Instagram’s new rules for AI influencers force a question the platform probably didn’t intend to ask.

Something quietly significant happened when Instagram announced it would require AI-generated accounts to carry a mandatory “AI-generated profile” label. The platform framed it as a fix for deception – users had told Instagram they disliked discovering that accounts they thought were human turned out to be entirely synthetic. The response was straightforward: make the disclosure compulsory, penalise accounts that refuse with reduced reach, offer an appeals process.

That is the news. But the more interesting question is what happens next, not to Instagram’s policies, but to our assumptions about what we actually want from the people we follow online.

The deception problem is real, and the policy is a meaningful step

AI influencer accounts are not a marginal curiosity. Aitana Lopez, a hyper-realistic AI model, created by a Barcelona-based agency, reportedly earns somewhere between $10,000 and $11,000 a month through brand deals – figures that originate from the agency itself and should be read as approximate commercial claims rather than audited data, but that nonetheless signal a functioning commercial market. Noonoouri, a more stylised digital persona, has accumulated nearly half a million Instagram followers and secured partnerships with Dior, Versace and Balenciaga, before signing a record deal with Warner Music in 2023.

These accounts operate within the same attention economy as human creators, competing for the same brand budgets and the same moments in someone’s daily scroll. Until now, the rules around disclosure have been voluntary, an “AI creator” tag existed but wasn’t required. Instagram’s decision to rename that label “AI-generated profile” and make it mandatory, with algorithmic penalties for non-compliance, represents a real escalation.

The explicit justification from an Instagram spokesperson was user feedback: people disliked discovering that they had been engaging with what they thought was a human. That felt sense of deception – the retrospective reframing of an interaction as something other than what it seemed – appears to have been the immediate trigger.

What the policy does not do is ban AI influencer accounts. It requires them to announce what they are. That distinction matters more than it might initially appear.

The question disclosure actually opens

Before mandatory labelling, two different problems were tangled together. One was active deception where users were being misled about the nature of what they were following. The other, more fundamental question was whether users would engage with a synthetic personality at all if they knew it was artificial.

Instagram’s policy addresses the first problem. In doing so, it creates the conditions to observe the second.

Here is what Instagram’s own evidence does and does not establish. It tells us that users disliked discovering, after the fact, that a profile they believed to be human was AI-generated. It does not tell us that users reject synthetic personas when the nature of those personas is known from the start. That is a genuinely different question, and the answer is not yet available.

There is good reason to think that the answer might be more complicated than either “people will always prefer humans” or “people will happily follow AI accounts once they know.” Consider how audiences already relate to animated characters, fictional personalities and constructed personas. People form genuine attachments to entities they know are not real. The prerequisite for connection is not always the belief that something is human – it is something more like finding that thing interesting, funny, appealing or engaging. If that is true for audiences, clearly labelled AI influencers could continue to attract followings and generate brand revenue without meaningfully disrupting the commercial market.

What being human might actually be worth

Accepting that AI personas can function as entertainment without requiring deception does not mean synthetic and human personalities become equivalent. There is a specific dimension of human influencer culture that AI cannot replicate – not because of technical limitations, but because of what the thing actually is.

A synthetic personality can generate images of travelling somewhere, overcoming something, developing a skill. It can produce the appearance of experience. What it cannot do is actually have those experiences, because experience requires a conscious subject who inhabits a body, encounters uncertainty and lives with consequences.

This matters because a significant part of what audiences find compelling in human creators is not just the output, the photograph, the video, the opinion, but the knowledge that a real person actually did or felt or achieved something. The admiration involved in watching someone become genuinely skilled, or seeing them fail and recover, or knowing that their expressed view cost them something to hold, derives partly from the fact that they are a human being operating under human constraints.

Sport offers a useful analogy, though it is our reasoning rather than evidence from research. A machine could theoretically outperform any human athlete in almost any measurable dimension. That greater performance would not make watching machines compete equivalent to watching extraordinary humans perform within limits we all share. The interest in human sporting achievement is partly constituted by those limits. Remove them, and something changes – not because the mechanical performance is less impressive, but because a different kind of meaning was always embedded in the human version.

The parallel for influencer culture is imperfect – Instagram is not the Olympics – but the underlying structure may be similar. As synthetic personality becomes cheaper, more abundant and more convincing, genuine human experience could become comparatively scarce and more legible as something distinct. AI personas might flourish in forms of content where the output itself is what matters: aesthetics, product demonstration, entertainment formats where the question of whether a real person is behind the screen is simply irrelevant to why you’re watching. Humans might retain a different kind of advantage where lived experience, genuine achievement, accountability, or the knowledge that another conscious person actually went through something forms part of why audiences care.

This is coexistence and differentiation, not replacement. It is also speculation — how audiences will actually respond to a world of clearly labelled AI personalities is not yet established by evidence.

The contradiction Meta has not fully resolved

There is a complicating background to all of this. Instagram’s new policy is being introduced by Meta, the same company that, according to secondary reporting, previously deployed its own internally created AI profiles on its platforms. Meta also announced plans to let users create AI characters that would operate as functional profiles on Instagram and Facebook. Whether Meta subsequently scaled back those first-party AI profile experiments is based on secondary sources and has not been independently confirmed.

The company is now mandating disclosure for third-party AI persona accounts while its own relationship with synthetic identity on its platforms remains unsettled. Whether this represents genuine course-correction, regulatory positioning ahead of AI transparency requirements in the EU and elsewhere, or something more commercially strategic is not clear from the available evidence. Probably some combination of all three.

What is clear is that labelling is an incremental governance step, not a comprehensive solution. Enforcement depends on detection mechanisms that the platform has not made transparent. Compliance depends on operators finding the penalties credible enough to deter evasion. And the practical effect on users depends on whether they notice and understand labels in the flow of normal scrolling – which, based on the general track record of content labels on social platforms, is not guaranteed.

The deeper question Instagram has accidentally surfaced is one the platform cannot answer through policy alone. If people knowingly continue following AI personas after disclosure, and there is no strong reason to assume they won’t, that reveals something about what online audiences are actually seeking. Not necessarily human connection in any deep sense, but perhaps novelty, aesthetic pleasure, entertainment, a sense of being part of something. Those things, it turns out, may not require a real person at all.

What remains uncertain is whether anything specifically human, the fact of genuine experience, the weight of real consequence, the knowledge that someone actually lived through what they’re sharing – retains a distinct value once the novelty of AI persona content settles into the ordinary. That question is just beginning to come into focus.

The paid Unclear Shift article goes further: examining the competing scenarios for how the influencer economy might bifurcate between AI and human creators, what signals to watch as Instagram’s labelling policy takes effect, and what this development could mean for the humans – models, creators, talent agencies -most directly in the path of these changes:

https://unclearshift.substack.com/p/after-the-label-what-the-influencer

Sources:

– The Verge — “Instagram addresses fake AI profile slop”: https://www.theverge.com/tech/986593/instagram-addresses-fake-ai-profile-slop

– Engadget — “Instagram will demote AI-generated influencers if they don’t clearly label their account”: https://www.engadget.com/2246914/instagram-will-demote-ai-generated-influencers-if-they-dont-clearly-label-their-account

– creatorflow.so — “AI Influencers on Instagram: Top 10 + Earnings (2026)”: https://creatorflow.so/blog/ai-influencers-instagram

– Business Insider (via Facebook) — Aitana Lopez profile: https://www.facebook.com/businessinsider/posts/aitana-l%C3%B3pez-is-an-ai-generated…

– TechTwisted — “Meta Embraces AI-Generated Social Media Accounts in Bold Platform Shift”: https://techtwisted.com/news/meta-embraces-ai-generated-social-media-accounts-in-bold-platform-shift

– Medium / Shivam More — “Why Meta is killing off its own AI-generated Instagram and Facebook profiles”: https://shivammore.medium.com/why-meta-is-killing-off-its-own-ai-generated-instagram-and-facebook-profiles-380296c8aa91