Influencers and AI: what really changed in 2026

An AI influencer is a content creator whose online presence, wholly or partly, rests on artificial intelligence. That one label hides three realities that raise different risks and different questions for a brand: the fully virtual character, the AI-assisted human creator and the digital twin of a real person.

The virtual character has no body: its face, its silhouette and sometimes its voice are computer-made, and a team runs it. One example documented in the press is Aitana Lopez, a Spanish model created by The Clueless Agency. According to a Euronews report dated November 2023 on its page, she had more than 343,000 Instagram followers, and her creators said she could earn up to €10,000 a month, about €3,000 on average. Those earnings are stated by the agency: we read no independent check of them, and the figures date from 2023-2024 (page updated in December 2024), not 2026. The case still shows what such a character aims for: playing the part of an influencer in brand campaigns. The same model exists in France: in June 2024, two French virtual influencers, Anne Kerdi (about 10,000 Instagram followers) and Lalina (more than 90,000), as of the article’s date,, were among the ten finalists of the “Miss AI” contest, picked from more than 1,500 entries, according to the trade site Gens d’Internet. And it is not reserved for independent creators: in October 2022, the British retailer Marks & Spencer introduced Mira, a computer-generated character with her own Instagram account, described in its press release as the first of its kind for UK retail.

The AI-assisted human creator is the one whose AI use is least noticeable. They film and sign their content, but use AI for editing, subtitles, translation, ideas or scripts. Here, it is the platforms that are moving: according to YouTube (page dated 15 May 2026), its automatic dubbing covered 27 languages, eight of them with expressive speech, and in December 2025 more than 6 million viewers a day watched more than ten minutes of dubbed content. A creator can thus be followed in other languages without speaking them. None of the studies we read measures properly how many creators use these tools: be wary of adoption percentages that come with no sample and no date.

The digital twin is a copy of a real person, made with their agreement. In June 2024, TikTok launched custom avatars in its advertising tool Symphony: an avatar that represents a creator or a brand spokesperson, alongside “stock” avatars played by paid actors. In fashion, H&M announced, according to the trade site FashionUnited (an article with no date shown, whose address suggests March 2025), that it was preparing “digital twins” of 30 consenting models, paid “in the same way as for their photographs”, who would keep control over their replica. That is a statement of intent, not a track record.

These three groups differ on one simple question: who is behind the image? In the first case, an agency or a brand, which will sign the contract. In the second, a human who may or may not disclose the help they get. In the third, a person whose face or voice is reproduced, with rights to settle. Platform rules, the law and audience trust do not apply to each of them in the same way.

All three major platforms have moved since 2024, but not in the same direction: YouTube acts on monetisation and likeness, Instagram on the visibility of generated profiles, TikTok on labelling and viewer choice. Here is what their official pages say, and what they leave out.

On 15 July 2025, YouTube renamed its repetitious content policy “inauthentic content” in its help centre. According to TechCrunch (article of 20 July 2026), the platform clarified on 16 July 2026, for members of its partner programme, three categories of inauthentic content: generic, repetitive or template-based content, off-putting or distressing content, and content where AI personas discuss sensitive topics such as health or finance. We did not read an official 2026 page on this point: the detail rests on TechCrunch alone. Another move: on 21 April 2026, YouTube extended its likeness detection, which spots AI-simulated faces in the manner of Content ID, to talent agencies and the celebrities they represent, with no channel needed. The tool is therefore not open to every creator.

On 31 August 2026, Instagram renamed its “AI creator” label “AI-generated profile”. According to its official page, a profile that features an AI-generated person and does not carry the label can lose eligibility for recommendations and see its reach limited. Creators who only use AI as a tool do not need to add it. The page states neither the countries concerned nor the size of the drop in reach: do not infer a figure.

On 24 November 2025, TikTok announced it was testing a setting that lets viewers see less or more AI-generated content in the For You feed, along with invisible watermarks and the adoption of Content Credentials (the C2PA standard). According to the platform, its labels have already covered more than 1.3 billion videos. We did not read any official TikTok announcement dated 2026 on this subject.

The timeline below puts these announcements in order, together with that of the European regulation, detailed further down.

The public says it is wary and asks for transparency, according to three surveys published in 2026 by Gartner and Clutch. These are self-reported surveys of different samples, and their publishers do not release the exact wording of the questions: the figures do not form a series and should not be added together.

According to a Gartner press release of 22 September 2026, based on 1,006 U.S. consumers surveyed in May and June 2026, 35% say they rely on influencers less for shopping because of the rise of AI, while 43% say they rely more on real people. In the same survey, 65% think brands produce too much AI content and 57% say they trust brand messages less. Emily Weiss, a Gartner analyst, sums it up: “The answer is not to avoid AI. The answer is to use it thoughtfully.” In other words, the publisher’s conclusion is not to ban AI.

Another Gartner press release, dated 9 June 2026 and based on a small sample of 307 U.S. consumers surveyed in March 2026, says 49% think generative AI has made the quality of available content worse. Among Gen Z and millennials the share rises to 57%, but the size of that subgroup is not given.

At Clutch, a marketplace of service providers, a survey of 601 consumers in August 2026, published on 8 September, found 53% less likely to buy from a brand they know uses AI in its social content, nearly 90% who want brands and creators to disclose AI-generated content, and 71% who would question a human influencer’s recommendation if AI had been used to create the content. The respondents’ country is not stated on the pages we read. Clutch does not describe the public as hostile, though: Anna Peck, of Clutch, notes that consumers are not necessarily opposed to AI; the call for transparency stands out in the survey.

What should you take from these figures? First, transparency is in high demand: nearly nine respondents in ten in the Clutch survey ask for it. Second, distrust of influencers themselves remains a minority view in the Gartner study (35%), and a larger share says it relies more on real people (43%). Third, these are stated intentions, not measured purchases, and both Gartner studies cover Americans only: apply them to your own audience with care, by reading the comments and reactions under the creator’s posts.

Two texts can apply together to AI content in a paid collaboration: France’s 2023 influencer law and the EU AI Act. What follows is a reference point for a brand, not legal advice, and it holds as of the day we read the texts, 3 November 2026.

In France, Article 5 of Law no. 2023-451 of 9 June 2023, in the version in force since 8 November 2024 (ordinance no. 2024-978 of 6 November 2024), requires in commercial influence the mention “Images virtuelles” (the French label for virtual images) for an image produced by artificial intelligence to represent a face or a silhouette, and the “retouched images” mention for an image altered to slim or thicken the silhouette. An equivalent, suitably adapted mention is accepted. The offence is punished by up to one year in prison and a fine of up to €4,500. We read this text through an automated summary of the Legifrance page, where the wording is in the plural: check it on Legifrance before writing it into a contract.

At European level, Article 50(4) of the AI Act requires whoever deploys an AI system that generates or manipulates an image, audio or video amounting to a deepfake, meaning content that resembles existing people, objects, places or events and could appear authentic, to disclose that the content has been artificially generated or manipulated. For an evidently artistic, creative, satirical or fictional work, the disclosure is limited to a form that does not hamper the work. According to the European Commission, these transparency obligations apply from 2 August 2026; systems already on the market before that date have until 2 December 2026 (that date concerns AI systems already on the market, not creators). The text targets the “deployer”, not the category “influencer”: how it applies to a lone creator is a matter of interpretation, and a brand or an agency can be a deployer. We read the text on the Future of Life Institute mirror, which is unofficial.

Answering a senator’s written question, the minister in charge said on 30 April 2026 that no national extension of an AI label to all advertising was announced, and pointed to Article 50 and its guidelines. The “images virtuelles” mention applies to commercial content on social networks.

AI does not change how you choose a creator; it adds one question to put to them in writing. The criteria stay the same: audience, engagement, content quality, fit with your product. What the 2026 trend adds is that an image or a voice may no longer be what it seems, and the platform, the law and part of the public expect that to be said.

Start by reading the creator’s profile: a BriefBench profile gives the account’s audience and engagement and, on Instagram, an authenticity indicator estimated from visible profile signals. These figures tell you whether the creator reaches the right people; the use of AI is something you ask the creator about, in writing.

Then put the question to the creator, before signing. These are the three that matter:

• Do you use AI for your visuals, your voice, your translations or your scripts, and which of my pieces of content would it affect?
• How do you disclose it today, on the platform and in the post?
• If a face or voice is reproduced, whose is it, with what written agreement, and who owns the rights?

Also look at what the creator already shows on recent posts: a mention already in place is a sign of good faith, and its absence is not proof of the opposite. For a virtual character, ask who the signatory is: it is the agency or company that runs it, not the character.

Finally, write the answer into the brief and the contract. A useful AI clause has four points: the creator declares in advance what they produce with AI; they add the required mention to every piece of content concerned, platform and legal ones included; they provide proof of agreement for any image or voice reproduced from a real person; and they warn you before changing practice mid-campaign. H&M announces for its digital twins pay for the models, control over their replica and a watermark on the first images: these are points to settle in writing.

A guided reading of a creator profile helps you note what can be seen from the outside, with no account and no score.

To draft the clause, see our article on reaching out and the written contract.

Before choosing, also see how to judge a creator’s engagement rate and how to spot fake followers on Instagram.

• Gartner, “Gartner Marketing Survey Finds 35% of Consumers Rely on Influencers Less Due to AI”, September 22, 2026
• Gartner, “Gartner Survey Finds 49% of U.S. Consumers Say GenAI Has Made Content Quality Worse”, June 9, 2026
• Clutch, “AI in social media report”, September 8, 2026
• Instagram for Creators, “Bringing More Transparency to AI-Generated Profiles on Instagram”, August 31, 2026
• YouTube, “Inauthentic content (help centre page)”, July 15, 2025
• TechCrunch, “YouTube clarifies policies around AI slop and upsetting videos”, July 20, 2026
• YouTube, “YouTube auto dubbing, explained”, May 15, 2026
• YouTube, “Expanding likeness detection to the entertainment industry”, April 21, 2026
• TikTok Newsroom, “More ways to spot, shape and understand AI-generated content”, November 24, 2025
• TikTok Newsroom, “Meet Symphony Avatars: TikTok’s New Creative AI Tool”, June 17, 2024
• FashionUnited, “H&M to create ‘twins’ of models with AI”, March 28, 2025
• Euronews, “Meet the first Spanish AI model earning up to €10,000 per month”, November 22, 2023
• Gens d’Internet, “Two French virtual influencers are finalists in the world Miss AI contest”, June 5, 2024
• Marks & Spencer, “Marks & Spencer experiments with AR to launch its first virtual influencer”, October 27, 2022
• Legifrance, “Article 5 of Law no. 2023-451 of 9 June 2023 (French influencer law), version in force since 8 November 2024”, November 8, 2024
• Future of Life Institute (unofficial mirror), “Article 50: Transparency Obligations for Providers and Deployers of Certain AI Systems (read on 3 November 2026)”, November 3, 2026
• European Commission, “Code of practice on the transparency of AI-generated content”, June 10, 2026
• French Senate, “Transparency of AI use in advertising and commercial images (written question and answer)”, May 1, 2026

Describe your product: BriefBench searches live on Instagram, TikTok and YouTube and rates each account’s engagement, and the authenticity of Instagram accounts. The AI clause is written in your own brief and contract.