Pubblicità e AI Generativa: Chi è responsabile?
1. Introduction
Generative artificial intelligence is reshaping the advertising industry not only because it accelerates content production, but because it profoundly alters the geography of liability.
The issue is no longer simply who “created” the message, but who enabled it, who selected it, who distributed it, who targeted it, and who ultimately benefited from its circulation.
The phenomenon emerges particularly clearly in misleading campaigns that exploit the images, voices, and likenesses of well-known individuals in order to disseminate false promotional or pseudo-informational messages. In such scenarios, generative AI is not merely a creative tool: it becomes a multiplier of credibility, capable of simulating the authority of journalists, economists, politicians, or public figures without any authorization and with a level of realism that makes it increasingly difficult for the average user to distinguish authentic content from manipulated material.
The traditional advertising model, based on a relatively linear supply chain, is therefore being replaced by a complex and layered ecosystem in which the final content arises from the interaction between human input, algorithmic automation, distribution platforms, and monetization infrastructures. When something goes wrong, copyright infringement, unauthorized exploitation of image rights, misleading advertising, unlawful processing of personal data, liability does not disappear; rather, it becomes fragmented and dispersed throughout the entire chain.
2. The AI-Driven Advertising Supply Chain
Advertising based on generative AI operates within a sophisticated and no longer linear supply chain, in which each participant contributes to the final outcome without exercising full and effective control over it. The traditional creative process is progressively disaggregated and distributed among multiple actors, each holding only a limited portion of the decision-making power.
The brand or final client continues to exercise strategic control by defining objectives, tone of voice, and advertising messages. In practice, however, its involvement usually stops before the technical generation phase.
The center of gravity therefore shifts toward the creative agency, whose role becomes increasingly central and hybrid. It is here that prompts are structured, AI tools are selected, and decisive choices are made regarding which outputs will ultimately be used.
At the top of the system are the AI providers, which supply both the technological infrastructure and the generation models. Although they do not directly participate in creative decisions, they concretely determine the characteristics, limitations, and risks of the output.
The result is a structure profoundly different from the traditional one, in which an increasingly evident dissociation emerges: those who “create” are not the same as those who “control,” and those who “control” are not necessarily those who will ultimately be held liable.
3. Liability for Generated Content
The first issue concerns the legal qualification of AI-generated output under copyright law.
Italian case law has begun to show significant openness. In particular, the Italian Supreme Court (Corte di Cassazione), with Order No. 1107/2026, stated that the use of artificial intelligence tools does not, in itself, exclude copyright protection, provided that an effective human creative contribution can be identified within the generative process.
In this context, prompts assume increasing relevance. Where they display a significant degree of structure, selection, and creative intentionality, they may be considered an integral part of the authorship process.
At the same time, infringement of third-party rights represents one of the most concrete operational risks. Generative AI now makes it possible to create content that evokes, in a recognizable way, the identity, tone, reputation, and professional context of real individuals.
From the perspective of external liability attribution, one essential principle remains unchanged: the fact that the harmful element was generated by the model does not interrupt the chain of liability for those who selected, approved, and disseminated the output.
4. Synthetic Advertising, Exploitation of Reputation and Trust Hijacking
One of the most insidious aspects of generative AI in advertising is the possibility of transforming another person’s identity into an artificial vector of trust.
The phenomenon may be described as a form of “trust hijacking”: the advertising message persuades not because of its intrinsic qualities, but because it simulates the authority of the person who apparently endorses it.
From a legal perspective, these scenarios trigger multiple forms of protection simultaneously, including image rights, personal identity, consumer protection law, misleading advertising rules, and privacy regulations.
5. The Critical Role of AI Platforms
AI platforms represent the technological core of the system, but they do not exhaust the issue.
AI providers determine the characteristics, limitations, filters, and risks of the models they develop, while generally excluding or limiting their own liability through contractual terms.
The provider enables the risk, but rarely assumes it.
The AI Act introduces transparency and accountability obligations, particularly for synthetic and manipulated content, but it does not fully resolve the issue of civil liability for damages caused by AI-generated outputs.
6. GDPR Profiles and Data Processing
The use of AI in advertising raises significant privacy concerns.
Even synthetic content may constitute personal data processing when it allows a message to be linked, directly or indirectly, to a specific real individual. The issue becomes even more sensitive when facial, vocal, or behavioral characteristics are involved.
The challenge does not concern only the final publication stage, but the entire lifecycle of the content: collection of input material, training or fine-tuning, output generation, internal review, dissemination, and storage.
7. Distributed Liability
The distinctive feature of the system is not merely the emergence of new risks, but the way those risks are distributed throughout the supply chain.
Every actor participates in the creation of the content, yet none fully governs the production process. As a result, liability becomes “distributed,” and the boundaries between legal positions grow increasingly uncertain.
AI does not create a vacuum of liability; rather, it creates a systemic dispersion of liability.
8. Risk Management
In this context, risk management cannot rely on formal solutions alone, but requires a structured approach.
It is essential to document the creative process, including prompts, revisions, and operational choices. AI providers should be carefully selected, contractual clauses should reflect the actual level of control exercised by each party, and internal AI policies should be implemented.
9. Conclusions
Generative artificial intelligence does not eliminate liability in the advertising sector; it redistributes it throughout an increasingly complex supply chain.
The key issue is no longer whether AI may be used in advertising, but who verified the content, who assessed the rights involved, who governed the supply chain, and who will be able tomorrow to reconstruct how that content reached the public.
Because in generative advertising, liability does not disappear behind the algorithm: it shifts, fragments, and often strikes those who failed to govern it properly.
This article was written by Managing Partner Niccolò Lasorsa Borgomaneri in collaboration with Marco Signorelli, Director of Strategy & Operations – DcP [dcpweb.it]