AI Act and generative content: European retail in the balance

The context: when the norm does not know its object

The European AI Act came into effect in 2024 and began its application rollout in 2025. However, one of its most critical points remains unresolved. The regulation imposes transparency obligations for synthetic content—including what can be defined as deepfakes. The problem is that the operational definition of deepfakes is not unambiguous in the legislative text.

This ambiguity is not a minor technical detail. In fact, it directly affects millions of euros in advertising investments in European retail. Companies that already produce product images, digital scenery, and virtual models through AI are faced with an unanswered question: are we obligated to disclose it?

The original source of this story is an analysis published by The Decoder, which highlighted the pressure exerted by Eurocommerce on the European Commission. The topic is set to remain central throughout 2026 and likely until 2027.

The Numbers Reinventing Retail Marketing

The most significant data comes from Zalando. The e-commerce platform states that the 90% Marketing Content published on their platform is already generated by artificial intelligence. This is not an exception: it is the direction of the entire industry.

Therefore, when discussing AI Act compliance in retail, it's not about edge cases. It's about the operational core of content strategies for large and medium-sized companies. Furthermore, the phenomenon doesn't just affect the big players. Italian SMEs in retail—often with limited budgets—are also adopting generative tools for producing product images, banners, and creative assets for campaigns.

According to data Gartner, by 2027, more than 30% of all global marketing content will be generated by AI. As a result, the issue of European regulation is not a future concern—it is already present in today’s creative pipelines.

Eurocommerce's position: a couch image is not a deepfake

Eurocommerce, the association representing giants like Amazon, H&M, and IKEA, has submitted a formal request to the European Commission. The goal is to obtain an explicit exemption for AI-generated images used for commercial advertising purposes.

The argument is logically sound. A virtual living room created by a generative model to contextualize a sofa for sale does not deceive anyone about the identity of a real person. It does not manipulate faces, alter statements, or create false identities. On the contrary, it is a digital visual merchandising tool, analogous to a studio photograph with a constructed set.

However, the European Commission has not yet formally incorporated this distinction. Thus, the regulatory gap persists. And in the absence of clarity, companies' legal and marketing teams tend to behave defensively—slowing down adoption or adding unnecessary disclaimers.

The operational gray area for marketing managers

For an Italian company's marketing or digital manager, the problem translates into concrete questions. Do you need to include a disclosure every time an AI-generated image is used in a Google Ads campaign? Does a banner with a virtual model violate transparency norms? Does an AI-retouched product photo fall under the definition of synthetic content?

At this moment, the answers are not clear-cut. Therefore, companies find themselves having to make compliance choices under conditions of uncertainty. This comes at a cost: it slows down creative processes, increases legal work, and in some cases leads to foregoing more efficient AI solutions to avoid unquantified regulatory risks.

We of SHM Studio we observe this phenomenon directly in our work with clients. Specifically, in the planning stages of Google Ads campaigns and of the LinkedIn campaign, the need to define an internal policy on AI-generated content is increasingly emerging, even before addressing the creative strategy.

Strategic Reading: Three Possible Scenarios for 2026-2027

Analyzing the evolution of the regulatory framework, it is possible to identify three plausible scenarios for the next 18 months.

• Scenario A — Commercial Exemption Approved: The Commission welcomes Eurocommerce's request and introduces an explicit distinction between deepfakes and AI-generated content for commercial advertising use. In this case, retail can continue to scale generative production without additional disclosure obligations.
• Scenario B - Generalized Labeling Obligation: The EU has decided to implement a labeling requirement for all AI-generated content, regardless of context. Consequently, every generative image in an advertisement will need to carry an explicit disclaimer. The impact on user experience and campaign CTR could be significant.
• Scenario C - Prolonged regulatory stalemate: The Commission does not provide operational clarifications in the short term. Therefore, each Member State interprets autonomously, creating a regulatory patchwork that complicates pan-European strategies.

Scenario C is, at the moment, the most likely in the short term. According to an analysis by Harvard Business Review, companies operating in ambiguous regulatory contexts tend to overestimate compliance risks, reducing the speed of technological adoption compared to competitors in markets with clearer rules.

Operational implications for content strategies

On a practical level, Italian marketing managers must face this scenario with a two-tiered strategy.

The first level concerns the Internal AI content governance. First of all, it is necessary to map which assets of your creative production are generated—even partially—by generative AI tools. This includes produced images, virtual backgrounds, digital models, generated texts, and synthetic videos. Subsequently, it is advisable to define a company policy that establishes when and how to declare the use of AI in advertising content.

The second level concerns the Campaign flexibility. The strategies of digital marketing must be designed with the ability to quickly adapt creative formats in case of new regulatory requirements. Similarly, strategies SEO generative content needs to consider the possibility of transparency requirements that impact organic ranking.

Besides this, who manages the production of copywriting Visual content should begin documenting the creative process. This documentation could become relevant in the event of a compliance audit.

The construction site is still open: what is really missing from European standards

The fundamental problem is not technical: it is conceptual. The AI Act was written with a legitimate concern—protecting citizens from information manipulation. However, the legislator did not distinguish with sufficient precision between identity manipulation and AI-assisted creative production.

This distinction is fundamental. A deepfake that alters a politician's face to attribute false statements to them is a democratic problem. An AI-generated image of an apartment used to sell a sofa is a visual merchandising tool. Treating them with the same regulatory framework creates operational distortions without producing real benefits for consumers.

Indeed, retail consumers don't ask if the living room in the photo is real or generated. They ask that the product matches the description. Therefore, the relevant transparency in this context is about the product's quality and features—not the origin of the contextual image.

We of SHM Studio We believe the European debate needs to shift towards a more precise taxonomy of AI-generated content, differentiating by purpose, context, and potential for deception. Without this distinction, the risk is slowing down innovation in marketing without truly protecting anyone.

Next moves: what to monitor in the coming months

For marketing managers and digital managers in Italian companies, there are some deadlines and developments to keep an eye on.

• AI Act Implementing Guidelines (Second Half of 2026): The European Commission is expected to produce operational guidance on the categories of content subject to transparency obligations. These documents will be crucial for campaign compliance.
• Advertising platform positioning Google and Meta are developing their own policies on AI-generated content in ads. In particular, their technical choices could anticipate or bypass the European regulatory framework.
• Evolution of tools AI for marketing Major vendors are already integrating watermarking and metadata features for generative content. Adopting tools that support native traceability is a prudent strategic choice.

Whoever wants to delve deeper into how to integrate these considerations into their strategy digital presence or in advertising campaigns can contact our team at Contact Us the insights of the SHM Studio Blog.