Computer-generated profiles are gaining traction with brands but may pose a threat to real life endorsers
In an eight second TikTok video, American university student Tinsley sobs, red-eyed, and tells viewers the sorority house she hoped to join had rejected her.
“Oh sweetie. There’s a house that wants you so please continue!!!” one TikTok user responds to the video, which has nearly 150,000 views. “Don’t give up,” offers another.
The words of consolation are misguided. Tinsley, who had garnered several hundred followers in a matter of days, is in fact a “100 per cent AI influencer”, according to her creator, Olivia Moore, a partner at Silicon Valley venture capital group, a16z.
Moore spent less than 10 minutes a day building the fake profile over the course of a week in August as an experiment, wielding just three existing artificial intelligence tools, she explained on X. “This may be the future of brands, marketing and even entertainment,” she gushed.
Cinematic-quality AI production tools are helping to launch thousands of digital avatars — or “AI influencers” — into the creator economy, which was estimated to be worth $250bn in 2023, whether newly generated personas or clones of existing influencers. The technology can be so convincing that one influencer, Mia Zelu, garnered more than 150,000 followers as social media clips appeared to show her attending Wimbledon this summer, before it was revealed she was actually an AI-generated personality in an artificial setting.
Proponents argue such tools can help existing influencers by automating some of their work — translating video content into other languages, responding to fans or deploying animated versions of themselves to clinch marketing deals, even while they sleep.
For brands, there is the promise of cheap, instant influencer marketing and full control over what an avatar looks like and says. It is, according to Alexandru Voica, head of corporate affairs and policy at media generation start-up Synthesia, the “great equaliser”, whereby even small brands without big budgets can afford “polished, studio-quality content at scale”.
But the shift has left some human influencers — particularly those who do not wish to use AI as part of their content creation processes — and their teams fearing displacement or disintermediation.
Already a survey of 500 marketing professionals conducted by the Influencer Marketing Hub and published in 2024, found nearly 60 per cent had experience using AI influencers in their campaigns and 15.5 per cent were planning to. The majority of those who did said the ability to customise or control the AI influencer’s behaviour was important.
“It’s dangerous for artists,” said creator and plus-size model Gabriella Halikas. “It’s dangerous for every creative involved in a photo shoot — whether that’s hair, make-up, stylist, model, photographer, videographer, set worker, producer.
“This has gone out of control . . . So you’re going to start seeing more people raise awareness,” she added. “I think we’re just starting out.”
The proliferation of AI-generated content — and wholly AI accounts — is a central pillar of the strategy of Silicon Valley social media platforms, according to executives. In October last year, Meta chief executive Mark Zuckerberg predicted AI would herald a “whole new category of content”. “This will be one of the important trends and one of the important applications,” he said.
Meta’s own AI studio allows creators to “build an AI extension of themselves”, using data on previous posts to “mimic your tone and expressions”. Many expect bulk personalisation for fans will be possible in time.
But Adrienne Lahens, former head of global content strategy and operations at TikTok, argues the biggest winners will be a new type of influencer: “AI artists.” These are people with advanced technical skills and knowledge of the AI video models developed by companies such as Google and OpenAI, who create characters with compelling storylines, she said.
Alex Mashrabov, founder of Higgsfield AI, said some creators were using his video platform to run “AI talent agencies”, directing followers to a roster of AI-generated influencers, and collecting sponsorship revenue without having to pay for physical shoots or travel.
“The next media empire won’t be built in Hollywood or Silicon Valley but . . . by a twenty-something with a small group of AI personalities,” said Mashrabov, the former head of AI at Snap. Individual creators with the right tools, he added, can “outmanoeuvre studios, out-create agencies, and out-trend Hollywood”.
Already, a brisk market catering to the AI creator economy is beginning to spring up, including studios focused on generating virtual personas and managing their narratives, and ad groups connecting brands with virtual influencers.
One of the first virtual influencers, Lil Miquela, was created by Los Angeles-based Brud, which charges up to hundreds of thousands of dollars for deals with brands such as Burberry, Prada and Givenchy.
Fameflow AI, which allows brands to license AI avatars of celebrities for marketing campaigns, said it had signed up more than 100 celebrities.
Others disintermediate real influencers altogether. Dylan Fournier, co-founder of Arcads, says paying fully AI-generated actors to create short adverts costs “€10 versus €300” with a real creator.
Several big brands are embracing the space. Last month, H&M revealed images of its first set of “digital twins”, or AI-generated clones of models, which it is using in adverts. Hugo Boss has worked with pink-haired virtual influencer Imma, who has about 500,000 followers.
But other larger advertising companies remain wary, particularly around the ethical question of if and how the use of AI should be disclosed.
The “jury is still out on consumer acceptability,” one executive at a big advertising agency said. “There are some obvious high-profile exceptions that have traded on the fact they are AI, but if that’s not the gimmick and it’s ‘just’ influencer marketing, I think there is a hesitancy on ‘fooling’ people. Even if transparent, it’s a tricky area.”
Some maintain AI will not fully replace the human influencer role as it cannot be related to, an assertion that appears borne out by data. According to a report from Twicsy, which analysed the earnings, revenue streams and audience engagement of more than 11,500 AI and real influencers, sponsored posts from human influencers got 2.7 times more engagement than sponsored posts from AI personas.
Human influencers earn an average of $78,777 per post compared with the $1,694 paid for AI, the report found.
“AI cannot experience things. It cannot be a tired parent. It cannot have a skin condition, it cannot taste food,” said Jago Sherman, head of strategy at The Goat Agency, the influencer subsidiary of WPP. He cited data from Emarketer that showed 65 per cent of adults felt uneasy about AI-generated adverts.
He predicted there would be an “uptick in influencer fraud”, whereby creators would be commissioned to produce real content but might use AI as a shortcut. “If we pay for human content and receive non-human content in return, that’s fools gold.”
Halikas warned AI influencers were creating unrealistic beauty standards, which could be damaging for mental health. “They are quite literally not real and I think that’s what’s very concerning for our youth and for women.”
The shift is also fuelling fears that platforms such as Meta will become inundated with so-called “AI slop” — low quality, bizarre or misleading AI-generated spam designed to juice user engagement for profit.
Canadian podcaster and creator Shira Lazar said this type of material would be the “fast food content” of the internet, prompting more social media users to explore where their content is coming from.
“There’s going to be a renaissance for tastemakers that you know to be true,” said Neil Waller, co-founder of the agency Whalar.
Source: https://www.ft.com/content/cb859409-dd3d-400d-9225-4a14d351bd20
