Instagram to limit reach of undisclosed AI-generated profiles

Instagram is tightening its rules around AI-generated personalities, with the platform set to limit the reach of profiles featuring AI-generated people if creators fail to disclose that the person users are seeing is synthetic. 

The move comes as AI influencers and digital avatars move from experimentation into brand marketing, putting greater pressure on platforms, creators and advertisers to make the distinction between human and AI-generated identities clearer.

As part of the change, Instagram is replacing its existing “AI creator” label with the more explicit “AI-generated profile”. The platform said the terminology is intended to make the disclosure easier for users to understand, particularly in cases where an account may appear to represent a real person but the person featured on it has been entirely or substantially created using AI.

The distinction is becoming increasingly relevant for marketers as synthetic personalities enter the creator economy. Instagram said the change was partly prompted by users discovering profiles that appeared to belong to real people, only to later realise that the person was entirely AI-generated. “They want to know when a profile features an AI-generated person,” the platform said.

India already has a growing roster of virtual personalities, from early names such as Kyra and Naina Avtr to brand-owned and category-focused profiles such as Maya and Kavya Mehra, with newer AI-first personalities including Vrutika and Aditi.AI entering the market. Kyra, created by FUTR Studios and launched in 2022, is widely regarded as one of India's first mainstream virtual influencers and has worked with brands including boAt, Amazon Prime Video and John Jacobs. Naina, developed by AVTR Meta Labs, has similarly built a presence around fashion and lifestyle content.

For creators, the consequence is more direct. Accounts that do not properly disclose an AI-generated profile could see their reach reduced, while creators using the new label will not be penalised simply for featuring an AI-generated person. The label, however, is not intended to cover every form of AI use. Creators using AI for tasks such as editing photographs, improving captions, creating graphics or making other creative changes will not be required to use the AI-generated profile label.

AI content - not necessarily cheap content

The policy change arrives at a time when the marketing industry is increasingly testing AI creators on the promise of lower production costs and greater scalability. But the economics of producing convincing AI-led content are proving more nuanced, particularly when brands expect polished, repeatable and campaign-ready output.

Varun Mayya, founder of Aeos and one of India’s prominent AI-first creators, recently put a number to that production challenge. He said his team spent $8,500, or around ₹7 lakh, in AI credits to produce a single high-quality video featuring his AI avatar, with repeated generations required before arriving at the final output.

The spending is not limited to individual videos. Mayya has also said his company spends more than $5,000 a month on AI-generated assets and testing, highlighting an often overlooked part of AI production economics: the cost of experimentation. While the final asset may look like a faster and cheaper alternative to conventional production, getting there can involve hundreds or thousands of generations, iterations and discarded outputs.

That cost becomes more interesting when placed against the commercial value of the digital likeness itself. Mayya has said he was offered ₹50,000 for allowing his AI avatar to be used in a television advertising campaign, suggesting that the value equation for AI creators is beginning to extend beyond content production to the licensing of synthetic identities.

For advertisers, this creates a new layer to the AI creator opportunity. Brands may be able to use digital likenesses for always-on, multilingual or highly scalable campaigns, but the cost of building credible AI content and the need to disclose its synthetic nature are emerging as equally important parts of the equation.

Transparency becomes part of the creator brief

Raj Mishra, MD & CEO, Chtrbox, said the scale of AI adoption is already changing how marketers approach content production, but argued that efficiency cannot come at the expense of the human qualities that drive creator marketing.

“The fact that nearly one in four social posts now involve AI isn't just a productivity milestone. It's a signal that marketing is entering a fundamentally new phase,” Mishra said.

According to him, AI is allowing brands to create, test and personalise content at scale, but “scale alone doesn't build affinity”. He sees AI more as a creative collaborator across ideation, production, localisation and optimisation rather than as a replacement for creators.

That distinction could become increasingly important as platforms introduce clearer rules around synthetic identities. For brands, the issue is no longer simply whether AI can produce content at a lower cost, but whether audiences understand what they are looking at and whether the creator identity being used in a campaign is transparent.

Mishra said AI-powered celebrity avatars and digital likenesses could enable “always-on, multilingual campaigns”, but argued that the rise of synthetic identities makes transparency and consent more important.

“AI can generate content, but it cannot replicate lived experiences, community trust or cultural context,” he said, adding that brands will need to use AI to amplify human creativity while remaining transparent and responsible in how they engage audiences.

Instagram's move therefore adds another consideration to the growing AI creator playbook. As synthetic personalities become easier to create and deploy, disclosure could increasingly become part of the basic campaign brief, alongside reach, engagement, production costs and usage rights.

For the creator economy, the development also signals a shift in how platforms may treat AI identities. The question is no longer only whether an AI-generated person can attract an audience, but whether that audience knows that the person they are following does not actually exist.