AI influencer: what a virtual influencer is, what it costs and when it makes sense for a brand

What is an AI influencer?

An AI influencer (in Polish often used in the female form, „AI influencerka”) is a character that does not physically exist but runs a social media profile like a real creator: it publishes photos and videos, comments on other people’s posts and works with brands. Its appearance, and often its voice and copy as well, are created with the help of AI, but a team of people decides what the character says, what views it holds and whom it works with.

Three phenomena are often confused. A virtual influencer is a character with its own “biography” and an independent profile. A brand avatar belongs to one company and speaks only on its behalf, a bit like a mascot, only more modern. Deepfake is a forged likeness of a real person and has nothing to do with marketing; we have covered it separately in our article Deepfake: what it is.

The phenomenon isn’t new. Lil Miquela was created in 2016 by the American studio Brud and, according to the European Youth Portal, has 2.5 million followers on Instagram; the only new thing is that today’s tools make it possible to create such a character with a much smaller team than a decade ago. Virtual creators are one of many threads we pull together in our guide to artificial intelligence in business.

What can we learn from examples of virtual influencers?

The most frequently cited European example is Aitana López, a virtual influencer created by the Spanish agency The Clueless. As Euronews reports, her creator Rubén Cruz came up with her when the agency was losing jobs because of unreliable models and influencers. According to the agency, the character earns around €3,000 a month on average, with a record of €10,000, and gained more than 343,000 followers in eighteen months. This is the creators’ own claim, not an independent measurement. Treat it as such.

The method is more interesting than the numbers. Every week the team decides what the character does and where it “goes”, and designers combine AI with manual retouching in graphics software. Cruz says plainly that people “follow lives, not images”, which is why the character was given a detailed personality and interests; AI produces the image, but influencer marketing still rests on a story written by people.

According to the same Youth Portal, the largest virtual profile in the world is the Brazilian character of the Magazine Luiza chain, with 7.2 million Instagram followers. It is an example of a brand avatar rather than an independent creator, because from the outset the character has spoken on behalf of a single company.

Who is an AI influencer right for, and in what situation?

In marketing, an AI influencer works best where the product is digital and audiences are used to virtual characters, for example in gaming, technology, digital fashion or campaigns aimed at younger audiences. It also makes sense as a permanent brand avatar serving many markets and languages for a company that does not want its image to depend on a single person.

It is a poorer fit for sectors where recommendations rest on experience: food, cosmetics, health, finance. A character that has never eaten, never used a face cream and never taken out a loan cannot honestly say “I recommend it because I’ve tried it”. In B2B communication a virtual influencer will rarely replace a named expert, because what counts there is personal brand of real people.

How to create an AI influencer for your brand, step by step

How do you create an AI influencer that doesn’t end up as an eye-catching profile with no followers? The order of the steps matters more than the tools.

1. STEP 01Goal and metric

   Write down why the brand needs the character: reach in a new group, a permanent face for campaigns, serving multiple markets. Also decide how you will measure the effect (enquiries, channel sales, awareness in a survey), because follower numbers alone prove nothing.

2. STEP 02Personality, not just looks

   Describe the character like the lead in a TV series, with interests, a way of speaking, off-limits topics and an attitude to the brand. Decide whether it is an independent creator or an openly declared company avatar. The second option is more honest and safer.

3. STEP 03Rules for appearance and labelling

   Write down a fixed description of the character’s appearance so that it doesn’t change its face between posts (these are the same consistency rules we describe for AI-generated graphics and logos), and decide how it will be labelled as AI-generated. Put the label in the profile bio and in every post that shows a realistic likeness.

4. STEP 04Small-scale test

   Before you build a profile, show the character in a few pieces of content on the brand’s existing channels and read the comments. Curiosity or dislike? A test like this costs less than withdrawing the whole project.

5. STEP 05Team and moderation

   Decide who writes the scripts, who produces the material, who replies to comments and who approves collaborations. The character is active every day, so without a permanent team the profile quickly dies.

6. STEP 06Measurement and a decision after one quarter

   Compare the results with a campaign featuring a real creator or with a baseline period. If the character doesn’t deliver, close the project rather than keeping it going out of habit.

How much does an AI influencer cost and what drives the budget?

The popular story goes that an AI influencer is cheaper than a human because it “doesn’t charge a fee”. In reality, the costs simply shift from the creator’s fee to the work of the team that keeps the character going every day. There are no reliable public price lists, so instead of figures we list the factors.

Labelling ads and AI content from a virtual influencer

A virtual character is subject to the same rules as a real creator, plus the regulations on AI content. Here are the most important ones as of September 2026.

• Labelling advertising. The recommendations issued by the President of UOKiK (the Polish competition and consumer protection authority) on 26 September 2022 call for clear labels, e.g. #reklama or #materiałreklamowy, at the start of the post, in Polish, and for “two-level” labelling: using the platform’s feature and in the content itself. They apply to every commercial partnership, and therefore to virtual characters as well.
• Labelling realistic AI-generated images. From 2 August 2026, Article 50 of the AI Act requires the disclosure of deepfakes, i.e. realistic images, audio or video recordings that imitate real people or events. A photorealistic character that audiences might take for a human should be clearly described as generated. You will find the details in our guide to labelling AI content and in the overview The AI Act in Poland.
• Platform rules. Since at least 2023, TikTok has required realistic AI content to be labelled, and since 9 May 2024 it has automatically labelled content carrying Content Credentials metadata (a digital “certificate of origin” for a file). Since July 2024, Meta has used an “AI info” label on content detected as generated.
• What comes next. The European Commission has announced a draft Digital Fairness Act for Q4 2026, which is set to cover, among other things, marketing carried out by influencers.

When a virtual character harms a brand

The biggest risk is to reputation, not a legal one. Audiences react ever more sharply to content they consider “artificial”, and in December 2025 the Polish brand Dafi withdrew an advert generated entirely by AI after a wave of criticism, as we described in our piece on AI advertising and big brands’ blunders. It is much the same with a virtual character. The scale is simply bigger, because the character “speaks” every day.

The second risk is accusations of manipulation when the character recommends a cosmetic it has never used or talks about experiences it has not had. The third is a crisis in the comments that the team is not ready for. How to prepare for this is covered in our guide to AI in crisis communication, while we discuss the ethical boundaries in our article on AI risks and ethics.

How to measure the impact of working with an AI influencer

Follower and like counts are not enough. Virtual profiles are easy to “pump up” with paid advertising, and reactions driven by curiosity do not translate into sales. Measure the same things as in any influencer collaboration: traffic from unique links and codes, the number of enquiries, sales, plus the change in awareness in a survey before and after the campaign.

Correlation is not impact. If sales rise during a campaign featuring the character, compare them with the same weeks of the previous season or with a market where the character did not appear, because only then can you tell the character’s effect apart from ordinary seasonality. Also track the mood in the comments. A predominance of complaints about “artificiality” is a warning sign that appears before results drop.

When to choose an AI influencer and when to stick with people?

A virtual character makes sense when the brand has a digital product or younger audiences, needs a consistent face across many countries, has the patience to build reach over many months and is prepared to label the character openly. A good moment is the launch of a new product line or entry into a new market, when you are building communication from scratch anyway.

Stick with people when the brand relies on authenticity and craftsmanship or on personal experience, when you need results within a few weeks, or when you don’t have a team to run the profile day to day. Nor is a virtual character a way around the rules on labelling advertising. The same rules apply to it.

A combination often works best: real creators for recommendations and experiences, and AI for producing variants and adaptations. How to build a consistent image in social media. If you want to assess whether a virtual character suits your brand and audience, take a look at the Commplace social media service.

Where brands stumble with virtual characters

Read next