AI ACT AND GOVERNANCE: WHAT CHANGES FOR BRANDS
Artificial intelligence has now become structurally embedded in marketing and communication processes: from copywriting to the production of images and videos, from adapting creative assets to modifying and optimizing existing content.
As of 2 August 2026, the provisions of the AI Act apply, including specific transparency obligations concerning content generated or manipulated through artificial intelligence.
This marks an important step for a sector in which tools, production methods and creative processes have evolved rapidly in recent years. For users, it means being able to understand more easily the nature of the content they encounter. For brands, agencies and other market players, it means having a clearer framework for assessing how AI can be used and what responsibilities are associated with different applications.
More broadly, the AI Act distinguishes between providers, meaning entities that develop or make AI systems available, and deployers, meaning organizations that use AI systems as part of their activities.
For brands, it is therefore becoming increasingly important to understand how AI is used within processes, what impact it has on outputs, and where responsibility for the final content lies.
STARTING FROM THE OUTPUT
The same AI tool can be used for a simple technical correction or to generate an entire scene. For images, videos and audio, it is therefore important to assess not only which tool has been used, but also what has actually been produced and how the content may be perceived by the user.
In marketing, this issue becomes particularly relevant when AI is used to modify product images, environments or people.
A product placed against a graphic background has a different impact from the same product placed within a photorealistic scene that never actually existed, especially when the setting may communicate information about proportions, usage or product characteristics.
The assessment should therefore focus on the effect that the generated or manipulated component has on the overall perception of the content.
This does not mean that every piece of content created using AI must automatically be labelled. Rather, it means that the use of artificial intelligence should be assessed in relation to the nature of the output, the type of intervention carried out and the specific circumstances in which the content is published.
Having clear criteria allows organizations to distinguish between ordinary uses and those requiring greater attention, while continuing to experiment with AI within a more clearly defined framework.
TRANSPARENCY AND DISCLOSURE
Article 50 of the AI Act sets out specific transparency obligations, including those relating to artificially generated or manipulated audiovisual content that may appear authentic, as well as AI-generated or manipulated text intended to inform the public on matters of public interest, under the conditions established by the regulation.
Where disclosure is required, it must be clear and distinguishable, enabling users to understand the artificial or manipulated nature of the content.
This principle primarily protects individuals by making it easier to recognize synthetic or significantly altered content. At the same time, it also provides greater certainty for those producing and publishing content, by making the relevant criteria more explicit across different scenarios.
Platforms may also use their own systems for labelling and identifying AI-generated or manipulated content. These tools, however, do not automatically replace the assessment that must be carried out as part of the content production and publishing process.
For this reason, checks should be integrated upstream into asset production and approval processes, rather than being treated solely as a compliance step after content has already been created.
WHAT CHANGES FOR TEXT CONTENT
Text-based content requires a particularly precise interpretation.
Article 50 does not introduce a general labelling requirement for every piece of text produced through a generative model. The provisions apply, in particular, to AI-generated or manipulated content intended to inform the public on matters of public interest, in the cases and under the conditions established by the regulation.
In this context, effective human review also plays an important role, together with the assumption of editorial responsibility for the content.
This reinforces the role of human oversight in processes where AI is used on an ongoing basis, particularly when the content has significant informational relevance or requires editorial judgment.
Here too, the key issue is not simply whether AI has been used, but how it has been used and who holds responsibility for the final content.
AI can support the creative and production process; however, the accuracy, quality and appropriateness of the output must remain part of a clearly defined editorial responsibility.
GOVERNANCE STARTS WITH PROCESSES
For companies producing large volumes of content, the real challenge is turning these assessments into a simple, repeatable and proportionate process.
It is therefore useful to define a shared framework that makes it possible to understand:
- what type of content is being produced;
- how AI has been used;
- which elements have been generated or significantly modified;
- what effect the intervention has on the final content;
- whether the conditions requiring disclosure are met;
- who is responsible for verification and final approval.
Regulation can provide the starting point, but governance also requires internal procedures and operational criteria that make these assessments applicable in day-to-day activities.
In this context, where appropriate, it becomes useful to track the origin of the asset, the type of AI intervention carried out and the decision taken regarding disclosure.
The same approach should also be applied to content produced by agencies, creators, influencers and other external partners. An asset coming from outside the organization should not automatically be considered outside the brand’s governance processes. In order to assess it correctly, it is necessary to understand, at least to the relevant extent, how it was produced and what AI interventions were involved.
It is precisely at this level that Intarget is integrating these principles into its own content production and approval workflows, defining criteria and control steps designed to make it easier for teams and clients to understand how an asset was created, which checks were carried out and when additional levels of attention may be required.
FROM COMPLIANCE TO GOVERNANCE
The AI Act makes more concrete a transition that many organizations are already addressing: defining how artificial intelligence enters business processes, which uses require greater attention and where responsibilities should sit.
For brands, this also represents an opportunity for greater certainty.
Clearer rules make it possible to use AI with greater awareness, reducing areas of uncertainty and creating more solid conditions for continued innovation.
The answer, therefore, is not to limit the use of AI, but to build a shared framework involving marketing, legal, IT and external partners, with clear criteria for the production, review and approval of content.
