The European AI Act, coming soon
- OneFifty Consultancy
- 1 day ago
- 2 min read
The AI Act is the first-ever legal documentation on AI, which addresses the rules and regulations around it, and what needs to be disclosed on social media. It becomes applicable on 2nd August 2026.
Who this applies to:
EU-based organisations and EU-based organisations operating in the UK
This applies directly. From 2 August 2026, you are legally considered an “operator” of any AI system used in your business activities, including AI-generated content produced by agencies on your behalf. Compliance sits with you as the brand.
UK-based organisations
If you’re operating in Europe: Brexit does not create an exemption. The EU AI Act has explicit extraterritorial scope: if your brand’s content, campaigns, or communications reach EU audiences, the regulation applies to that activity. UK businesses working with EU markets, subsidiaries, or consumers are firmly in scope.
If you’re only operating in the UK: This wouldn’t apply. However, many of the provisions are reasonable good practice, in the way that images already often feature disclaimers if they’re not the exact model you’d receive in a showroom
How it works, and what it changes:
AI-generated or manipulated images, video, and audio must be clearly labelled if they could reasonably be perceived as real by an average viewer - think realistic car imagery, AI-generated people, or AI voiceovers. Clearly stylised or obviously unrealistic content is typically exempt.
AI-generated text does not require a disclaimer if it has been reviewed, edited where necessary, and approved by a human. Standard AI-assisted copywriting with human sign-off is fine
AI chatbots, voicebots, and human-like avatars must disclose to users that they are interacting with AI, not a person
Type of Content | Transparency Obligation applicable? | Sample Disclaimer |
AI-generated images, videos and audio (“deepfakes”) | Yes, if content is a deepfake (= content appears authentic or truthful)
No, if unrealistic or artistic representations à In case of doubt, transparency obligations should be fulfilled.
| “AI-generated image”
“This video was created using AI”
“AI-generated content” |
AI-generated text | Yes, if texts are published without human oversight or content is clearly presented as purely AI-generated without editorial responsibility
No, if text is reviewed, adjusted if necessary and approved by a human à human assumes responsibility; AI may be used for drafting or ideation
| “This text was created with the assistance of AI”
“AI-generated Text” |
AI-Chat- and Voicebots | Yes, if user is interacting with an AI-Chat- or Voicebot
| “You are chatting with an AI-powered chatbot”.
“You are interacting with an AI-powered voicebot”. |
AI-Avatars | cf. transparency obligations for AI-generated images, videos and audio (“deepfakes”)
cf. transparency obligations for AI chat- and voicebots where an AI avatar is used as a chat- or voicebot. | cf. transparency obligations for AI-generated images, videos and audio (“deepfakes”)
cf. transparency obligations for AI chat- and voicebots where an AI avatar is used as a chat- or voicebot. |
What's complicated with the act?
It’s not very clear what using standard editing tools in Adobe which utilise AI means in practice. E.g. if one colour grades a video, and lets the software optimise it, does that count as AI? Or an image being auto resized? We suspect there will be further detailed guidance as the deadline approaches on some of these nuanced points.
Comments