Synthetic Personas
The Invisible Influencer
Synthetic Personas, Structural Deception, and the Global Disclosure Vacuum in AI-Mediated Marketing
An independent research project examining how synthetic identities acquire commercial authority, how audiences understand the relationships they offer, and what meaningful transparency should require.
A synthetic influencer can present a face, a personality, and the appearance of a life. It can express vulnerability, cultivate familiarity, and recommend products through the language of personal experience. Yet the identity being represented may have no corresponding person beyond the content through which it appears. Synthetic Personas begins with the ethical tension inside that encounter: what does it mean to give someone your trust when the nature of that “someone” has not been disclosed?
Developed through the research paper The Invisible Influencer, this project investigates AI-generated influencers as a problem of commercial communication, audience autonomy, and platform accountability. Its central argument is that undisclosed synthetic identity cannot be understood simply as an omitted label. The paper approaches it as a structural imbalance of knowledge: brands, agencies, and platforms participate in producing and distributing a commercially constructed persona, while audiences may encounter that persona without the information needed to understand the relationship being offered. The concern is not artificiality itself, but persuasion conducted through ambiguity about who, or what, is speaking.
The construction of trust
The research brings together surveillance capitalism, parasocial relationship theory, dramaturgical analysis, critical data studies, and platform governance to examine different dimensions of the same encounter. Within its theoretical framework, the synthetic influencer is both a cultural representation and a commercial instrument: a designed identity through which familiarity, aspiration, and emotional attachment can become resources for advertising. This shifts the analysis away from whether a digital character looks convincing and toward the economic arrangements that make convincing personhood valuable.
A central conceptual distinction draws on Erving Goffman’s account of social performance. Audiences may recognize that a human influencer’s public presence is curated while still assuming that a person, with experiences and a life outside the frame, exists behind it. The paper describes the wholly synthetic persona as a “front stage without a back stage.” The distinction concerns the identity portrayed, rather than the human labor involved in producing it: there is no lived biography corresponding to the apparent person whose experiences are being presented. Disclosure therefore communicates something more fundamental than how extensively an image has been edited. It establishes what kind of presence the audience is encountering.
The study also distinguishes AI-generated personas, which represent wholly synthetic identities, from AI-enhanced personas, in which an existing person’s image, voice, or output is augmented through AI. Maintaining this distinction prevents different forms of mediation from being treated as an interchangeable category and gives the disclosure argument greater precision.
Representation is equally important to the analysis. Drawing on Safiya Umoja Noble’s work, the paper asks whose appearance, identity, and aspirations become commercially desirable within synthetic media, and who controls and profits from those representations. Disclosure does not resolve representational inequality. In the project’s argument, however, it makes a crucial difference to how representation can be examined: what appears to be an individual’s naturally occurring identity becomes recognizable as a set of design decisions for which someone can be held accountable.
Locating the research in Lebanon and MENA
Lebanon and the wider Middle East and North Africa region are positioned as primary sites of inquiry rather than supplementary examples. The project challenges an approach to synthetic-media governance in which questions are defined elsewhere and regional audiences appear only afterward. It asks how disclosure, commercial influence, and trust should be understood through the media environments and institutional conditions in which those audiences actually encounter them.
This regional focus informs both the audience research and the policy discussion. The paper compares approaches to endorsement disclosure, synthetic-media transparency, and influencer licensing, examining the different questions these instruments address. Its concern is whether acknowledging advertising or AI use necessarily gives an audience sufficient information about the identity delivering a commercial message. The regulatory comparison is a time-bounded component of the research, rather than a continuously updated statement of the law.
Research approach and evidence
The project combines theoretical analysis, comparative regulatory review, documentary analysis of published industry and regulatory positions, an exploratory audience survey, and an illustrative content-analysis exercise. Its practitioner discussion draws on the public record rather than original interviews, keeping published institutional positions distinct from independently gathered expert testimony.
The survey collected 114 valid responses through a 22-question instrument distributed across university networks, professional contacts, and social media channels in Lebanon and the wider MENA region. Within this sample, 65.8% reported awareness of AI-generated influencers, while 75.4% supported mandatory disclosure. These figures describe a non-probability convenience sample, not regional public opinion. The study explicitly acknowledges that recruitment and self-selection may overrepresent digitally engaged, AI-aware respondents and those already inclined to support transparency.
The qualitative responses give the figures their interpretive significance. Participants’ concerns were not necessarily directed against synthetic influencers as a category. Many instead expressed a desire to know when an identity was artificial before investing trust or making purchasing decisions. The paper interprets this as a position of conditional acceptance: engagement with synthetic media need not depend on concealing its construction.
The research preserves an important distinction between awareness and recognition. Knowing that AI influencers exist does not, by itself, establish whether a respondent can correctly identify one. Because the survey did not test classification accuracy, it cannot demonstrate recognition ability or establish the causal effects of disclosure on behavior. Its contribution is an exploratory account of awareness and stated attitudes toward transparency.
The account-analysis component requires a separate qualification. Structured around twelve accounts and thirty posts per account, it develops categories for examining disclosure placement, commercial content, emotional register, engagement, and audience response. The manuscript explicitly identifies the reported account-level values as an illustrative coding dataset, not independently verified measurements. Its value lies in demonstrating an analytical framework; the numerical patterns would require systematic verification before being presented as confirmed platform findings.
From disclosure to accountability
The project’s policy contribution is a proposed three-tier disclosure framework organized around commercial intent and audience risk. It distinguishes creative and entertainment personas from commercially active synthetic endorsers, then identifies higher-risk uses involving minors, sensitive sectors, or heightened emotional manipulation. The corresponding obligations range from account-level identification to prominent on-post disclosure and additional review. For commercial content, the proposal emphasizes that disclosure should remain understandable and visible as material is shared beyond its original context.
Underlying this framework is a shift in responsibility. Audiences should not bear the entire burden of detecting synthetic identity while the commercial actors deploying it possess that information from the outset. The proposal therefore assigns obligations to platforms and advertisers, supported by media-literacy infrastructure. It distinguishes legally required platform enforcement from voluntary self-regulation and explores how coordinated obligations could operate without making a universal treaty the precondition for action. These are proposed governance mechanisms, not an implemented or empirically tested enforcement system.
The broader contribution
Synthetic Personas connects the intimate experience of encountering a persuasive digital figure to the larger structures that make that encounter possible: ownership, optimization, advertising, representation, and governance. Its contribution is to bring these dimensions into the same field of analysis while distinguishing theoretical arguments, exploratory evidence, illustrative methods, and policy proposals.
The project does not call for an information environment without artificial characters. It argues for one in which audiences can understand the character of the relationships offered to them. In that sense, disclosure becomes more than a compliance requirement. It is part of the conditions under which trust can be meaningfully given, withheld, or revised.
The central question is not whether a synthetic persona can appear human. It is whether the person encountering it has been given the information necessary to choose how to relate to it.