Abstract
This article examines a novel form of AI-enabled creative digital communication content, termed softfake, which has emerged alongside the rapid diffusion of generative artificial intelligence across contemporary communication ecosystems. The study conceptualises softfake as a distinct communicative form and a creative artefact, co-produced through human–AI collaboration and designated to mediate presence, voice and authenticity in digital contexts. Drawing on a systematic review of media coverage, social media discourse and peer-reviewed literature, the article analyses illustrative cases of AI-generated video, audio and visual content used in strategic communication, branding and political campaigning in 2024. Building on an existing body of knowledge on deepfakes and cheapfakes, the study proposes a definition of ‘softfake’ as professionally produced, AI-generated content that intentionally conveys an authentic message from an individual or organisation created with their consent but without their direct engagement. By positioning softfakes as a new genre of creative communication, the article contributes to emerging debates on human–AI co-creation, mediated authenticity and the evolving nature of communicative authorship. The findings further identify key ethical boundaries governing the usage of softfakes, emphasising transparent labelling, organisational accountability and alignment with ethical AI principles. Beyond electoral communication, the study highlights broader societal and business implications, including applications in brand communication, organisational storytelling and influencer-led content creation. Overall, the article advances the conceptual understanding of how generative AI is reshaping creative communication practices, offering a foundation for future research on AI-enabled co-creation, communication governance and cultural meaning-making in digital environments.
Introduction
In recent years, the rapid advancement of artificial intelligence (AI) has fundamentally reshaped how communication is conceived, produced and experienced across creative industries, media systems and organisational contexts. In the global outlook, accompanied by other features of the BANI world—brittle, anxious, non-linear and incomprehensible (de Godoy & Ribas Filho, 2021), impacts communication reality. Beyond automating routine tasks, generative AI increasingly participates in creative communication practices, contributing to ideation, narrative construction and symbolic representation in domains such as advertising, branding, journalism and broadly understood digital content creation. As AI-generated and AI-augmented communication proliferates, question arises not only about efficiency and ethics but also about how communication itself is being reconfigured as a creative, co-produced (human–AI) process, especially keeping in mind that different forms of AI play a critical role in filtering and gatekeeping, with an essential involvement in informing creative communication practices (Atkinson & Barker, 2023).
The growing body of knowledge highlights AI’s expanding role in shaping creative workflows, communicative authorship and mediated presence, raising critical theoretical questions in communication research. When AI generates voices, faces and narratives perceived as authentic, the traditional distinctions underlying the concept of the communication process, such as sender, medium and message, become increasingly blurred. From this perspective, AI not merely supports communication, it begins to function as a communicative agent and creative collaborator, mediating and thereby actively shaping the meaning-making process, audience interpretation and perceptions of authenticity (Rezwana & Maher, 2022).
Building on this foundation, in addition to the already known personalisation tools, deepfakes and other forms of AI-generated content that have become ever-present (Battista, 2024; Neyazi et al., 2025), a novel form of AI-enabled communication emerged and remains underexplored in both communication theory and creative communication research. In recent years, professionally produced AI-generated video, audio and visual content has been used to convey messages attributed to identifiable individuals and organisations, created with their consent, but without their direct physical involvement. This research conceptualises this emerging practice, grounding it in political communication as a softfake (Chowdhury, 2024): a distinct communicative form and creative artefact produced through human–AI co-creation.
With the proliferation of user-friendly content-editing applications and the increasing popularity of generative AI, a broader population could create such content (Khan et al., 2023). While the preceding research already extensively examined deepfakes (Kietzmann et al., 2020; Mustak et al., 2023; Chesney & Citron, 2019; Selaković, 2021) and cheapfakes as deceptive or malicious media practices (Vries, 2020), softfakes differ fundamentally in communicative intent, authorship and ethical orientation (Jensen et al., 2025). Rather than aiming to mislead, softfakes are designed to strategically mediate presence, voice and identity within a legitimate communication context. Therefore, they represent not simply a technological novelty but a new genre of creative communication that challenges the established understanding of authenticity, authorship and representation in digital environments. Empirically, softfakes have become particularly visible in political campaigns where they have been used to extend communicative presence under conditions of constraint. However, similar dynamics are increasingly evident in branding, organisational communication and influencer-led content creation, signalling broader implications for creative industries and strategic communication practices. By examining softfakes across the contexts, this study seeks to advance conceptual understanding of AI-enabled co-creation and mediated authenticity (Vitanov et al., 2024).
While the existing AI-generated content, such as digital avatars or virtual influencers, simulate presence, they do not fully capture the specific communication goals of softfakes. Softfakes represent a distinct category as they provide a unique alignment of authorship, intent and mediation. Unlike avatars, virtual influencers or synthetic spokespersons which function as fictional entities, softfakes are created as a representation of real individuals whose authentic message is conveyed through an AI-generated form. This maintains the continuity of communicative intent while using technology as a mediator in a synthetic form that mediates the delivery of an authentic form. Therefore, the authors claim that softfakes are not simply an extension of the existing forms of synthetic media; they introduce a new paradigm in which the full authenticity of the represented individual is preserved and extended through the use of AI tools.
This article aims to define and theorise softfakes as a novel form of communication within contemporary creative communication; distinguish softfakes from other AI-generated media, such as deepfakes and cheapfakes; and identify key ethical boundaries for their responsible use in strategic, organisational and cultural communication contexts. In doing so, this study contributes to the emerging debate on human–AI co-creation, communication governance and the cultural implications of generative AI in the age of creative communication. While there are many research studies on deepfakes, there is a clear research gap related to the nature, scope and ethics of softfake content—the form of AI-mediated communication that presents authentic messages through AI generation. Unlike deepfakes, which fabricate false evidence (Win et al., 2024), softfakes pose an emerging ethical challenge in strategic communication, particularly in branding campaigns and political communication. The current research aims to determine whether softfakes are substantially different in nature from other types of fabricated video content, and what the ethical considerations are in using them in strategic communications.
The research questions include the following:
RQ1: What is the scope of softfakes, and what are the key distinctive elements that differentiate softfakes from other types of generated video content? RQ2: In the era of AI, what ethical boundaries need to be set to ensure the ethical use of softfakes in strategic communication, branding and electoral campaigns?
Conceptual Background
In the post-truth world, marked by the widespread proliferation of fake news, the imperative to use AI ethically in strategic communication became essential (Dorosh et al., 2022). This era is characterised by the challenge of discerning truth from falsehood, where information can be manipulated or content fabricated to serve various agendas and political or corporate objectives. In such an environment, the ethical use of AI in strategic communication can help combat the negative impacts of misinformation and disinformation by ensuring that information disseminated to the public is accurate and dependable (Montasari, 2024). Such an approach is critical, as AI-driven tools are increasingly used to parse vast amounts of data and deliver news and information to audiences (Kevin-Alerechi et al., 2025).
Despite numerous appeals in the scientific literature for the ethical use of AI, a proliferation of cases of unethical AI use creating deceptive or false content is evident. Notable examples include deepfakes—digitally altered hyper-realistic visual media—and those of cheapfakes include low-quality altered visual media (Hägle, 2022; Mustak et al., 2023; Nour & Gelfand, 2022). The ability of every internet user to create content and upload and share it on social networks (Kondamudi et al., 2023), the public’s desire for sensationalism (Hamby et al., 2024), accompanied by the need for immediate transmission of information (Pang et al., 2018), are enabling factors for the propagation of deepfakes and cheapfakes. The number of deepfake content posted on social networks is growing geometrically: research by the Reuters service DeepMedia shows that in 2023 about 500,000 audio and video deepfake posts were shared on social networks. To illustrate, just two years earlier, in 2021, 14,678 deepfake examples had been identified (Ulmer & Tong, 2023). Both deepfakes and cheapfakes (fake content created with cheaper software) are frequently used intentionally to deceive audiences, especially in political campaigns. Investigations in the current body of literature clearly link both categories to malicious actors and intentions. At the same time, the new phenomenon of softfakes, initially described by Chowdhury (2024), has not been examined. No other literature sources are found to describe softfakes. Moreover, no research offers distinctive criteria to distinguish softfakes from deepfakes and cheapfakes, nor are there ethical guidelines for the use of softfakes in strategic communications or election campaigns.
Rhetorical field theory, proposed by Kornprobst and Senn (2016), can offer insights into how AI can reshape communication strategies across diverse contexts. The theory suggests two types of communication—background and foreground. Background communication refers to the underlying assumptions and beliefs, and foreground communication refers to the strategic communicative acts through which these ideas are communicated to target audiences. Softfakes are not designed to fabricate false events or evidence; they operate at the intersection of background and foreground communication, conveying substantive messages (background ideas) through AI-mediated form (foreground communication) that may, however, lack transparency about their AI-generated origins. Softfakes, nevertheless, represent a novel phenomenon in which AI-generated content conveys authentic background ideas, underlying beliefs and values, but does so through strategically created AI content.
Softfakes introduce a structural separation between background and foreground communication, thus reconfiguring the concept of mediated authorship. The individual remains the author of the message at the level of underlying assumptions and beliefs, yet the foreground communication is enacted by the AI-generated avatar. This dissociation between background and foreground communication challenges conventional approaches in which authorship and delivery are typically aligned and raises new questions about communication ownership and accountability.
Moreover, the concept of softfakes also affects authenticity. While traditional communication is based on physical presence, authenticity here shifts as the message itself is dissociated from the individual delivering it. Nevertheless, a message can be considered authentic even when its form is synthetically generated, under the condition that it fairly represents the author’s intent. This reframing shifts authenticity from the integrated expression and meaning to meaning alone; it can consequently raise additional questions about the role of non-verbal communication in cases where the message is delivered in person versus by an AI-generated entity.
From this perspective, softfakes actively reconfigure the structure of the rhetorical field, transforming the relationship between background meaning and foreground representation, and how to approach and interpret agency, authorship and authenticity in the context of AI-generated communication (Selaković et al., 2024).
This distinction can be further applied to highlight the fundamental difference between deepfakes and softfakes. Deepfakes, created with the purpose of misleading, operate primarily at the foreground level visibly fabricating events that never occurred in the form of video, audio or imagery. While deepfakes create false evidence of actions or statements (Win et al., 2024), from the perspective of rhetorical field theory, these can be understood as primarily manipulating the surface of communication, thereby undermining trust in the foreground content that the audience encounters.
To further clarify the distinction between softfakes and similar forms of AI-generated content, a set of conditions that distinguish them is proposed. An AI-generated content can be qualified as softfake if the following conditions are met:
Confirmed message authenticity: The message originates from a real, identifiable individual or organisation. AI-generated content: The message is conveyed through AI-generated content not through the physical presence of the represented individual. Consent: The individual or organisation whose identity is represented has explicitly approved the creation and dissemination of the content. Non-deceptive intent: The content is created for legitimate communication purposes not with the purpose of audience deception or misinformation.
These conditions need to be met for AI-generated content to be classified as softfake: removing any one of them places the content outside the category of softfakes.
The proposed distinction can be applied across multiple contexts and disciplines. In organisational communication, health advocacy, educational settings and strategic communication, the use of AI-generated softfakes raises questions about authenticity not at the level of what is communicated, but at the level of how and through which medium the message is delivered. Where deepfakes deceive by falsifying events, softfakes complicate the relationship between message substance and artificially generated form of communication, raising ethical questions about transparency, consent and the role of human mediation in constructing meaning (Kornprobst & Senn, 2016). Moreover, the complexity of fake news opened a number of additional ethical dilemmas. While mainstream media have the power to enforce certain norms, the proliferation of new media and user-generated content is creating another dimension of asymmetrical communication. Although models of social learning, which analyse the formation of beliefs through social networks, implicitly assume that signals are shared in the same manner, the reality is different, creating a power asymmetry: more powerful and influential content creators can create more negative impact through sharing of the fake content (Abiri & Buchheim, 2022; Bahar & Hasan, 2025; Buechel et al., 2022). The power asymmetry situation can become more complex if platform owners and institutional adopters are taken into consideration: they have the power to steer the discourse through ranking, recommendation, labelling and moderation pipelines that determine visibility, credibility and reach, effectively governing the visibility of the content without clear oversight mechanisms (Gillespie, 2018). These power asymmetries are enabling micro-level manipulation of attention and narrative control. For audiences, the ethical issues in AI-mediated communication increase as generative systems blur distinctions between authentic and synthetic expressions. Genuine speech can be dismissed by biased audiences as fabricated, while fabricated speech can be presented and perceived as genuine. This type of manipulation creates audience perception biases and makes it difficult to differentiate between reality and falsity, undermining trust, accountability and shared epistemic ground (Bahar & Hasan, 2025; Chesney & Citron, 2019; Thi & Ibrahim, 2025). With strong development of AI-powered synthetic personas that have the capability to influence the entire paradigm of public relations, these issues become critically important (Luttrell & Welch, 2025).
While there are many research studies on deepfakes, there is a clear research gap related to the nature, scope and ethics of softfake content—the form of AI-mediated communication that presents authentic messages through AI generation. Unlike deepfakes, which fabricate false evidence (Win et al., 2024), softfakes pose an emerging ethical challenge in strategic communication, particularly in branding campaigns and political communication. Softfakes synthetically mediate the delivery of an authentic message that originates from the represented individual or organisation. The current research aims to determine whether softfakes are substantially different in nature from other types of fabricated video content, and what the ethical considerations are in using them in strategic communications.
Methodology
An exploratory, multi-source approach that combines literature review, media analysis and social media mapping has been employed. The objective was to identify, synthesise and interpret patterns of AI-mediated communication practices across scholarly publications, media reports and within the social media discourse. This approach is appropriate considering that the present study focuses on early-stage conceptual development, where terminology is evolving and empirical examples are distributed across different sources. While not a fully systematic review in the strict PRISMA sense, the study adopts a structured approach consistent with a qualitative synthesis approach in emerging research domains.
The present study followed a structured multi-stage process across all three data sources (media reports, social media content and academic literature). As a first step, a comprehensive set of keywords related to AI use in campaigns has been identified, and a comprehensive search across digital news repositories, social media channels and academic sources has been conducted using predefined keyword combinations (Table 2).
The research was performed in three phases: the first phase involved gathering data from established media outlets to identify relevant articles and reports published in the last year.
This process resulted in a curated dataset of cases and discussions directly relevant to the study’s research questions. The second phase focused on analysing social media platforms for public discourse related to the use of AI in campaigns. The third phase encompassed reviewing scholarly articles published in academic databases from April 2023 to December 2025. Content was further evaluated for its relevance based on the presence of keywords and the context in which AI was discussed in relation to business and election processes. Sources were screened based on relevance criteria, including explicit reference to AI-generated or AI-mediated communication; relevance to political campaigns, branding or strategic communication; publication within the specific time frame and publication language. Content was further refined through exclusion criteria, removing technical discussions unrelated to communication contexts and duplicate content. An overview of all three phases, along with sources, time period and selection criteria, is presented in Table 1.
Overview of Research Phases and Data Sources.
The objective of the first phase was to identify real-world examples and industry discourse surrounding softfakes in recent campaigns. A search was conducted across established media outlets, industry sources and organisations monitoring election integrity. A manual search of major news outlets using keywords related to campaign applications and geographic scope was conducted, limited to countries with primary 2024 elections (Pakistan, Bangladesh, Indonesia, the United States and India).
The second phase focused on social media conversations mentioning AI technologies in campaigning. It involved monitoring hashtags and trends to capture public discourse and user-generated commentaries. The third phase involved a semantic search aiming at a structured exploratory review of scientific databases and scholarly articles from April 2023 to December 2025. The search queries were conducted across three academic databases: EBSCO, ProQuest and Emerald. An overview of the Boolean search strings used for the title, abstract and keywords is presented in Table 2.
Boolean Search Strings.
Relevant content was then reviewed and analysed to identify patterns, examples and strategies for AI use in recent campaigns. The study adopts an exploratory qualitative synthesis approach intended to support conceptual development rather than a formal systematic literature review. A qualitative content analysis approach was employed to examine the identified material. Each case was analysed according to the following criteria: content of the message (authentic vs fabricated), type of representation (AI-generated vs. human-delivered), consent, intent (strategic vs deceptive) and context of use. This triangulation of data collected from different sources strengthened the findings by integrating insights from media narratives, public discourse and academic literature. This approach enabled systematic comparison across different identified cases and enabled the differentiation between softfakes, deepfakes and other AI-generated content.
Additionally, the analysis of the actual use of AI in political campaigning provided a comprehensive understanding of current practices and public sentiment surrounding this evolving topic. This approach, although extensive, has several limitations. Given the novelty of the terminology, sources may use different terms, so search strategies may not account for all synonyms. Additionally, social media platforms’ policies and content moderation may affect the availability and visibility of content, especially on sensitive topics such as elections.
Application of Softfake in Communication
There are numerous notable cases of AI-generated video content, both in business and in political campaigns. In the business context, companies are increasingly using AI-generated avatars in their campaigns (Ramadan & Ramadan, 2025). Moreover, Oliveira et al. (2025) argue that nowadays, digital human avatars used in advertising campaigns exhibit high levels of anthropomorphism, emulating human movement, speech and facial expressions. In the business context, for example, Adidas demonstrated that an entire product campaign can be based on AI-generated content (Lee et al., 2025). From the observed business attempts, the favourable and desired use of AI for creation in advertising campaigns is usually limited to digital brand mascots or human avatars. It is not associated with authentic human influencers or campaign holders. Thus, softfake, as a novel form of communication, is currently mostly visible in political communication. However, it should not be ruled out as a consideration for future widespread use in business communications.
A paradigmatic example for the creation and usage of softfakes comes from the 2024 elections in a South Asian democracy. The primary opposition movement, whose leadership was detained during the electoral period, faced the challenge of maintaining campaign visibility without its candidates’ physical presence. Political strategists resorted to an innovative solution: the jailed opposition leader would deliver written statements through legal representatives, which AI experts would then transform into video and audio messages featuring the leader’s appearance and voice, communicating reactions to election-related events. Following publication on social networks, the content would be disseminated to mainstream media to reach broader audiences (Folkman, 2024; Shahzad & Shahid, 2024).
The ethical challenge is that the video and audio messages of the detained leader of the opposition in one of the South Asian Nations, by their very nature, depict someone saying or doing something an individual did not say or do (De Ruiter, 2021), and thus fall into the category of deepfake content. Although the text is authentic and most likely original, that is, delivered by a lawyer, the video and audio contents are not entirely genuine because they were generated with the help of AI. The foreground performance is synthetically generated, the underlying communicative intent and message originate from the represented individual, distinguishing softfakes from deceptive deepfake fabrication. However, in this case, the content is non-disputable from the perspective of a political candidate; however, according to current definitions and the deepfake paradigm, it technically falls into the category of fake content. With such content given legitimacy, the need to differentiate from deepfakes currently present in the digital space arises. Reliance on the AI-powered solutions is also noticeable in the 2024 campaign in Southeast Asia. The electoral candidate relied heavily on generative AI, creating and promoting cartoonish avatars to rebrand himself and build stronger bonds and greater acceptance of his campaign among younger generations (Chowdhury, 2024; Tan & Husada, 2024). In this case, the use of AI-powered solutions to generate video content went beyond avatarising presidential candidates, including the digital resurrection of former president Suharto (Chen, 2024). Such developments are opening numerous ethical challenges similar to those previously described in the South Asian nation’s electoral campaign case.
In another election case in Asia, generative AI was used to outline the candidates’ key messages (Chowdhury, 2024). The elections have been marked overall by numerous disruptions and confusion stemming from the massive spread of disinformation through digital channels (Parkin, 2023). Based on the examples identified in the research, the need to differentiate softfakes from other types of generated video content became obvious. Based on the available literature and further elaborating on Chowdhury’s (2024) initial considerations, the following definition of softfake has been derived: softfake is a professionally produced image, a video or audio content crafted intentionally to convey an authentic message from an individual without the individual’s direct engagement in its creation. Findings of the study indicate that in the current state of global technological development, the majority of softfakes used in the electoral campaigns are AI-generated.
Consent of an individual is a primary and crucial distinction between softfakes and deepfakes. Softfake content is approved by an individual: candidate, manager, influencer or celebrity, and/or by its professional team, and is an integral part of the campaign, both business-versed and politically oriented. In both cases, softfake aims to enhance the appeal, narrative or likability of a person, brand or political candidate. It is significantly different from deepfake content, which is created with the clear intent to deceive the audience. Moreover, softfake content is professionally produced, a critical distinguishing feature between softfakes and cheapfakes. In softfakes, the synthetic element lies in the mediated performance of communication rather than in the fabrication of communicative intent. With solid convenience and the potential to reach broad audiences, it is expected that some strategists will decide to use AI to convey authentic messages, as in the case of the previously mentioned 2024 South Asian Nation’s elections, to save time and increase the volume of content that the candidate broadcasts. However, the development of softfakes does not rule out the possibility that the existing deepfakes will continue to flourish as a means of business and political competition. Competitors might begin producing hyper-realistic, completely fake audio and video contents to discredit opponents, confuse the public and damage reputations. Considering the extremely high level of pragmatism and extremely low level of ethics in today’s BANI world (de Godoy & Ribas Filho, 2021), such development does not seem unlikely.
Unwanted and unethical development would lead to complete confusion among both the mainstream media and consumers/voters, who would no longer be able to recognise authentic content. This highlights the importance of establishing ethical boundaries and guidelines aligned with the ethical AI principles outlined in the literature (Table 3 highlights the crucial differences). Given the need to identify the authenticity of AI-generated content, two key ethical boundaries can distinguish the ethical use of softfakes from other types of AI-generated video content. Softfake should be considered ethical if the published content is clearly labelled as AI-generated and created by the campaign team of an organisation promoting a product, a service or a political candidate. Such an approach aligns with good practices in ethical AI (Cheng & Jiang, 2022; Luoma-Aho & Badham, 2023). However, disclosure alone may not fully mitigate audience misinterpretation. Prior research on synthetic media suggests that audiences may continue to perceive AI-mediated communication as authentic even when informed of its artificial construction. Consequently, future communication governance frameworks may need to address not only transparency obligations but also audience literacy, contextual framing and perceptual bias in AI-mediated environment. The moral aspects of softfake creation and their use are essential, as the convenience of softfakes creates space for further proliferation not only in political agendas but also across various spheres of government and business communications.
Comparative Framework: Deepfakes, Cheapfakes and Softfakes Comparison.
Avenues for Future Research
The emergence of softfakes represents merely the visible surface of a far more profound transformation in business communication. As this study has established the conceptual foundations and ethical boundaries of consensual AI-mediated communication, it becomes imperative to acknowledge the accelerating technological and cultural shifts that will reshape the entire communication paradigm. The convergence of generative AI proliferation, autonomous agent commerce and what researchers term the ‘Dead Internet Theory’, where synthetic content increasingly dominates digital spaces, signals that softfakes are not an isolated phenomenon but rather an early indicator of systematic change. Current trends reveal that non-human traffic now exceeds 50% of all Web activity, virtual influencers achieve engagement rates triple those of human counterparts and AI agents are beginning to conduct autonomous commercial negotiations. Against this backdrop, the strategic communication landscape demands forward-looking research questions that go beyond current practice to anticipate how brands, organisations and consumers will navigate an environment in which the boundaries between human and AI-generated communication continue to dissolve. Table 4 highlights the themes and potential research questions that future researchers may consider investigating empirically.
Potential Themes and Research Questions.
These research questions collectively address three interconnected trajectories that will define the future of business communication: the automation of emotional labour and persuasion, the crisis of authenticity in increasingly synthetic digital environments and the fundamental restructuring of how brands build relationships when their primary interlocutors may be algorithms rather than humans. While a few questions explore near-term operational challenges, such as how brands communicate with AI shopping agents, or manage bot-amplified reputation crises, others probe existential tensions around trust, agency and the very nature of communication when both the sender and the receiver may be artificial. The questions deliberately span technical considerations (authentication protocols, content provenance), strategic imperatives (differentiation in algorithm-driven markets), ethical boundaries (disclosure requirements for AI-to-AI persuasion) and sociological implications (the emergence of human-verified communication as luxury positioning). Empirical investigation of these questions will require innovative methodologies that combine computational analysis, experimental design, longitudinal observation and cross-cultural comparison. Most critically, these questions recognise that the softfake phenomenon documented in this study is not a destination but a waypoint in an ongoing transformation in which the fundamental assumptions underlying strategic communication, human authorship, authentic presence and relational trust are systematically interrogated and reconstructed. Future scholarship must move beyond debating whether AI should participate in communication towards understanding how human–AI communication ecosystems can be designed to preserve meaning, maintain accountability and serve a genuine human, thereby flourishing even as the technological substrate of communication becomes increasingly synthetic.
Way Forward
Softfakes represent a broader shift in how presence, voice and authenticity are creatively constructed in AI-mediated communication. The study delineates softfakes as professionally produced AI-generated content that conveys messages directly from the influential person/organisation, with their consent, distinguishing it clearly from other types of generated video content, such as deepfakes and cheapfakes, which are intended to deceive. This differentiation and strategic use of generative AI tools in business and political campaigns are crucial for understanding the landscape of AI-enhanced communication, as demonstrated in the context of political communication through examples from recent elections in Asia and Southeast Asia. In contrast, in business, the use of softfake is still in its nascent stage.
The findings indicate that while softfakes can enhance campaign strategies, they also pose significant ethical challenges. Using AI to create content that appears authentic, yet is not directly presented by the individual portrays, blurs the lines between authentic and non-authentic content. The second research question addresses the ethical boundaries necessary for using softfakes. It is suggested that ethical use should include clear labelling of AI-generated content and creation under the direct oversight of the organisation’s team, and that any use of AI-generated content in campaigns should be openly and clearly disclosed. The validity of this study is supported by its systematic approach to data collection and analysis across multiple sources; however, it faces limitations due to reliance on publicly available data, which may not capture all instances of softfake use, especially those not disclosed by campaigns or covered by the media. For future research, it would be beneficial to explore the potential psychological impact of softfakes on consumer behaviour or voter choices, and how these perceptions might influence the success of product marketing campaigns or election outcomes.
Additionally, in the years to come, it will be crucial to explore further the regulatory frameworks that could govern the use of AI in advertising, political campaigning and the electoral process. The present study has contributed to understanding AI’s role in contemporary campaign strategies by offering a definition of ‘softfake’, clearly differentiating it from other AI-generated content types, and providing initial ethical guidelines for its use. By further exploring this area, insights may be gained into the current state of strategic communication and technology utilisation, and it may also pave the way for more responsible applications of technology in future campaigns, both in political and business contexts.
Footnotes
Declaration of Conflicting Interests
The authors declared no potential conflicts of interest with respect to the research, authorship and/or publication of this article.
Funding
The authors received no financial support for the research, authorship and/or publication of this article.