AI Transparency Rules for Marketing Content
From 2 August 2026, Article 50 of the EU AI Act introduced transparency duties for selected uses of AI-generated and AI-manipulated content. 1
This does not mean labels are needed for every AI-assisted image, email or edit. It does mean commercial teams need a more deliberate review process.
The central concept is the deepfake: image, audio or video generated or manipulated by AI that resembles existing people, objects, places, entities or events and would falsely appear authentic to a viewer. 2
If a business deploys AI to create this type of content, it must disclose that the material has been artificially generated or manipulated. The disclosure must be clear, distinguishable and accessible no later than the viewer’s first exposure.
For marketing teams, the question is where editing ends and manipulation begins. Generating a wholly fictional character is not automatically the same as presenting a real person as having said or done something they did not. Basic editing is not necessarily the target either: Article 50 recognises an exception for AI used as an assistive function for standard editing, or where input is not substantially altered.
The line is contextual. An AI-enhanced product visual, transformed before-and-after image, or travel video that makes a real destination look materially different deserves closer scrutiny.
Our recommendation, echoed by lawyer Nicholas Ørum Keller in a recent episode of Marketingpod, is to focus first on the risk of deception rather than the mere presence of AI.
- Ask whether the asset presents a real person, product, place or event in a way that may be mistaken for reality.
- Ask whether the alteration changes the core consumer impression.
- And ask whether the disclosure is visible where the audience actually sees the asset, rather than buried on a landing page or at the end of a long video.
Teams should map their AI-enabled creative workflows, establish an approval rule for high-risk assets, and retain a record of what was generated or materially changed. Where disclosure is required, the European Commission’s optional AI icons can be helpful, but an icon alone does not establish compliance. The requirement is effective disclosure, not one particular graphic.
AI-generated text is narrower. Article 50 addresses text published to inform the public on matters of public interest when it has not received human review or editorial control. It does not create a blanket duty to label routine marketing copy, and human review by a responsible editor is expressly relevant to the exemption.
Transparency under the AI Act also sits alongside existing consumer and marketing law. A label is no licence to mislead. Brands will navigate the change best by using AI where it adds value, preserving human judgement, and treating authenticity as part of their creative standard.
This article provides general information rather than legal advice. Assess specific campaigns with qualified legal counsel.
