Guides · For brands and retailers
Which AI tools are there for influencer marketing, and what are you allowed to automate with them?
AI tools for influencer marketing cover creator search (HypeAuditor, Kolsquare, Modash), outreach (GRIN, Upfluence) and ready-made videos (Creatify, HeyGen). Search, drafts and reports can largely be automated, but saying yes or no to creators and signing off content should stay with a person. Since 2 August 2026, Article 50 of the EU AI Act has applied: if an AI character in an advert looks real, it must be disclosed as artificial. Another route is wyrld Boost, where creators show your product to their own community themselves.
By the wyrld team · Last updated: · 10 min read
With wyrld Boost, many creators feature your products at the same time on Instagram, TikTok or YouTube. Your budget is fixed up front, and the creators are paid from it by results.
Key points
- Article 50 of the AI Act has applied since 2 August 2026. Fines: up to €15m or 3% of global turnover, whichever is higher; for SMEs and small mid-caps, the lower.
- Under EU guidelines of 20 July 2026, a realistic AI influencer that only demonstrates a sponsored product’s functions is a deep fake, not a creative work.
- Under the EU guidelines, merely hiring an agency without deciding if or how it uses AI does not make you a deployer. Deciding and controlling it can.
- Germany’s Data Protection Conference prefers AI services that do not train on your inputs. Removing names and addresses is regularly not enough.
On this page
- What AI can take on for each task, and what you check yourself
- Seven AI tools and a campaign service: function and price as stated by each provider
- Deep fake or not: classifying AI content under Article 50(4) of the AI Act
- Who labels AI content: the brand, the agency or the creator
- Invented testimonials: an AI label won’t rescue them
- Creator data in ChatGPT and other AI services: five questions to settle first
- When you want real people to show your product
What AI can take on for each task, and what you check yourself
| Task | What AI takes on | What you check | Legal basis |
|---|---|---|---|
| Finding creators | Plain-language search, lookalike profiles, shortlists | Audience, numbers, past ads | GDPR: legal basis, notice under Art. 14 |
| Vetting profiles | Scanning posts for risks | Look at the flags, make the yes or no yourself | Art. 22 GDPR, where a decision has legal effect |
| Outreach | Drafts, personalisation, sending, follow-ups | Recipient list, consent, claims in the text | § 7(2) no. 2 UWG, Art. 50(1) AI Act |
| Briefing | Draft from product data and goal | Product claims, ad disclosure, AI rules for creators | § 5 and § 5a(4) UWG |
| Reporting | Capturing posts, building reports | Spot checks against shop figures and returns | § 5 UWG, once you advertise the figures |
| Generating videos | Avatar, voice, editing, variants | Deep fake question, consent, true product image | Art. 50(4) AI Act, § 22 KunstUrhG, § 5 UWG |
| Testimonials | Nothing: an AI has no experience | Only real statements from real people | § 5 UWG, Annex no. 23c UWG |
With AI in influencer marketing, the line falls between groundwork and decisions. Anything that just gives you material can largely be automated. As soon as a person is affected or something goes public, the final check belongs with a human. GRIN says you choose for yourself how much of its AI agent Gia’s work you review. We would start with search and reporting, and only automate outreach and video once consent and the deep fake question are settled. We would not leave saying yes or no to creators to any tool.
If an AI agent writes to people on its own, the EU guidelines say it must disclose that it is artificial and on whose behalf it is acting, for example with an AI notice at the top of the email. That duty under Article 50(1) of the AI Act lies with the agent’s provider, so ask them about it.
Seven AI tools and a campaign service: function and price as stated by each provider
| Provider | Task | Function, as stated by the provider | Price, as stated by the provider |
|---|---|---|---|
| wyrld Boost | Campaign with many creators | Creators are selected for you; those who join feature your product in their own videos | You set the budget, usually from €10,000 plus VAT |
| Creatify | Ad videos | AI avatars, realistic “AI Actors” | Standard $10 a month ($15 struck through), discount on the annual plan |
| GRIN | Programmes from search to reporting | AI agent Gia finds creators, writes and sends outreach | Free $0 plus a 20% fee on wallet deposits; Starter $200 a month plus 10% |
| HeyGen | Avatar videos | AI avatars, over 500 ready-made presenters | Free $0 with watermark; Creator $29 a month |
| HypeAuditor | Search, vetting, outreach | AI Scout from a brief, one-click AI outreach | no price on its pricing page; demo available |
| Kolsquare | Search, reporting | AI search with image and speech recognition | no price given; demo available |
| Modash | Search, management | AI search, connection to AI tools via MCP | Essentials $199 a month, billed annually |
| Upfluence | Programmes from search to payment | AI agent Jaice AI, AI review of posts | custom quote by module |
The selection comes from wyrld, which publishes this page and is a provider itself. It is not complete and not a ranking: after our own offer, the order is alphabetical. Functions and prices are taken from the providers’ own pages, read on 4 November 2026.
Deep fake or not: classifying AI content under Article 50(4) of the AI Act
| Content | Deep fake? | What it means for you | Reference |
|---|---|---|---|
| A realistic AI influencer only demonstrates a sponsored product’s functions, or simulated customers show it teleshopping-style | Yes, not a creative work | Disclose; check § 5 UWG | Art. 50(4), guidelines 6.1.3 |
| AI depiction of a well-known influencer in an advert | Yes | Disclose; settle consent beforehand | Guidelines 6.1.1, § 22 KunstUrhG |
| AI image makes the product look nicer or better than it is | Yes | Disclosure does not cure misleading content | Guidelines 6.1.1, § 5 UWG |
| Real product against an AI background, no false impression | No | No duty under para. 4 | Guidelines 6.1.1 |
| Lighting, sound or colour improved in a real video | Probably not, depends on the case | Check if the impression of the product changes | Guidelines para. 116 |
| AI ad copy without claims on, say, health, consumer safety or sustainability | Not a para. 4 case | No AI label under para. 4 | Guidelines 6.2.1 |
| AI copy with such claims | May be covered | Disclose, unless a person reviews it and takes responsibility | Art. 50(4), guidelines 6.2.1 |
References as of: AI Act, Regulation (EU) 2024/1689, consolidated version of 27 July 2026; European Commission guidelines C(2026) 5054 of 20 July 2026; UWG as last amended by the act of 22 October 2026; all read on 4 November 2026. The guidelines are not binding: only the Court of Justice of the European Union can interpret the AI Act authoritatively.
Article 3(60) of the AI Act defines a deep fake as AI-generated or manipulated image, audio or video content that resembles existing persons, objects, places, entities or events and would falsely appear authentic. According to the guidelines, an entirely invented but realistic person is enough, and your intention is irrelevant. Their example of something that is not a deep fake: mice squabbling in a cheese advert. The lighter disclosure rule for creative works covers advertising only in certain cases.
Being an EU regulation, the AI Act applies in Germany too. The guidelines say deep fakes generated before 2 August 2026 need not be labelled retrospectively. The transition period to 2 December 2026 under Article 111(4) covers only providers of existing systems, and only the technical marking under Article 50(2), not disclosure by deployers.
Who labels AI content: the brand, the agency or the creator
Under Article 50(4) of the AI Act, disclosing a deep fake is the job of the deployer, meaning whoever uses an AI system under their own authority in a professional capacity. If your team generates the video, that is your brand, even when freelancers run the tool under your control. If you only hire an agency and do not decide if or how it uses AI, the guidelines say you are not a deployer. If a creator makes the post with AI and decides that for themselves, the definition points to them.
Even as a client only, a brand is not entirely off the hook: under § 8(2) UWG (the German Unfair Competition Act), claims for an injunction and removal can be brought against it too if someone it has engaged breaches the UWG. So the brief should say if AI may be used and who labels it. The flip side: whoever decides and controls if and how AI is used can become a deployer under the guidelines.
A deep fake has to be disclosed at the first exposure at the latest, for instance at the start of the video or while scrolling through the feed. A tool’s invisible marking is not enough; the notice has to be perceptible without technical aids. The EU icon for AI content is optional; the duty is not. The advertising label is still needed as well: Article 50(6) leaves other transparency obligations untouched, and § 5a(4) UWG still requires the commercial purpose to be made clear unless it is immediately apparent from the circumstances.
Platform rules apply on top of the AI Act. TikTok requires a label on AI-generated or significantly edited content showing realistic-looking scenes or people; YouTube requires disclosure for, among other things, realistic-looking scenes that never happened. According to the guidelines, a deployer may use a very large platform’s labelling tool, but the responsibility stays with the deployer.
Invented testimonials: an AI label won’t rescue them
An AI avatar has never worn your jacket. If it says “I’ve been wearing it for three months” anyway, it is describing an experience that never happened. Under § 5(2) UWG, a statement is misleading if it contains untrue information or information likely to deceive, for example about the main characteristics of the goods.
Annex no. 23c UWG lists commissioning fake consumer recommendations and misrepresenting them on social media. Under § 3(3) UWG, anything in the annex is always unlawful towards consumers, and simulated customers talking about their own use can fall under it. The EU guidelines on the AI Act make clear that the labelling duty does not make misleading advertising lawful.
Creator data in ChatGPT and other AI services: five questions to settle first
Creators’ names, profiles and email addresses are usually personal data. In its guidance of 6 May 2024, the DSK (Datenschutzkonferenz, the joint body of Germany’s data protection authorities) sets out what to clarify before you start. Article 4 of the AI Act also requires measures that build your team’s AI literacy.
- Are inputs and outputs used for training? The DSK says services that don’t do this are preferable.
- Do you have a legal basis, and do the creators know? The DSK requires both for entering data and passing it to the provider.
- Is the provider your processor? With cloud services it often is, in which case you sign an agreement under Art. 28(3) GDPR.
- Is the input really free of personal data? Removing names and addresses is regularly not enough, as the context can still identify someone.
- Does a person make the final decision? Under Art. 22 GDPR, decisions with legal effect may, as a rule, only be made by people, and purely nominal involvement is not enough.
When you want real people to show your product
AI can find and contact creators, but posting is up to them, and an avatar from a video tool brings no community of its own. With wyrld Boost, the creators who take part show your product to their followers in their own videos; wyrld picks them, and they are paid from your budget by the KPIs reached. It is the wrong route if you need videos for your own ads or want to rework them with AI. The videos run on the creators’ channels. The right to reuse them yourself is not part of the package.
wyrld Boost at a glance
- What it is
- The campaign service from wyrld, a social commerce platform based in Hamburg, Germany. Many creators feature your products at the same time, so you do not have to book each one. People buy in your own shop, on Amazon or through price comparison: you decide.
- Price
- You set the budget. Campaigns usually start from €10,000 plus VAT. Creators are paid from the budget according to the KPIs they reach, such as purchases, clicks or watch time. The package price also covers selection, payouts, reporting and a short consultation.
- Creators and videos
- wyrld selects suitable creators, and each creator decides whether to take part. Creators label campaign videos as advertising. The videos run on the creators’ channels. The right to reuse them yourself is not part of the package.
- How it runs
- Depending on the campaign, you approve the videos before they go live, or the creators post from your brief. We set out the minimum scope of the campaign in your offer. We do not promise reach or revenue. After the campaign you can see which creator delivered which results.
- Who it is for
- wyrld Boost is open to brands of every size and in all categories, including brands from abroad that want to sell in Germany, Austria or Switzerland, provided the product and its advertising are lawful. Agencies book as normal customers.
- Brand and shop subscription
- Separately, this shows you every day which creators feature your products and, where the shop reports purchases, who is selling. Price on request.
Frequently asked questions.
Do I have to label AI-generated influencer videos?
Yes, if the video is a deep fake under Article 3(60) of the EU AI Act, meaning it falsely appears real. The EU guidelines give two examples: a realistic AI influencer that only demonstrates a sponsored product’s functions, and simulated customers in a teleshopping style. The deployer of the AI system has to disclose it, and the advertising label under § 5a(4) UWG comes on top.
Can I use AI to contact creators automatically?
The sending is more of an issue than the tool. In Germany, § 7(2) no. 2 UWG treats advertising by email without prior express consent as an unreasonable nuisance; para. 3 has a narrow exception for existing customers. Get legal advice beforehand on when a collaboration request counts as advertising. Application pages where creators put themselves forward are less of a problem.
Can I recreate a creator’s face or voice with AI?
As a rule, only with consent that covers exactly that purpose. Under § 22 KunstUrhG (the German Art Copyright Act), portraits may only be distributed with consent, and personality rights and data protection come on top. Don’t assume the fee for a filmed video covers an AI replica, and put the purpose and duration in writing. In advertising, the result is usually a deep fake.
Last updated:
Sources
Legal information as of 4 November 2026. This page explains the AI Act, the UWG and data protection in Germany in general terms and is not legal advice. It is published by wyrld, which is itself a provider through wyrld Boost, and the list of tools is not complete. Accessed on :
- AI Act, Regulation (EU) 2024/1689, Article 50, consolidated version of 27 July 2026, European Commission AI Act Service Desk
- AI Act, Article 3 (definitions of deployer and deep fake)
- AI Act, Article 4 (AI literacy)
- AI Act, Article 99 (penalties)
- AI Act, Article 111 (transitional rules)
- AI Act, Article 113 (application)
- European Commission: guidelines on the Article 50 transparency obligations, C(2026) 5054 of 20 July 2026
- European Commission: EU icons for labelling AI-generated content
- UWG (German Unfair Competition Act), § 3
- UWG, § 5
- UWG, § 5a
- UWG, § 7
- UWG, § 8
- UWG, annex to § 3(3)
- German Art Copyright Act (KunstUrhG), § 22
- German data protection conference (DSK): guidance on AI and data protection, version 1.0 of 6 May 2024
- TikTok: Community Guidelines, integrity and authenticity
- YouTube Help: disclosing the use of content made with generative AI
- Creatify: pricing
- GRIN: pricing
- HeyGen: pricing
- HypeAuditor: platform
- HypeAuditor: pricing
- Kolsquare: pricing
- Modash: pricing
- Upfluence: pricing