AI in advertising in Belgium: the Advertising Council publishes new guidelines

On 24 August 2026, the Belgian Advertising Council (Conseil de la Publicité) and its member associations, including the UBA, published a new version of their guidelines on the use of artificial intelligence in advertising in Belgium. The text clarifies two things: when AI-generated or AI-altered advertising content must be disclosed as such, and which principles govern responsible use. Here is what this framework changes for businesses that advertise in Belgium.

1. Using AI does not automatically mean labelling it

The guidelines' first message: using AI to produce or alter an advertising visual, video or sound does not automatically require an AI label. Disclosure becomes necessary when the content looks realistic enough to be mistaken for authentic, and could influence how the consumer understands or perceives the ad. The guide proposes three simple questions to decide: to what extent was AI used, does the content look realistic and plausible, and could it influence the consumer's perception.

2. A decision tree for real-world cases

Beyond these three questions, the guidelines provide a decision tree and concrete disclosure examples for images, videos, audio content, out-of-home advertising (OOH) and AI-based interactions, such as a chatbot or virtual assistant on a website. This clears up doubt on frequent cases: a retouched product image, a synthetic voice on the radio, or a virtual spokesperson.

3. 12 principles that go beyond transparency alone

Transparency is only one part of the framework. The updated text also sets out 12 principles for responsible AI, covering in particular human oversight, internal AI governance and culture, inclusion, protection of vulnerable audiences, environmental impact, privacy, intellectual property, usage monitoring, fair competition and incident management. Together, they map out a balance between innovation, transparency, consumer protection and accountability.

4. What this concretely changes for Belgian businesses

For a business advertising in Belgium, this framework does not ban anything: it sets a test to apply before publication. An AI-generated product image, a synthetic voice for a radio spot, a virtual model or an advertising chatbot each deserve to be run through the guide's three questions. This framework builds on the transparency obligations of the European AI Act, applicable since 2 August 2026, which already require labelling for certain AI-generated content.

The AI view, humans first

What these guidelines formalise, Vistalaro already applies internally: human oversight, the first of the 12 principles, is not a box to tick, it is a production step. An AI tool can generate a draft visual or a synthetic voice in seconds. But deciding whether that content should be disclosed, reworked or rejected remains a human judgement, based on the brand's context and the intent of the message. AI executes the production. Expertise decides, and keeps watch so that transparency stays the rule rather than the exception.