Gisler Report on AI in Advertising: Why Many Are Calling for a "Fairness" Label
A new study by the Gisler Protocol and the ZHAW shows that, from consumers’ perspective, AI in advertising creates standardized and stereotypical images of people. The majority would like to see more realistic portrayals and a seal of approval for them.
AI in advertising creates stereotypical images of people
Advertisements are increasingly featuring people who don’t actually exist. Generative AI creates faces, bodies, and role models in seconds—based on training data that reinforces existing beauty standards. As a new study by the Gisler Protocol in collaboration with the Institute for Marketing Management at ZHAW shows, Swiss consumers are well aware of this effect of AI in advertising: 84 percent of respondents agree that AI-generated images tend to promote standardized ideals of beauty and body image. 76 percent say that AI-powered advertising often portrays an unrealistic body image.
When it comes to stereotypes, the verdict is also clear: 86 percent believe it is possible for AI to generate stereotypical representations, and 81 percent believe it reinforces existing clichés. Eighty-seven percent of respondents agree that people are often portrayed in a clichéd manner in advertising anyway. A new technology is thus encountering an old problem here.
Impact on Self-Image and Brand Trust
According to the respondents, this is not without consequences: 77 percent agree that AI-generated images can negatively influence self-image, and 72 percent say that stereotypical images could undermine their trust in brands. Respondents estimate that stereotypes influence 62 percent of the Swiss population, but only 40 percent of themselves.
This contrasts with a clear preference: 82 percent believe that advertising should depict people as realistically as possible. For 70 percent, a brand appears more authentic when it shows everyday body types, and for 63 percent, when it shows people with imperfections. Authenticity, therefore, does not come from flawless images—yet that is precisely what technology trained for optimization delivers.
Consciously, yes; recognizably, no
94 percent of respondents are aware that advertising uses AI-generated images, but only 58 percent feel confident that they can recognize them. This gap affects trust: For 84 percent, it seems dishonest when a brand uses AI images without labeling them. However, labeling requirements, such as those stipulated by the EU AI Act, only regulate whether an image is identified as AI-generated; they do not address how people are depicted in it.
On the provider side, the technology is already part of everyday life: 68 percent of the experts surveyed use AI frequently or very frequently, 93 percent consider stereotypes to be a relevant issue, and 79 percent rate the risk of stereotypical AI representations as high. Regulatory oversight is not keeping pace: 57 percent say they do not have clear guidelines for the use of AI with imagery featuring people, and 61 percent say they do not have defined processes to prevent stereotypical representations. Only 32 percent believe their clients are closely monitoring this issue.
«Brands don’t risk their credibility by using AI, but by failing to control its use. 72 percent of consumers say that stereotypical portrayals damage their trust in a brand, and 82 percent want to see realistic depictions of people. When credibility is lost, the flawless AI face becomes a reputational risk,» says Dr. Bettina Beurer-Züllig of the Institute for Marketing Management at ZHAW.
Two-thirds want a seal of approval for fair representation
66 percent of respondents would support the introduction of an AI label in Swiss advertising, and 67 percent see the benefit in being able to better understand advertising. However, the idea that receives the most support is an expansion of this concept: 69 percent believe that a label for the fair portrayal of people should apply to advertising in general, not just to AI-generated images. Accordingly, the expectation is not for a label focused solely on technology, but rather for a standard ensuring the fair portrayal of people overall.
Nina Bieli, president of the Gisler Protocol, commented: «When a technology determines how people look in advertising, it is no longer a technical issue, but a social one. Four out of five people recognize that AI narrows the portrayal of humanity and want something different.» The «FairFrame Certificate,» a quality seal proposed in the study, could, according to its creators, meet this need.
What's Next
The Gisler Protocol is now working with the ZHAW to develop the assessment framework for the «FairFrame Certificate.» The certification will cover companies and their processes—including recertification—rather than individual advertising materials. The ongoing development process also includes evaluating potential issuers of the certificate. «We have deliberately left open the question of who will issue a »FairFrame Certificate’ in the future. The goal is to find an issuer with the highest possible profile and credibility. To that end, we are in contact with various testing and academic partners,” said Nina Bieli.
For this next phase of development, we are also seeking early adopters to test the standard in practice. Interested companies and agencies can contact the Gisler Protocol.
The study was designed, conducted, and analyzed by the Institute for Marketing Management (IMM) at the ZHAW School of Management and Law as the research partner; Gislerprotokoll was responsible for field access and recruitment. The project was funded by an innovation grant from Innosuisse. The survey included 398 consumers and 28 experts from advertising companies and agencies, supplemented by desk research and five in-depth interviews. The consumer sample is broadly diverse but not statistically representative; due to the small sample size, the results of the provider survey should be interpreted as trends.
For more information about the study and the «FairFrame Certificate,» visit www.gislerprotokoll.ch.
