorealistic content" (YouTube Help). A realistic person who does not exist, talking to camera in a place that does not exist, fits that description. Animated or clearly unrealistic content and minor edits such as beauty filters do not need the label, and using AI to write a script or a title does not either.
What happens if you skip it: YouTube says creators who consistently fail to disclose may have a label applied for them, have content removed or be suspended from the YouTube Partner Program. It also says that disclosing "won't limit a video's audience or impact its eligibility to earn money". Which would you rather explain to a sponsor: a small label under your video, or a channel removed from the Partner Program? The label costs nothing.
Meta labels content it detects as AI-generated or that creators disclose. In February 2024 it said it would require people to use its disclosure tool "when they post organic content with a photorealistic video or realistic-sounding audio that was digitally created or altered", and that it "may apply penalties if they fail to do so" (Meta). On 1 July 2024 it renamed its "Made with AI" label to "AI info" after photographers complained that lightly edited real photos were being tagged (Meta; TechCrunch).
TikTok: AI-generated content labels
TikTok's Community Guidelines say it requires creators "to label AI-generated or significantly edited content that shows realistic-looking scenes or people", and that it does not allow AI content that misleads about matters of public importance or harms individuals (TikTok Integrity and Authenticity). TikTok can also add an "AI-generated" label itself.
The FTC Endorsement Guides (United States)
The FTC revised its Endorsement Guides in 2023. The definition of an endorser now covers anyone who could be "or appear to be an individual, group, or institution", and in its notes the Commission said the new language reaches fabricated endorsers and was written to cover more than virtual influencers alone (Federal Register, 26 July 2023). In plain terms: when your AI persona praises a product it was paid or gifted to promote, the same "#ad" disclosure rules apply as for a human creator. The Guides are FTC guidance on how it reads the law against deception, and the FTC can act against deceptive endorsements under that law.
The FTC fake reviews and testimonials rule (United States)
On 14 August 2024 the FTC announced a final rule banning fake reviews and testimonials, including ones that "misrepresent that they are by someone who does not exist, such as AI-generated fake reviews", and allowing civil penalties against knowing violators (FTC). The rule took effect on 21 October 2024 (Federal Register). It also bars selling or buying fake indicators of social media influence, such as followers or views generated by bots.
Read that against the Elias Yoder shop: reviews its own code calls "NOT real reader quotes", and a seller persona presented as a real Amish couple vouching for their own book. Whether a specific page breaks the rule is for a regulator to decide, but you can see why an operator would not want to be the test case.
The EU AI Act: deepfake transparency
In the European Union, Article 50(4) of the AI Act says deployers of an AI system that generates or manipulates image, audio or video content "constituting a deep fake" must disclose that the content has been artificially generated or manipulated (AI Act, Article 50; official text on EUR-Lex). The Act defines a deep fake as AI-generated or manipulated content that "resembles existing persons, objects, places, entities or events and would falsely appear to a person to be authentic or truthful". The disclosure has to be clear and given at the latest at the first exposure. For evidently artistic, creative, satirical or fictional work, the duty shrinks to flagging that generated content exists, in a way that does not spoil the work.
Article 50 applies from 2 August 2026. Breaches of the Article 50 transparency duties can bring fines of up to €15 million or 3% of worldwide annual turnover, whichever is higher (AI Act, Article 99). Whether a given persona counts as a deep fake depends on the facts, and a photorealistic character presented to Europeans as a real human is the case most likely to be caught. If you have followers in the EU, label.
Every platform and regulator here allows an AI persona that tells people what it is, and all of them push back on one that hides it, fakes reviews or borrows a real identity.
Ethics and risk: what the rules do not cover
Several risks land on you as the operator whatever any regulator decides.
Deception has a delayed bill. An undisclosed persona builds trust it cannot keep. Pleasant Green's video on Elias Yoder was watched 829,682 times in about two weeks, according to the case study. When an audience learns the face was invented, the people who bought from it feel the most deceived, and they are the ones who leave reviews and file chargebacks.
Borrowed cultures and borrowed names. Elias Yoder is presented as Amish, a real religious community with real families who share that surname. The real Eli Yoder had to put a disclaimer on his own site. Meta's "Liv" character drew criticism over the identity it claimed, according to NBC News. Before you give a persona an ethnicity, a faith, a disability or a profession, picture a member of that group reading the bio. Could you explain your choice to them face to face? If the answer is no, change the persona.
Advice with nobody behind it. A persona that gives health, money, heating or safety advice still gives advice. Pleasant Green warned that a heating method in the Elias e-book could hurt someone who relied on it in a cold snap. A fictional farmer cannot be held responsible. You can.
Sensitive topics backfire. Miquela's leukemia storyline was for a real charity with a real goal, and commenters still called it disrespectful to patients (People). A synthetic character borrowing a human experience is a hard sell even when the cause is good.
Platform dependence. One review under YouTube's disclosure or inauthentic-content rules can switch off ad income and shop traffic on the same day.
Copycats. Within weeks of Elias Yoder taking off, lookalike channels appeared. Anything built on a borrowed identity can be borrowed again, and you will have no better claim to it than they do.
Picture explaining your persona to your mum over Sunday lunch, including how it makes money. If you would happily show her the bio, you are probably on the right side of every line on this page.
Building your own: an honest assessment
Here is the honest assessment, as plainly as I can put it.
What the evidence supports. AI personas can attract real audiences, brands have paid for them, the tools are widely available, and a disclosed persona is allowed on every major platform.
What the evidence does not support. That a solo beginner can expect income on any particular timeline. The famous names were built by agencies and funded startups, one of them with a 32-person team. The only published monthly earnings figures are self-reported by an agency promoting its own persona. Undisclosed personas, the ones that grew fastest in the public examples, carry the most risk under every rule above.
Who it suits. People who enjoy character design, writing in a voice and steady posting, who are happy for the persona to be openly synthetic, and who can treat the first six months as practice.
Who should skip it. Anyone who needs money this month, anyone planning to pass a fake human off as real, and anyone hoping the persona can give expert advice they could not give in their own name.
Be honest with yourself here: is it the persona you are excited about, or the idea of income without showing your face? If it is the second, there are better-evidenced ways to get there, and they are listed below.
If you recognised yourself in the first group, start soon. Each week more accounts launch with the same tools and the same prompts, and the ones that already have a voice and a steady posting rhythm collect the attention. Early, honest and consistent is a short list, and right now there is still room on it.
A disclosed-persona starter plan (first 30 days)
If you still want to try, here is a plan that keeps you on the right side of the rules from day one.
Week 1: Paper. Write the persona sheet from Step 1: name (checked against real people), niche, voice, backstory and the one-sentence disclosure. Draft the bio with the disclosure in it, for example: "AI-generated character. Made by [your name or studio]." List 30 post ideas.
Week 2: Look. Generate 40 to 60 candidate images, pick 10 to 15 that look like the same person, and save them as your reference sheet. Read the commercial-use terms of every tool you used and save a copy.
Week 3: First posts. Publish five to seven posts on one platform. Turn on the platform's AI label for every realistic image or video. Make one short video with a voice and caption it.
Week 4: Measure. Look at saves, shares and comments more than likes. Write down whether people responded to the character or to the information. Note it, cut what flopped, and plan the next month.
Before any brand deal. Write your own media kit with real numbers from your analytics. Put "#ad" or the platform's paid-partnership tag on every paid post. Never use invented reviews, bought followers or fake "recent purchase" pop-ups. Those are the exact practices the FTC's 2024 rule names.
Could you keep this plan going for 30 days alongside everything else in your week? If yes, those four weeks will teach you more than any course sold on the topic.
Better-evidenced alternatives on this site
If what draws you is the craft (AI images, AI video, presenting without your own face), these guides cover routes where the money has clearer, documented paths. Each figure comes from the guide it links to.
| Route | Income range a month | Start-up cost | Time to profit | Why it might suit you better |
|---|
| AI video production | $2,000 to $20,000 | $0 to $1,000 | 2 to 4 weeks | Clients pay for the skill, with no audience needed |
| Faceless YouTube | $100 to $50,000 | $0 to $500 | 3 to 6 months | Same no-face appeal, your own voice or an openly AI presenter |
| UGC creator | $500 to $20,000 | $0 to $300 | 2 to 4 weeks | Brands pay for content directly, follower count never comes up |
| AI art print on demand | $300 to $5,000 | $0 to $100 | 2 to 6 weeks | Uses the same image tools, sells products instead of a persona |
| Clip channel | $500 to $15,000 | $0 to $200 | 2 to 4 weeks | Editing skill that every persona and creator needs |
These ranges are wide on purpose. Plan around the low end, because the top is what a few people reach after long practice. Which of these five could you start this weekend with the laptop and phone you already own?
The UGC creator guide is worth a special look, because it is close to what Aitana's agency sells: content that brands run in their ads. The difference is that you film it yourself, as yourself, so there is nothing to disclose beyond the usual paid-content rules. The clip channel guide sets the starting point at 15 to 20 clips a month for $500.
If a clip side income like that reached its $500 starting point, think about what it would quietly change: the school trip form signed the same night it comes home, the electric bill paid the day it arrives instead of on the last possible day. Small money, real breathing room, and every dollar of it earned under your own name.
The site's best AI tools for making money post compares the tools, and how to build a personal brand covers the human version of what a persona imitates.
What to avoid
The AI influencer boom has its own set of traps, and most of them target exactly the person reading this page.
- "Make $10K a month with your AI model" courses. The headline figures usually trace back to self-reported agency claims or a single outlier. Ask the seller for their own income proof and where the number came from. The case study found a vendor video titled "6 Figure" whose six-figure estimate was about a different AI presenter.
- Undisclosed personas presented as real people. It is the fastest way to lose an audience, a platform account and possibly a legal argument, all at once.
- Faces or voices that resemble real people. Never generate a persona from a real person's photos, and never clone a voice without written permission.
- Real cultures, faiths and surnames used as costume. Elias Yoder shows the cost to real people who share the name.
- Fake reviews, fake buyers, bought followers. The FTC's 2024 rule names fake testimonials from people who do not exist and fake social media indicators specifically.
- Expert advice from a fictional expert. Health, money, legal, heating and safety advice needs a real, accountable person behind it.
- Mass-produced, templated uploads. YouTube's July 2025 inauthentic-content wording targets them directly.
- "Done-for-you" persona accounts for sale. A bought account carries platform-ban risk and whatever the previous owner did with it.
- Adult subscription content without reading the rules. Age, consent and impersonation rules are strictest on subscription platforms, and breaking them can cost you your account and payouts.
Tonight, before you pay for any tool or course, open a notes app and write one paragraph about your persona: who it is, who it talks to, and the label you will put on every post. Then make three test images on a free plan and check whether the face stays the same. That hour will teach you more than any sales video, and it costs nothing.
Your first step today
Open the AI influencer you have seen most often in your own feed, or open the Elias Yoder case study if none comes to mind. Spend 20 minutes noting on paper whether it discloses that it is AI and where, what it sells, who its brands are, and whether its audience would stay if the bio said "AI-generated" in the first line.
Then write your own persona's one-sentence disclosure before you write anything else about it. If you cannot write that sentence comfortably, take it as your answer, and start with the AI video production or UGC creator guide instead. If you can, you already have the foundation most AI influencers skipped.