Since 2 August 2026, the European AI Act has required transparency for advertising content classed as a “deepfake”, generated or manipulated by artificial intelligence. On 25 August 2026, the Belgian Advertising Council translated these rules into a practical framework for the industry. For a Belgian brand using AI in its creative work, the question is no longer whether disclosure is needed, but when.
1. What became mandatory on 2 August
2 August 2026 marked the start of a new phase of the European AI Act: the transparency obligations under its Article 50 became binding. Any advertising content classed as a “deepfake”, a visual, video or audio piece generated or manipulated by AI realistic enough to be mistaken for authentic, must now be disclosed to the public from its first exposure. A fictional customer testimonial presented as real, a synthetic voice imitating a real person, or a product demo entirely generated by AI can all fall within this scope. The disclosure must be clear and visible, without the public having to dig through metadata to find it.
2. Belgium's 25 August framework: a decision tree
On 25 August 2026, the Belgian Advertising Council (Conseil de la Publicité / Raad voor Reclame) published updated guidelines translating these European rules into concrete tools. The new framework proposes a three-factor assessment: how much AI shaped the final content, how plausible and realistic it looks, and its potential influence on consumer perception. Once these three factors point to a risk of confusion about the message's authenticity, disclosure becomes necessary, with recommendations tailored to the medium: online image, outdoor advertising, video or audio.
3. What stays outside the disclosure requirement
Not every use of AI is covered. Preparatory steps, such as idea generation, scriptwriting, storyboarding, translation, optimisation or technical touch-ups like noise reduction or colour correction, generally require no disclosure at all. The line sits at the final content shown to the public: once AI-generated material could be mistaken for authentic and distort how a product or message is perceived, transparency becomes mandatory again, on top of existing rules against misleading advertising, which remain fully in force.
The AI lens, humans first
An AI tool can generate a campaign visual in minutes, propose ten variations of a headline, or flag that an image has been altered. But judging whether that content crosses the AI Act's “deepfake” line, deciding where to place the disclosure, and checking that the message stays honest, remains a human, case-by-case call. AI executes the production. Expertise decides what must be disclosed, and owns compliance.
This week
Review your last few visual or video campaigns: identify any that use AI-generated visuals or voices, and check whether disclosure is required under the Advertising Council's decision tree.
With Vistalaro Studio, we apply this framework from the moment we create your visuals and videos: when AI plays a role in production, we assess with you whether disclosure is required, without slowing down the creative work. With Vistalaro Pilot, your dedicated marketer keeps continuous watch on these regulatory obligations and adjusts your content strategy accordingly.
Is your AI content compliant?
We review your visuals, videos and campaigns together, no jargon.
Let's talk