AI Virtual Influencers Rise: Real Faces Won't Disappear — Credibility Is What Matters

Brands Pour Billions into AI Influencers, But Consumers Care About More Than "Real or Fake"

Virtual influencers are becoming a new destination for brand marketing budgets. According to aggregated data, brand spending related to AI influencers is projected to reach $1.37 billion in 2026 and climb to $7.4 billion by 2030. That growth rate is enough to make any marketing professional take notice.

But the real question worth asking isn't whether human influencers will be replaced — human faces in the feed haven't decreased. The core tension is: when content implies a human life that doesn't actually exist, will audiences still buy in?

A study on human influencers posting AI-generated content attempts to answer this question.

The Cleverness and Limitations of the Experimental Design

Three Conditions, One AI Face

The study set up three experimental conditions, but all three used the same AI-generated image. In the "real person" condition, the image was only lightly retouched. This means the core variable of the experiment wasn't the realism of the image itself, but rather the human identity the content claims.

This design approach is actually quite clever: it strips away the interference of "does the image look like a real person" and directly tests the impact of "claiming there's a real person behind this" on audience judgment.

This experimental approach of "controlling the image, manipulating the claim" originates from the deception detection research tradition in social psychology. The classic paradigm holds that humans rely on two pathways when judging information authenticity: first, cue processing — looking for flaws in the content itself; second, a priori reasoning — making credibility inferences based on source identity. When visual cues are fixed, only the a priori reasoning pathway remains active in the experiment, which is precisely the variable researchers want to measure in isolation. In corporate practice, this means that even if brands invest heavily in visual realism, trust will still collapse if problems arise at the identity claim level.

Research Limitations Must Be Acknowledged

However, the design of sharing the same AI image across three conditions also brings obvious limitations. Since the image material didn't change, the study cannot compare how AI faces of different realism levels affect audiences differently. The comparison between the lightly retouched "real person" condition and the pure AI condition reflects differences at the psychological cognition level rather than the effect of visual realism. In other words, this study tested "claims," not "perception."

This involves a key distinction: claimed authenticity vs. perceived authenticity. The former refers to whether content claims to have a real human identity and life experiences; the latter refers to whether audiences subjectively feel the content "seems real." The two can vary independently — a highly realistic AI image (high perceived authenticity) may not claim to have a real person (low claimed authenticity), and vice versa. Because the image material was fixed, this study actually only measured the claim dimension and cannot answer the question of "when AI faces become realistic enough, will audiences relax their vigilance toward claims?" For brand practice, this means the risk of synthetic personas comes not only from whether they lie, but also from the cognitive gray zone created by technological evolution.

Where Is the Trust Red Line for Synthetic Personas?

Categories Requiring "Usage Experience" Are Minefields

The practical recommendation from the study is very clear: keep synthetic personas away from categories that require real usage experience.

The logic isn't hard to understand. When an AI influencer recommends a skincare product and claims "I used it for three months and my skin improved noticeably," once audiences discover this person doesn't exist at all, the entire recommendation logic collapses — it's not the product being questioned, but the recommendation act itself losing its basis of legitimacy.

Categories requiring personal trial — beauty, food, fitness equipment, healthcare — naturally have higher demands for "human verification." When synthetic personas make recommendations in these areas, they are essentially asking audiences for a kind of trust that cannot be fulfilled.

This type of recommendation requiring personal experience endorsement is called experience goods in marketing — consumers cannot directly verify quality before purchase and can only rely on others' trial experiences as substitute signals. In contrast, search goods such as electronic product specs or flight prices can have their quality verified through external information before purchase. The trust structure of synthetic personas naturally suits the latter: they can provide objective comparisons and data organization, but cannot provide "I personally used it" — the most core signal of experience goods. This explains the fundamental reason why AI influencers frequently fail in beauty and food categories.

So Do Human Creators Win?

Not necessarily. The study's implicit conclusion is more nuanced: what audiences care about isn't who produces the content, but whether the content is honest. A human influencer using AI tools to assist creation, as long as they don't fabricate experiences that didn't happen, won't see trust discounted because they "used AI." Conversely, an AI persona that doesn't claim to have a real life but honestly exists as a "virtual character" can equally build its own audience relationships.

In other words, what audiences detect isn't "AI traces" but "deception signals."

If You're Running a Synthetic Persona: Three Practical Recommendations

First, be restrained in category selection. Avoid products whose core selling point relies on "I personally experienced it." Synthetic personas are better suited for recommending categories whose effects can be externally verified and don't depend on personal experience endorsement.

Second, be transparent about identity claims. Don't claim to have a physical body, real name, or verifiable personal experiences. Treat the virtual identity as part of the character setting, not a secret to be hidden.

Third, content value must be real. Audiences can accept a non-existent person providing genuinely useful information — such as product spec comparisons, use-case analysis, price tracking. But when synthetic personas start fabricating "my usage stories," value becomes deception.

The Trust Logic Behind a $7.4 Billion Market

The projected growth of the AI influencer market from $1.37 billion to $7.4 billion reflects brands' strong demand for "controllable content producers." Human influencers have schedules, moods, and risk of scandals; AI personas won't be late, won't breach contracts, won't post controversial tweets at 3 AM.

But controllability has a cost. As audiences become increasingly familiar with the characteristics of AI-generated content, their sensitivity to "authenticity claims" will only increase. What brands are really buying isn't a good-looking AI face, but the trust commitment that can be continuously fulfilled behind that face.

Real faces won't disappear. What will disappear is content that can't withstand the basic question of "does this exist?"

Background Supplement

The conclusion that audiences detect "deception signals" rather than "AI traces" aligns closely with the source credibility model in communication studies. This model holds that credibility consists of three dimensions: expertise, trustworthiness, and goodwill. The use of AI tools itself doesn't directly damage these three dimensions, but if fabricated experiences are discovered, trustworthiness immediately drops to zero, which in turn undermines judgments of expertise. This also means that synthetic personas that transparently declare their virtual identity are actually proactively maintaining the trustworthiness dimension, and their long-term trust assets may paradoxically be more solid than accounts pretending to be real people.

One-line Summary

Brands are betting heavily on AI influencers, but the key to trust isn't real or fake faces — it's whether content honestly declares its identity.

Paragraph Summary

AI influencer brand spending is projected to grow from $1.37 billion in 2026 to $7.4 billion by 2030, a rapid increase. But a study reveals that what audiences really care about isn't whether content is AI-generated, but whether it claims a human identity that doesn't exist. The experiment used the same AI image to test three conditions, proving that what affects trust is "identity claims" rather than "visual realism." The study recommends that synthetic personas avoid beauty, food, fitness, and other categories requiring real experience, while maintaining identity transparency and providing externally verifiable content value. What brands are buying isn't an AI face, but a fulfillable trust commitment.