Fake people, real money: How AI influencers are fooling millions of followers

AI-generated avatars are flooding social media feeds, promoting products and building loyal audiences, often without disclosing that they are not real people. Charlie Warzel reports for The Atlantic, drawing on an interview with New York Times technology reporter Tiffany Hsu, who has investigated the rise of synthetic influencers selling supplements and other consumer products.

One example is Melanskia, an AI avatar with more than 300,000 followers. She appears to be an Amish mother promoting clean living and wellness advice. She is not a real person. Her creator, entrepreneur Josemaria Silvestrini, manages a network of AI avatars that promote his supplement brand. He outsources the actual creation of the avatars to others and sees the practice as a cost-effective marketing strategy.

Hsu explains why the wellness industry is particularly vulnerable. Most consumers lack the scientific background to distinguish legitimate health products from fraudulent ones. A synthetic persona speaking with apparent authority and personal experience can be highly persuasive. Tim Caulfield, a health science professor in Canada, tells Hsu that scammers have recognised AI avatars as an easy and cheap entry point into this already fraud-prone space.

The economic logic is straightforward. Creating an AI avatar eliminates costs associated with human influencers: no appearance fees, no product shipments, no studio bookings. Online tutorials promise monthly earnings of up to 30,000 dollars. The venture capital firm Andreessen Horowitz has invested in Doublespeed, a company that specialises in bulk AI content creation. Its website carries the slogan “Never pay a human again.”

The sheer volume of synthetic content is producing what Hsu describes as epistemic exhaustion. Audiences are so overwhelmed by the task of distinguishing real from fake that many have stopped trying. During hurricane coverage in North Carolina, a local official shared an AI-generated image of a child on a flood raft and responded to criticism by saying the feeling it evoked was real, regardless of its origin. Hsu sees this attitude spreading.

Some human influencers are responding. The lingerie brand Aerie posted a pledge to never use AI-generated bodies, and it became their most popular post of the year. Meanwhile, AI avatar Aitana Lopez, a fictional fitness influencer based in Barcelona with nearly 400,000 followers, has secured brand deals with companies including Alo Yoga.

Legislation is beginning to emerge. New York has passed a law requiring disclosure of synthetic performers in advertisements. Hsu is cautious about its impact. Many creators are anonymous, operate from outside the country, and the law is difficult to enforce at scale.

Hsu draws a parallel to the so-called “liar’s dividend,” a phenomenon where the widespread presence of AI makes it easier to dismiss authentic footage as fake. Israeli Prime Minister Benjamin Netanyahu recently posted a verified proof-of-life video after speculation about his appearance in a recording. Significant parts of the public rejected it as AI-generated regardless.

The boundary between real and synthetic influence is eroding faster than regulation or public literacy can respond.

Full interview