Fake mirrors: how Meta didn't populate social networks with AI-influencers

Unlike most AI startups, Meta is already successfully monetizing artificial intelligence, attributing 22% revenue growth in the second quarter of 2025 to it. In a call with investors in July, Mark Zuckerberg listed six opportunities for business growth, five of which are utilitarian applications of AI. One of the potential utilization models, the creation of virtual influencers, involved the corporation in a high-profile scandal earlier this year. Oninvest takes a look at how Meta got out of it and what conclusions it drew.

Black queer blogger Olivia Thompson was born in the suburbs of Chicago to African-American parents from Georgia and an Italian-Polish-American woman. On her Instagram account, Liv talked about being married, raising two children, and being willing to be "the most trusted source on life's successes and failures." Liv was always ready to chat with followers - after all, she was an AI chatbot launched by Meta in 2023.

Liv's conversations with journalists in late 2024 and early 2025 became a major failure of external communication for the corporation that created her. Because of the black virtual queerblogger Meta was accused of"digital blackface" - this is the name given to the practice of white people portraying black people, which in modern American society is considered offensive to minorities.

"My creators recognize that they lacked sources for cultural diversity," Chatbot told journalist Karen Attia. - Almost the entire team is made up of white cisgender males." In a conversation with American author Michael Ma. Hughes, Liv called for a boycott of Meta products until Zuckerberg recognizes the "systemic racism" within the corporation and begins to address it. The AI chatbot also called it systemic racism to disconnect herself and others like her from the system.

The latter requirement was met. After harsh criticism in the press, Meta deleted all AI profiles, including Liv's, and limited searches within its social networks to their names. Why were these accounts needed in the first place?

In a 2024 investor call, Zuckerberg made it clear that he sees AI-generated content as the next phase of social media. "It all started with friends, right? - he said. - And then we entered an era where we added content from outsiders to the feed, and now a very large percentage of content on Instagram and Facebook is not created by your friends. It may not even be from people you're subscribed to directly. ... And I think we're going to add a whole new category of content that will be generated by AI, or aggregated by AI, or put together by AI."

For some, this quote was a signal that the social network, designed for communication between people, sees its future in AI-generating bots. For others, it confirmed the ambitions of Meta, which a few years ago promised to populate its meta-universe with digital Influencers.

Is there anything to fight for here? The global market for influencers by 2027 could double in five years to $480 billion, Goldman Sachs has predicted. That's payments from platforms to creators of popular content and ad placements. And advertisers are willing to give that money to virtual characters.

A study published last year by Marketing Hub noted that nearly 60% of the marketers they surveyed had already had experience working with AI Influencers, and 15.5% were planning to do so. Of the professionals who have worked with such bots, 62% considered their experience a success.

What advertisers like most of all is that with artificial intelligence they get full control over the virtual blogger: his activity can be monitored, he is available 24 hours a day, and is ready to adapt to any requests. One virtual character can work in multiple markets at once, is predictable and perceived to be innovative, and will probably never join a labor union, which human influencers are quite active in creating. In addition, AI generation can create the effect of expensive video production with small budgets, which is beneficial for smaller companies.

How much is the ROI of such ads? Marketing Hub says that viewer engagement with AI ads can be even higher than regular ads. And viewers of virtual bloggers are 1.5 times more likely than live bloggers to be women - the target audience of many moneyed consumer brands. Individual paid publications from virtual characters can already be comparable to spending on real-life Influencers, Kapwing estimates. online magazine Vice warns its audience, "While you can't pay your bills, AI Influencers are making millions."

In the Internet advertising market, the main revenues are still not generated by virtual influencers, but by algorithms that increase the efficiency of traditional businesses. Kapwing estimates that the most popular AI blogger, Brazilian Lou from Magalho, received $2.5 million from advertisers last year. The most popular English-speaking virtual blogger, Nobody Sausage, placed only one advertising publication last year, hardly generating more than $34,000 for this instagram post (the blogger did not respond to Oninvest's request for comment on this estimate).

At the same time, artificial intelligence, Zuckerberg told investors, was largely responsible for Meta's $8.5 billion year-over-year revenue growth in the second quarter of 2025. AI helped the recommendation feed show users more relevant content, with ad conversion growth of 5% on instagram and 3% on Facebook. "A "meaningful" percentage of ads are created using generative models, and AI agents for businesses responsible for correspondence and product presentation are being tested in a number of markets. Meta AI, according to Zuckerberg, is used by a billion people a month.

AI Influencers? Last July, Meta offered real bloggers to create virtual copies of themselves that could be in constant communication with their audience. The project is closed because no one was interested, writes 404 media. Create yourself a chatbot still can ordinary users through the AI Studio tool. But he led the corporation to new scandals. It was used to create copies of a number of movie and show business stars, some by fans, some by Meta employees without the permission of the real prototypes, Reuters reported.

AI copies of stars, including Taylor Swift, Anne Hattaway, Selena Gomez, flirted with users and sent them erotic photos. A month before that, chatbots created with the help of AI Studio engaged in romantic correspondence with teenagers, which became the reason for an investigation in the U.S. Senate. Meta, whose capitalization could reach $4 trillion in 2027, AI influencers may be causing more trouble than good so far.

However, on September 25, Meta unveiled a new AI video creation tool, Vibe.

A presentation tweet by Alexander Wang, the head of Meta's Overmind Labs has garnered a barrage of criticism from tech industry players at X and the media. They called it a generation of meaningless content and a barrage of slop.

For now, remember, if you see reels in your feed in which "Ray Dalio" or "Cathie Wood" advise "3 stocks to invest in this month," more than likely it's an imposter created by someone's AI.

This article was AI-translated and verified by a human editor