An AI influencer is a computer-generated persona with its own social media accounts that posts content and promotes products like a human creator. Brands use them in three main ways: paying an existing virtual character to post sponsored content, building their own virtual ambassador, or using AI avatars to make UGC-style ads. Whichever you pick, a paid post must be disclosed, and platforms now expect labels on realistic AI content.
This guide covers what AI influencers are, a few well-documented examples, how brands use them, the trade-offs, the disclosure rules, and how to start. It also covers using AI to find human influencers, which is often the safer choice.
What is an AI influencer?
“AI influencer” and “virtual influencer” are used for the same idea: a fictional character built with computer graphics, AI image and video tools, or both. The character has a name, a look and a voice. It posts on Instagram, TikTok or YouTube and may work with brands.
The term hides a wide range of setups:
- A designed character run by people. A team writes the captions, sets the story and approves every image. AI or 3D tools may help produce the visuals, but the personality is directed by humans.
- An AI-generated model. Images and video are produced with generative tools from prompts. This is cheaper to start and easier to scale, but it is harder to keep the face and style consistent.
- A digital twin of a real person. A real creator or employee is recorded once and an AI avatar speaks new scripts in their likeness. That isn’t a fictional character, but it often gets lumped into the same category.
The point to remember is that there is always a company or person behind the account. The character does not make decisions, and the brand paying for the post is still responsible for what it says.
Well-known virtual influencers
You will see a lot of “top virtual influencer” lists with follower counts that are out of date within months. Two examples have solid, citable coverage.
Lil Miquela
TIME put Lil Miquela on its 2018 list of the 25 most influential people on the internet. TIME described her as a virtual avatar “whose origins and purpose are mysterious,” and noted that Prada had used her to promote its animated GIFs on Instagram earlier that year. She was the only virtual character on that list. The same article says she was built by an entity called Cain Intelligence, “which may or may not exist,” before being taken over by the Los Angeles software company Brud, and adds that a publicist said she is no longer managed by Brud. Treat the details of who runs her today as unconfirmed, since TIME’s piece is from 2018.
What this example shows: a virtual character can get mainstream attention and fashion-brand work, even when nobody can point to a human face.
Lu do Magalu
Lu do Magalu is the clearest example of a brand-owned virtual ambassador. Axios reported in June 2026 that Lu started in 2003 as a virtual shopping assistant for Brazilian retailer Magazine Luiza, replacing in-store salespeople as the company moved online. Axios says she built trust over decades by being genuinely useful.
Axios published these follower figures in a chart, dated June 10, 2026 (the chart credits Axios research):
| Platform | Followers (June 10, 2026) |
|---|---|
| 14 million | |
| 9.1 million | |
| TikTok | 7.5 million |
| YouTube | 3 million |
| X | 1.2 million |
Axios also said that outfits she wore in music videos with popular Brazilian artists sold out on Magalu’s app. Magalu has not published earnings for Lu in the sources used here, so we don’t give a revenue figure.
What this example shows: the strongest results came from a character the brand owns, built over many years, that does a real job for customers. It was not built overnight as a campaign.
How brands use AI influencers
Sponsored content from an existing virtual influencer
The brand pays a virtual character’s owner for posts, the same way it would pay a human creator. This is the quickest route, and your risk is tied to someone else’s account. You also depend on their audience, which may be smaller or less engaged than the headline follower count suggests. Ask for audience demographics and past campaign results, as you would with any creator.
A brand-owned virtual ambassador
The Lu do Magalu model: the brand creates and controls the character. Axios framed the appeal as owning “your own trusted brand voice” instead of paying human influencers whose audiences and prices shift. The cost is that you carry everything: design, content calendar, community management and reputation. It suits brands with a long-term content budget and a clear role for the character, such as shopping help or product education.
AI avatars for UGC-style ads
This is where most small teams will start. You use an AI avatar as an on-screen presenter, then test different hooks and scripts without a new shoot each time. HeyGen’s own site describes making product videos, ads and demos, and generating variations from a single script without reshooting. Synthesia lists marketing launches as a use case. (See the AI avatars tag and the UGC tag for more tools.)
Note that this is an ad format, not an influencer with an audience. Viewers see a presenter in an ad, not a creator they follow. It can work well for explainers and testimonials-style scripts, but see the disclosure section: an AI avatar must not be used to fake a real customer’s testimonial.
Pros and cons
| Pros | Cons | |
|---|---|---|
| Control | You decide the look, tone and schedule. No creator goes off-script. | You own all the work and all the mistakes. |
| Cost | Possibly lower per piece once a character or avatar exists. | Upfront design and ongoing production cost. No public benchmark exists. |
| Scale | The same avatar can deliver many script variants. HeyGen and Synthesia both list many languages on their pricing pages. | Low-effort output looks generic and can get ignored. |
| Risk | No scheduling conflicts or personal scandals. | Audiences may react badly to undisclosed or “fake” content. Platform labels and FTC rules apply. |
| Trust | A useful, consistent character can earn trust over time, as Lu has. | Many people trust real people more. Nothing in the sources above measures this, so test it with your audience. |
A fair summary: AI influencers fit brands that want control and consistency. Human creators fit brands that need credibility, real product use and a community that already exists.
Disclosure rules you need to follow
This section is general information, not legal advice. Check with a lawyer for campaigns that carry risk.
FTC: virtual influencers are covered
The FTC’s updated Endorsement Guides, announced in June 2023, changed the definition of “endorsements” to clarify the extent to which it includes fake reviews, virtual influencers and tags in social media. In practice, a virtual influencer who is paid to promote a product is treated like any other endorser.
The FTC’s influencer guidance says disclosures must be easy to notice. They should appear with the endorsement, not buried in a profile or behind a “more” click. For video, put the disclosure in the video itself, not only in the description. Use plain words such as “ad,” “advertisement” or “sponsored,” and avoid vague tags like “collab.” The FTC also says not to assume a platform’s disclosure tool is good enough on its own, and to consider using it in addition to your own clear disclosure.
FTC: fake reviews and fake social proof
The FTC’s final rule on fake reviews and testimonials, announced August 14, 2024, bans creating, selling or buying reviews that misrepresent they are by someone who does not exist, such as AI-generated fake reviews. It also bans buying or selling followers or views generated by a bot or hijacked account when the buyer knew or should have known they were fake and used them to misrepresent influence. For you, this means:
- Do not use an AI avatar to play a “real customer” who never used the product.
- Do not pad a virtual influencer’s audience with bought followers.
Platform rules
- Instagram and Facebook (Meta). Meta’s branded content policies require branded content to be posted using the branded content tool, with the paid partnership label. Meta separately labels AI-generated content as “AI info,” using detected industry signals or self-disclosure.
- TikTok. TikTok requires creators to label AI-generated content that contains realistic images, audio or video. Creators can use its label or other disclosures such as a sticker or caption.
- YouTube. Creators must disclose realistic content that was meaningfully altered or generated with AI, using the “AI use” setting in YouTube Studio. YouTube says non-realistic content, such as obvious animation or fantasy, doesn’t need the label. Paid promotions need a separate check box in YouTube Studio.
A sponsored AI influencer post can therefore need two labels: one for the paid relationship and one for realistic AI content. Build both into your approval checklist.
How a brand can start with AI avatars
If you want to test the UGC-style ad format, a low-risk start looks like this:
- Pick one job. For example, a 30-second product explainer you currently pay to shoot.
- Write the script first. The avatar only delivers it. State only claims you can support.
- Choose a tool and check the plan. Compare the options below.
- Get consent for any real likeness. Use a signed release for employees or creators.
- Label the result. Add the paid label or “ad” disclosure, plus the platform’s AI label where it applies.
- Test against a human-made version on your own account and compare your own numbers.
HeyGen and Synthesia at a glance
Prices below are from each company’s official pricing page, US dollars, as of October 2026. They can vary by region and billing cycle, and plans change, so confirm on the site.
| HeyGen | Synthesia | |
|---|---|---|
| Free plan | $0/month: 3 videos a month, up to 1 minute each | Basic, $0/month: 500 credits, 10 minutes of video creation or dubbing, 9 AI avatars, watermark |
| Entry paid plan | Creator, $29/month: 600 monthly credits, videos up to 30 minutes, 1080p, no watermark | Starter, $29/month monthly or $18/month billed yearly: 125+ AI avatars, up to about 12 minutes of video a month on monthly billing |
| Mid plan | Pro, $49/month: 1,000 monthly credits, 4K export | Pro, $89/month monthly or $64/month yearly: 180+ AI avatars, 5 personal avatars, up to about 60 minutes a month on monthly billing |
| Team plan | Business, $149/month: 1,500 credits, 5 custom avatars, extra seats $20/seat/month | Enterprise, custom pricing |
| Custom avatar | Built from a photo or a short video recording | Personal avatars from a photo or a recorded consent video |
| Consent | Custom avatars need verified consent from the person recreated | The person in the consent video must match the avatar and be 18 or over; consent is recorded live |
Both tools make the consent step part of the product, which is useful: it puts a guardrail on the “digital twin” use case. Neither page says anything about how an avatar ad performs against a human creator, so do your own test.
Choose HeyGen if you want a free plan with a low-cost step up and plan to build avatars from photos or short recordings. Choose Synthesia if you need a larger avatar library on paid plans or have a team that wants personal avatars with a formal consent flow. If you are comparing more video options, see the AI video generator list.
Using AI to find human influencers
Not every brand needs a virtual character. Many get more from using AI to improve how they find and vet real creators. Typical jobs are searching creator databases by niche, checking whether an audience looks genuine, and drafting first-contact messages. Browse the influencer marketing tag for tools that cover this work.
Keep a few rules in mind:
- Review the content yourself. Scores and audience reports are helpful, but a creator’s actual posts and comments tell you more about fit.
- Watch for fake followers. Since the FTC’s 2024 rule covers bought followers and views, checking audience quality protects you as well as your results.
- Disclose every paid deal. The same FTC guidance applies to human creators, and it says disclosure is the influencer’s responsibility. Put disclosure terms in your contract and check posts before and after they go live.
For a wider look at tools in this space, see the social media category. To plan how these fit alongside the rest of your work, read our guide to an AI marketing strategy.
Which route should you take?
- Need a quick campaign with a known audience: pay a human creator, and use AI tools to shortlist them.
- Want a long-term brand character: build your own virtual ambassador, but plan for years of content, not a single post. Lu do Magalu is the model to study.
- Need more ad variations on a small budget: use an AI avatar for UGC-style ads, get consent, and label the result.
- Want to run a sponsored campaign with an existing virtual influencer: vet their audience as you would any creator, and require paid-partnership labels in the contract.
Whatever you choose, make disclosure part of the brief before anything is posted.