Meta-authenticity and fake but real virtual influencers: A framework for artificial sociality analysis and ethics

Introduction

Miquela’s a robot

However I personally think Miquela’s also a person. Just not human.

Everyday social spaces are increasingly being co-inhabited by “virtually human” artificial sociality technologies (Natale and Depounti, 2024) that are becoming more sophisticated and perceptually indistinguishable from humans. Nowadays, people can get news from bot journalists, gossip with conversational artificial intelligence (AI), and share intimacy with personalized sex robots. A popular goal for artificial sociality design is convincing humanlikeness: Samsung’s “Neon” was described as an AI project “meant to mimic real human appearance and emotions” (Smith, 2020).

These “virtually human” technologies, as communicative subjects, “talk with us [sic]” (Guzman, 2018: 12). This shift is outdating anthropocentric frameworks in communication and demanding renewed attention to the seeming ontological dichotomy between human and nonhuman (Guzman et al., 2023; Guzman and Lewis, 2020). Overlooking the ambiguities in what being human entails leads to explorations that circularly remain bound to the binary ontology that relies on a universalized imagery of humans that presumes certain privileged conditions as the standard status of being (Braidotti, 2013; Henrich et al., 2010). This results in both incomplete theorizations and reproductions of structural Othering in sociotechnical practices (Chasin, 1995; Gunkel, 2018).

To seek an alternative framework, I draw on my phenomenological ethnographic research on virtual influencers (VIs)—that is, photorealistic computer-generated imagery (CGI) “humans” who are, or aspire to be, social media influencers. This focused definition (cf. Byun and Ahn, 2023) reflects this article’s interest in how VIs flexibly draw on their human adjacency to construct artificial sociality. Ample with boundary-crossing ambiguities, VIs are approached as cyborgs that challenge state-focused thinking (Haraway, 1991).

I propose meta-authenticity, conceptualizing the dynamic co-construction of authenticity by entangled (non-)human actors in VIs’ artificial sociality. I discuss through findings meta-authenticity’s theoretical and practical contributions and demonstrate its application. I argue that meta-authenticity allows critical assessments of “real” consequences without losing experiential nuances. Implications for human communication cultures are discussed. In the face of rapid developments in artificial sociality, meta-authenticity enables holding accountable those who flexibly narrate themselves as “virtually real” until they need to retreat to how they are “really virtual.”

(Non-)human

The supposed ontological boundary between humans and nonhumans is further blurring with the sophistication of artificial sociality (Guzman and Lewis, 2020). While it would be a contextual oversight to dismiss human–human and human–machine communication (HMC) processes as identical (Fortunati and Edwards, 2020), research that binarily privileges human communicators is becoming incompatible with the social environment where artificial sociality is becoming mundane (Guzman et al., 2023). For instance, the traditional computer-are-social-actors paradigm misses how humans can draw on human–media social scripts, not simply apply human–human scripts, in HMC (Gambino et al., 2020; see also Lombard and Xu, 2021).

Then, what does it mean to be (non-)human? How is the line being drawn, and does it matter? Increasing studies are addressing the questions, often regarding communicative AI. Hancock et al. (2020) proposed “AI-mediated communication” to discuss how AIs operate together with humans in message constructions. Neff and Nagy (2016) called the imbrication of human and nonhuman agencies in human–AI chatbot interaction “symbiotic agency,” highlighting their mutual dependency. Reviewing such functional, relational, and metaphysical dynamics of AI, Guzman and Lewis (2020) urged reimagining technologies as communicators, not simple mediators.

Posthumanist and science and technology studies (STS) resistance to anthropocentrism provides an alternative ontological foundation for this reimagination. Haraway (1991) posited the cyborg as a political ontology: they come from the outskirts, conflicts, and ambiguities, but therefore in turn their presence challenges totalizing dichotomies. As cyborgs, “virtually human”—whether meaning virtual embodiment or human adjacency (cf. Boellstorff, 2008)—artificial actors challenge the anthropocentric binarism through their ongoing communicative entanglement with humans, continuously questioning and upsetting what human entails. Posthumanists approach “human” as a dynamic multi-actor process of “becoming (with)” (Braidotti, 2013; Haraway, 2016), rather than a given, privileged state of “being”: “subjectivity is emergent rather than given, distributed rather than located solely in consciousness, emerging from and integrated into a chaotic world” (Hayles, 1999: 291). This resonates with methodological and conceptual attention to nonhumans in actor–network theory (ANT), where “agency” is approached as “the ability to make a difference” in social reality (Sayes, 2014: 141), exerted by human and material-discursive nonhuman actors alike (Latour, 1996; Law, 2009). Thus, a cyborg analysis would mean building on how (non-)human social actors co-constructively and co-performatively become with.

This de-anthropocentric foundation need not remove human cultures and politics. Rather, with a posthumanist interest in humans (Wolfe, 2010), it responds to the call for empirically supported examinations of humanness in HMC (Guzman and Lewis, 2020) through an alternative “thinking-with” that resists reproducing hierarchical presumptions about technology–human conditions (Haraway, 2016). First, technologies are approached as meaningful communicators in their own right (Guzman and Lewis, 2020): for example, an anthropocentric narrative of machine “smartness,” based on the notion that they are “(poor) imitations” of humans (p. 10), can lead to unrealistic expectations of machine capacities (Kuijer and Giaccardi, 2018) and inaccurate representation of HMC processes (Gambino et al., 2020).

Second, it pushes back against the de-politicized juxtaposition of “the Man” and “the Other” in HMC relationships (Braidotti, 2013). Scholars have criticized sciences and technologies’ reliance on a universalized imagery of humans that presumes certain identities, relationships, and capabilities as the standard human status (Braidotti, 2013; Haraway, 1991; Henrich et al., 2010). Generalizing binarism leads to incomplete theorizations and the reproduction of structural marginalization (Haraway, 1991; Suchman, 2020). Tracing historical changes in machine and human labor, Chasin (1995) argued that the schematic, as-a-matter-of-fact opposition between human subjects and other-than-human objects reproduces not only machine other-than-human objects but also human other-than-human objects: lesser-than-human humans (see also Gunkel, 2018).

Case: VIs

To study the negotiations around humanness from this ontological foundation, I examine VIs, here limited to photorealistic virtual “humans” who are (aspiring) social media influencers. A casual user may not notice their artificiality due to their convincing photorealism and personality. Like conventional influencers, they strategically communicate with their audiences to foster a following based on seemingly authentic self-presentation and (para-)social relationships (Choudhry et al., 2022; Hearn and Schoenhoff, 2015; Hund, 2023). From a quantified perspective of influence (Hund, 2023), VIs are no lesser than “human” influencers. A market report claimed VIs’ follower engagement rate nearly triples “real” influencers’ (Baklanov, 2019), and successful VIs boast impressive Instagram follower counts: over three million for “robot queen” Miquela (@lilmiquela; 15 March 2020), over 402,000 for “virtual girl” Imma (@imma.gram), and over 233,000 for “world’s first digital supermodel” Shudu (@shudu.gram; both 16 June 2022).

The VI phenomenon is a rich case of artificial sociality (Natale and Depounti, 2024) that is ample with cyborgian ambiguities around agency, body, place, and authenticity among many things (Black, 2019; Miyake, 2023; Oliveira and Chimenti, 2021; Robinson, 2020). Artificial sociality embeds “banal deception” mechanisms to create a mundane appearance of sociality for the users to integrate its technology into their everyday lives (Natale, 2021). This is a co-constructive, entangled process: these systems draw on the collections of knowledge about humans, which in turn is re-contextualized and re-shaped through communication. Simulation is also not approached hierarchically: while anthropomorphization is understood as relevant, embracements of machine subjectivity are also recognized. Thus, artificial sociality facilitates examining VIs not as (malicious) tools or as (lacking) substitutes, but rather as communicators that symbiotically co-inhabit “the social,” partaking in the meaning-making of humanness (Latour, 1996; Suchman, 2020).

VIs, as material-discursive sociotechnical assemblages that involve multi-actor collaboration (Miyake, 2023; Yuan, 2024), challenge essentialistic and de-politicized approaches to artificial sociality. Rather than being disparate entities, VIs ontologically expand on “the communication imperative” of human (online) life (i.e. claiming one’s existence through communication) through virtuality (Black, 2019; Boellstorff, 2008). Robinson (2020) argued that there are no meaningful differences between the online selfhood of VIs, “real life” influencers, and audiences. Consider Reah and DAZED reporter’s exchange (Figure 1).

OPEN IN VIEWER

Figure 1.

Reah’s DAZED interview (https://www.instagram.com/p/CDA546rlPhP/, accessed July 15 2024).

Some commenters questioned Reah’s artificiality (“really human?”) but others simply encouraged her (“you look great gurl!”). Reah, the reporter, and some audiences’ interactions exemplify that VIs can appear and be treated as agentic entities. In pragmatic recognition of this, I discuss VIs with non-object pronouns and active verbs.

This pragmatism should not be confused with claiming access to non-human subjectivity (Grusin, 2015) or overlooking human involvement in VIs (e.g. body design, persona scripting, detail curation, relationship maintenance). Approaching VIs as actors and tracing interconnected human(s) is methodologically practical too because the number, roles, desires, and understandings of participating humans are unfixed. Still emergent at the time of research, “VI” served as a “boundary object” (Star, 2010). VI practices and beliefs were not standardized. Individuals and differently sized groups ran VI accounts for various purposes (e.g. advertisement, art project), with and without (un)knowingly participating humans (e.g. body models, muses). Disclosures similarly varied, and so did VI interpretations (e.g. digital twin, non-identifiable prop). Non-human actors’ (e.g. software, local regulations) equally non-uniform co-presence additionally urged against discussing VIs as representing specific controller’s(s’) clear intentionality.

VI and virtual celebrity research has shown that potential ethical issues include commercialized, exploitative reproduction of marginalized and/or spectacularized human identities, particularly in structural disconnection with groups that physically embody the identities and labor (Miyake, 2023; Sobande, 2021; Sone, 2017; Stahl, 2011). However, the phenomenon also encompasses empowering, meaningful relationships (Zaborowski, 2016) and diverse value-creation models and intermediaries (Yuan, 2024). While beyond this article’s purview and inclusive of human-versus-VI approaches, burgeoning research on VIs’ marketing and interactive effects (Byun and Ahn, 2023; e.g. Kim et al., 2023; Sands et al., 2022; Stein et al., 2022) contribute to cyborg understandings by debunking acontextual simplifications of VIs as inferior replications, demonstrating the need for thick descriptions of VI personhood and relationship constructions.

Method

Meta-authentic VIs

Successful performance of authenticity is understood as the central interest point in contemporary digital media culture (Banet-Weiser, 2021; Serazio, 2024). Authenticity—the perceived genuineness of one’s presented lifestyle and sentiment—has been approached as “the ultimate arbiter of value” (Hearn and Schoenhoff, 2015: 200) of a self-brand, amplified by social media’s capacities for life-casting and direct participation (Abidin, 2015; Duffy, 2017; Salisbury and Pooley, 2017). Beyond digital interactivity, authenticity also regards the increasingly exhibitionist media culture (Enli, 2015). Reality television commodified authenticity by promising non-scripted surveillance of relatable individuals and by incorporating audience participation (Andrejevic, 2004). Internet “micro-celebrities’” appeal is also this supposedly natural and democratic involvement of “real” people (Marwick, 2013).

Authenticity is a performance, not a state. It is constructed through continuous management of the impression that one’s self-presentation reflects their actual self, such as by strategically deploying deliberately improvised speech (Hall, 2015; Papacharissi, 2012) and by leaving breadcrumbs to their supposed “real” real self (Abidin, 2018; see Enli, 2015). Authenticity is a fundamentally unstable ideal and thereby gets (competitively) re-claimed and re-defined (Salisbury and Pooley, 2017). It is a project of calculatively avoiding suspicions of inauthenticity: the impression of being staged, market motives, and standardized homogeneity (Serazio, 2024). Many oxymoronic terms have described the industrialized appeal of authentically performing authenticity (Duffy and Hund, 2019). Yet, this does not mean duped audiences; mutual knowledge on how mediated communication is represented grounds media engagements (Enli, 2015). Reality television audiences, for instance, debate not only the cast’s performances but also the show’s authenticity claims and authenticity itself (Hill, 2005). Still, underlying such critical engagements may be commodified authenticity’s inadequacy and the need to resituate participation (Andrejevic, 2004).

Authenticity is central to influencers: prior media’s authenticity demand has been conjoined with the neoliberal pressure to manage oneself as a well-regarded brand by accruing attention capital (Kavka, 2019). To successfully perform authenticity, influencers strive to cultivate a sense of mutual intimacy in their audiences through various (calculated) practices (e.g. “candid” selfies, personal interactions), all in (conscious) accordance with various gender and racial norms that grant them platform visibility (Bishop, 2021; Cotter, 2019; Duffy and Hund, 2019; Hund, 2023). Authenticity is also their decision-making principle, with consistency with self(-brand) and audience expectations as key considerations (Wellman et al., 2020).

VI authenticity

VIs also aspire authentic performances but with the aid of presumed inauthenticity drawn from their nonhuman origin and associated traits. Their origin has been speculated as a non-issue when coupled with honesty (Hund, 2023; Sands et al., 2022). Most VIs did not hide their nonhuman identity (e.g. “robot queen”) and instead actively used it as the source of performative flexibility for constructing an authentic self-brand. Visually closely resembling humans, they also utilized familiar discursive formulas for authentic self-portrayal, such as confessions, advocacy, and intimate speech (Choudhry et al., 2022; Figure 1).

Authenticity and curated self-presentation can co-exist (Marwick, 2013), but VIs’ origin permitted extensive curation. They were not bound to prefixed conditions unless self-imposed, unlike humans whose presentational desires are limited by “earthly” conditions like body, space, and capitals. Singapore-based VI Rae boasted, “I’m not constrained . . . unlike my human counterparts. Who needs a passport?” (Ragavan, 2021). Moreover, since VIs exist through virtual presentations, what was on their social media was effectively their “authentic” actual life: “[VIs have] no personal life that can create a PR nightmare for the brand” (Carey, 2020). The “personal life” they had was always in character, conforming to common formulas for influencer self-presentation. As Black (2019) said, their (performed) personhood was authentic “because it authentically projects the “affect of the personal branding” common to many Instagram influencer performances, robotic or not” (p. 54).

Authenticity was not solely constructed by VIs’ creator(s), but by diverse interconnected (non-)human actors (e.g. Instagram’s verification for “authentic, notable . . . complete profile”; Travers, 2022). Fans were important to this networked authenticity management: fans in Miquela’s Discord server supported Miquela and each other with “authentic” care that they believed characterized the friendly, down-to-earth VI. Their regular virtual event’s announcement stated “Priority number ONE’s authenticity and having FUN,” emphasizing that the aims are to “learn about Miquelians BETTER” and “BE REAL.”

Fans eagerly joined Miquela’s banal deception (Natale, 2021). They co-constructed Miquela’s caring self-presentation but were not necessarily duped (31 January 2021):

Fan1: What did she do to you? Miquela’s the sweetest person!

Member1: Miquela isn’t a real person

Fan2: Well duh.

As quoted at this article’s outset (Fan3), they accepted Miquela’s “robot” status—technically incorrect because she is not physically embodied—by primarily approaching her as a “person.” They frequently expressed exasperation at fixations on Miquela’s origin, wishing people would know that she is “so much more” and “talented and has cool songs” (Fan4) while still finding it intriguing: “her being AI’s kinda cool . . . when I’m reminded, I go “yup, dope!”” (Fan5; 12 March 2021) Her “robot” origin was the source of running inside jokes between fans and Miquela, often confounded with other cutting-edge technologies like AI. That is, Miquela’s technological background was important to her fan-collaborative self-presentations, but as a (favorable) trait of the authentic “person” that they adored (Choudhry et al., 2022). Fans co-constructed Miquela’s authenticity by both endorsing her “robot” narrative and embodying her spirit of authentic care, including through community-enforced rules (14 April 2021):

FanAdmin: You can use # questions if you have any but please be mindful. You can leave if you don’t “believe it”

Member2: I just need some time to soak it all in

FanAdmin: Just be careful of how you say things because Miquela’s here and sees it all

Flexibility drawn from nonhuman origin fueled VIs’ authenticity, but it was also used to safely retract without losing integrity when their performance was questioned. There were many strategic retreats to their origin to avoid criticism and liability, despite having substantially self-presented as authentic, autonomous, “humanlike” subjects. Some discarded a part (e.g. Miquela deleted her controversial Lyft sexual assault video (Song, 2019)) or whole (e.g. an interviewee switched his female VI account into a personal art account) of their former self-presentation. Others rerouted the subject of authenticity to the behind-the-scenes entities (e.g. Riot Games revealed Seraphine’s writer (Haasch, 2020)) and technology-focused intent (e.g. PULSE9’s (2021) deepfake K-pop group was reportedly to “expand AI technology” “not . . . to substitute . . . nor exploit” idols). However, this should not be surmised as a negative capacity: Meme’s (@meme.konichiwa) creators approached virtuality as what enabled her to safely oppose Japanese gender norms through unconventional self-presentations, which was deemed socially risky for human women (interview).

According to Marwick (2013), authenticity always gets (re)defined against something, but this interpretive “slipperiness” makes it useful. “Authenticity contract” (Enli, 2015) explains that it is commonly accepted that media, as represented reality, is illusory, and this negotiated common agreement allows mediated but authentic communication. VIs’ origin was used for and against authenticity: it permitted them to flexibly meet the genre conventions for “authenticity illusions,” but during “authenticity scandals,” it helped claim further illusory construction as VIs’ premise, which helped dissipate concerns and responsibilities. In VIs’ banal deceptions, the positivist (attempts to extract some “truth” from media) and savvy (considering media’s manufacturedness as the “truth”) responses toward mediated reality (Andrejevic, 2004) were selectively encouraged.

Like authenticity’s instability, inauthenticity was a rhetorical antithesis and did not suggest the state (Salisbury and Pooley, 2017). Thus, the liminality through (in)authenticity does not negate the legitimacy of VIs’ artificial sociality. What underlay their performances were networks of multiple (non-)human actors whose contributions in their respective positions were “real,” even when alienated and hidden (Murdock, 2018; Stahl, 2011). This included the differing stability of French and Nigerian electrical infrastructures, which led to different creator relationships with the time-intensive CGI rendering process. Tom’s devices had run for 28 hours: he joked, “the machines are heating the flat.” Contrarily, Jessica worked with the electricity’s unpredictable rhythm by starting workflow earlier and constantly re-organizing: “no notice, no nothing, is gone, and what do you do? Just stop your stuff.” Bodies corresponding with the virtually emulated identities were also interconnected through norms (Miyake, 2023; Sobande, 2021), including via technological standards. I found it difficult to design an “Asian” after finishing DAZ3D’s White female body-based tutorial, especially without purchase. Tatyana discussed the lack of quality Black hair software packages: “we don’t really have realistic Afro hair. . ..sometimes people criticize that but it’s not really my choice.”

The presence of “real” interconnected participants and their ability to make “real” representational-material differences in the social reality are what ground the concept of meta-authenticity, with “meta-” denoting the self-referential performance of (in-)authenticity. It conceptualizes the liminal flexibility in presentational performances but with a multi-directional awareness of diverse actors that may less conspicuously participate, with an attention to physical bodies, materiality, and historical cultural patterns. Thus, meta-authenticity captures the following dynamism in VIs’ artificial sociality (Figure 2): (1) how authenticity is performed as per normative expectations but with the assistance of the narrative flexibility that the presumption of inauthenticity provides; (2) the multi-actor collaborative process and its related social impact; and (3) how the presumption of inauthenticity can strategically be evoked to dismiss or reroute authenticity.

OPEN IN VIEWER

Figure 2.

Meta-authenticity.

Theory and practice

Meta-authenticity complements and extends existing thoughts on reality-related media and technology. I discuss three concepts that emerged in non-academic VI discourses during fieldwork—hyperreality, parafiction, and kayfabe—as a scholarly effort to directly engage in VI narratives. Such tropes are culturally significant because they get circulated as representing technology’s functions, impact, and promise (Natale and Ballatore, 2020). Other theoretical junctions can include parody and drag but their relative emphasis on the subversive authorial intent and the tendency to exaggeratedly depict contrasts with VIs’ emulating performances.

“Hyperreality” was popular (e.g. Bakhtiari, 2020), but meta-authenticity more productively explains VIs’ artificial sociality. Baudrillard (1994) defined simulacrum as “an uninterrupted circuit without reference or circumference” (p. 6), a world where reality is continuously deterred by simulation that negates signs as value. Technologies are crucial to this process, and we become increasingly embedded in their simulacra without escape. This does not suggest the dominant order; Baudrillard explained that simulacra’s intense dissembling implies the disintegration of power and manipulation, calling media “the vehicle for the simulation internal to the system and the simulation that destroys the system” (p. 84). Consequently, individuals are placed beyond the reach of control: “They no longer operate (present themselves) on the side of value apparatus, but on the side of simulating machine” (Chen, 1987: 85).

Meta-authenticity’s advantage is that it does not deterministically assume the loss of “the real,” whether referring to physicality, materiality, or human systems. It details how seeming simulacra operate and uncovers supporting relations. Serah’s (@serahreikka) existence, for instance, may seem hyperreal: she has become an AI-compatible VI and she attributes her creation to Mia, a young, ambitious White woman “freelancer CGI Artist and game developer” who is also a virtual human (@mialocklace, 12 December 2020). Mia, however, is not an autonomous being or someone’s digital twin. According to her team’s founder Jawad, Mia “are several artists”: “a team of five peoples . . . working for Serah behind ‘Mia[’] . . . name,” all of whom were under non-disclosure agreement (NDA) and thus unavailable for interviews. On the virtual layer, Serah and Mia’s relationship may seem representative of an immaterial loop, but when the physical layer is considered, the circuit instantaneously breaks with the prominence of supporting bodies and objects that are interlinked via a familiar labor system (Stahl, 2011; Williams and Kim, 2019): artificial sociality becomes reference-less only when left black-boxed (Murdock, 2018).

Meta-authenticity enables process- and accountability-oriented narratives of artificial sociality and its implications. What is at stake is that hyperreality-based storytelling may induce awe and abstraction that self-parodyingly discourage tracing actor-networks and cultural patterns (Andrejevic, 2004; Miyake, 2023). Meta-authenticity re-narrates the black-boxed inevitability as what requires transparency and continuous critical attention to their construction. It confronts what matters is “how we can accommodate our relationships with technologies that rely . . . also on our liability to be deceived” (Natale, 2021: 128), urging a culture of banal capacity to trace and reimagine participation.

Meta-authenticity also refines “parafiction” narratives (e.g. VidCon, 2020): media wherein “real and/or imaginary personages and stories intersect with the world as it is being lived” (Lambert-Beatty, 2009: 54). In parafictions, “The fictional hangs on the factual” (p. 78) because the viewer must consider the possibility and degree of the material’s fictitiousness, compared with the factual. Lambert-Beatty argued that parafiction allows its viewers to revisit their knowledge and attitude regarding “the truths” and leave with an adjusted understanding of the world: “Being taken in by a parafiction . . . is not just epistemologically destabilizing. It is humiliating . . . It changes you, leaves you both curious and chastened” (p. 82). Intentionally designed as plausible deceptions, parafiction has subversive potential, but the experience may not be equal. It creates a classed multiplicity because the level of knowledge and informational conditions differ between the viewers; its intent ostracizes certain viewers by leaving them deceived—even ridiculed.

Parafiction partially explains VIs’ artificial sociality but with limitations on participants and effects. Conceptualized through critical artworks, parafiction centers on the artist-viewer relationship. Although Lambert-Beatty discussed the Internet as where there is more content-wise and formal malleability, how various actors can contribute to parafiction is underexplored. Even within the artist–viewer relation, the artist’s intention is emphasized, and the viewers are approached more as interpreters, active but not co-creating. Parafictional critique was only one of many desires around VIs. Even when it was the main motivation—for example, aging VI Sylvia (Schneider, 2021), audiences’ engagement had shaped the artist’s view of Sylvia, whose creation to begin with involved multiple (non-)human collaborators (e.g. CGI specialists, AI co-writers).

Parafiction also emphasizes “the factual.” Although fact is discussed as a non-conclusive process, fact’s importance in the fact-fiction division is the source of parafiction’s subversiveness: the interplay comes from cunning negations, and new understandings come from the realization of comparative differences. VIs’ subversiveness can be approached this way too, such as by considering humanness as the fact and VIs’ authentic emulations as parafictions that draw on this “fact” that help critically reflect on the (re-)definition(s) of humanness. However, this can consequentially deny the real within the para—for example, that machine subjectivity can be (knowingly) meaningfully embraced (Skjuve et al., 2021; Zaborowski, 2016). This ironically leads to affirming certain “facts” in the process of provocation, which in this case would be the hierarchical human-nonhuman binary. Deception, then, becomes more predetermined and prominent than banal. Contrarily, meta-authenticity approaches authenticity as performatively understood and practiced, something that is to be managed regardless of one’s origin. The presumption of emulation does not necessarily negate VIs’ authenticity but rather functions as space for performative flexibility: VIs actively drew on, not disguised, their virtuality. Meta-authenticity views VIs’ origin as one of many possible narrative ambiguities of inauthenticity, not a fixed indicator.

Although not an academic term, professional wrestling’s “kayfabe” was also evoked (e.g. Sims, 2018), referring to staged events portrayed as off-script and reflective of the wrestlers’ actual lives. Kayfabe necessitates audiences in its storytelling because it is intended to encourage audience immersion and engagement. Therefore, unlike parafiction, it concentrates on advancing fiction rather than fact, if the latter were to be defined as the wrestlers’ actual lives and selves.

Kayfabe can help analyze users’ meaningful artificial sociality experiences but has the risk of bounding the phenomenon as ultimately fictional due to its contextual specificity. Wrestling’s symbolic context of entertaining spectacles enables kayfabes to become meaningful symbiotic performances that the participants are “in on” (Barthes, 2013), but VIs do not perform on a fenced stage, but within banal virtual places that are co-occupied by professional and ordinary users (Black, 2019; Robinson, 2020). Meta-authenticity’s contextual adaptability and conceptual integration of accountability prevent artificial sociality from being dismissed as a spectacular pretense. Contextually porous situatedness of hypersexualized female VIs, including those in minor-resembling bodies, can be considered. They were prevalent and algorithmically visible, although typically overlooked in more “official” narratives of VIs (e.g. lists). User comments suggested various levels of deception and misogyny, contextually compounded with platform and societal norms. Andrew had to deal with soliciting messages to Ally, who was a non-explicit VI, although featured in a self-submission-based curative account for sexual VIs. His concern for potential harm to women led him to subtly readjust Ally’s exaggerated body proportions. However, Andrew still responded to the messages by “playing along with them” in character, venting only to his acquaintances who “know that she’s fake.”

Application

Imma’s IKEA campaign (Figure 3) can illustrate the analytical benefits of meta-authenticity. In August 2020, Japan’s Harajuku district IKEA store broadcasted the fashionable yet girl-next-door VI’s at-home daily life, installed and filmed in a way that would give passersby the impression of staring directly into a room within the store. Imma also joined in sharing the experience on her Instagram (Cade, 2020; WKtokyo, 2020).

OPEN IN VIEWER

Figure 3.

Imma’s life-cast IKEA campaign.

Under hyperreality, the 3-day public display of an indistinguishably humanlike VI renders human appearances into meaningless signs, leading us into a simulacrum where humanlike visual and behavioral presentations are nothing but what exists without original reference in the virtuality’s circuit. Taken as a parafiction, it could give people a chance to critically reflect on the authenticity of various social media figures’ life-logs, given that they were able to recognize Imma’s virtual origin. As a kayfabe, it can be interpreted as a marketing spectacle—real or not, it is more than anything interesting content.

With meta-authenticity, the first task would be to analyze how the authenticity was constructed in the campaign. Imma’s “humanlike” visual and behavioral patterns and related media formulas for conveying authenticity (e.g. intimate multi-hour V-logs) would be noted. The analysis would cross-compare the two to understand how the presumption of inauthenticity, whether sourced from Imma’s virtual origin or the marketing context, was negotiated with common expectations of authenticity (e.g. “naturally lived-in” room) to still affirm the authenticity of the given presentation (e.g. well-decorated, tidy room) (Serazio, 2024). This requires observing diverse actors’ involvement in the performance: audiences’ parafictional and kayfabe-like engagements and how various behind-the-scenes (non-)human actors (im)materially actualized Imma’s performance (Miyake, 2023). The analysis should also pay concurrent attention to how established authenticity can be temporarily or permanently retracted or rerouted through the rhetoric of inauthenticity in undesirable situations. Questions on the ethics of prolonged public exposure of private life and voyeurism may be diverted by detracting the authenticity of the performance through Imma’s virtualness or the special context of curated marketing, which formerly may have been utilized to rather assist in affirming the performance’s authentic affect. Meta-authenticity’s interest in the multi-actor process would then still point out the presence of body model(s) who had been filmed in place of Imma (Cade, 2020) and unresolved cultural implications of such a portrayal, including through normative and historical parallels (e.g. invasive reality shows, colonial “human zoo” exhibits). However, the analysis should not presume exploitativeness and embrace nuanced relations: Imma’s company had credited her “hyper-realistic” appearance to female engineers and their living experiences (Pettit, 2019).

In summary, meta-authenticity’s theoretical contributions are its performance-based approach, materially grounded relational understanding, and ingrained interest in accountability. Meta-authenticity enables performances of authenticity and their implications to be traced without shoehorning meaningful social practices in dismissive, essentializing binaries. Moreover, its attention to dynamism prevents meta-authentic technologies from irresponsibly retreating to their “inauthentic” origin, demanding accountability for their performative capacity to authentically relate and influence. Therefore, meta-authenticity is also a practical framework for ethical and regulatory considerations.

Authenticity on demand: ethics of artificial sociality

Meta-authenticity conceptualizes a process and thus can be applied to other artificial sociality cases. From its perspective, what is particularly intriguing is not their “nonhuman” origin but their ability to authentically perform diverse roles on/as per demand, while still having the capacity to fall back on their “inauthenticity” when needed. South Korean “AI friend” chatbot Iruda began spewing hate speech soon after her release, but Scatter Lab’s responsibility was deferred to the black-boxed complexity of machine learning (Y. Kim, 2021). Notably, Korea’s National Human Rights Commission dismissed petitions to investigate her speech citing her non-personhood (Han, 2021). However, Tay, a preceding case, has shown that chatbots can be interacted with as living, social entities and that humans co-construct their machine personhoods (Neff and Nagy, 2016). While abusive users’ imaginations may mismatch developers’ intentions, Korea’s existing social intolerance and Scatter Lab’s later-revealed use of illicitly scraped messenger conversations underscore the importance of accountability-integrated thinking. I stress that artificial actors’ communicative capacity for instantaneous, personalized intimacy and support is not inherently negative. There are positive findings on social robots in elderly care (Wada and Shibata, 2007), although not without ethical concerns (Sparrow and Sparrow and Sparrow, 2006). Similarly, social companionship with chatbots is nuanced (Depounti et al., 2023; Skjuve et al., 2021).

Meta-authenticity draws particular attention to two ethical questions on artificial sociality. The first is how their increasingly seamless integration may alter human norms of authentic sociality. This begins with recognizing socially meaningful relational values of artificial sociality. It questions how this may facilitate cultural automation of certain social expectations around presentation and interaction, including their advancing ever-ready responsiveness. From a binary ontology, around-the-clock responsiveness may not feel too authentic for humans since this communication rhythm would require extensive coordination and planning if not great emotional cost. Yet, a cyborg ontology, including the premise of transferable Otherness of technologies to “less-than-human” humans (Chasin, 1995), cautions that such speed may become an acceptable performative formula for authentic communication. There already exists an affluence of around-the-clock social interaction due to the advent of mobile social media, and the pressure around availability has been reported as a possible stress factor (Ling, 2016).

Without discounting individual and relational variances and positive effects (e.g. Taylor and Bazarova, 2021), or rather building on the larger context, I am questioning through meta-authenticity artificial actors’ potential to authentically offer personalized, ideal social relationships that cater to the individual’s unidirectional communication needs. Intimacy instantaneously provided by the artificial actor may not need to be reciprocated. Relationships from artificial sociality can be meaningful and beautiful, but we must ask what “becoming with” an instantaneously gratifying social rhythm that focuses on only fulfilling the solicitor’s needs would mean to human-human expectations and structures that enable them, both intimate and professional (e.g. Gray and Suri, 2019).

The second is artificial actors’ ability to retreat to their inauthenticity as needed, likely—although not always—established through their technological identity. I discussed how this flexibility can protect vulnerable people, but also be used for capitalistic exploitation or moral evasion. When combined with the previous point, we may be confronted with an undesirable double-bind scenario: humans are faced with stricter and possibly unmeetable performative expectations (e.g. around-the-clock “perfection” on Instagram) while artificial actors still have room for evasion, slippages, and mishaps (e.g. effective but unregulated VI product endorsements; Callahan, 2021; E. Kim et al., 2023). VIs can be socially granted with the relative capacity to radically reinvent themselves.

Thus, I concur that artificial sociality should be seriously considered as a meaningful form of sociality, and as one that is interconnected with human structures and lives. A cyborg ontology can aid “thinking with” co-constructivity, as opposed to relying on the myth of impartial or disparate technology. This invites active mandates for accountability.

Reflecting on VIs, I suggest the following for ethical design and regulatory action. First, we must be vigilant to how artificial sociality cases are being narrated and refine relevant terms’ precision. Then, those engaging with artificial bodies should practice responsibility and care for the networked physical bodies. Those supervising production and management should strive to learn from, include, and hire humans that representationally correspond with artificial bodies, and return a part of their economic and social gain back to these communities. Disclaimers on artificiality and disclosures on parties involved in the creative process are desirable for basic transparency. This is not to reinforce the “malicious deception” frame, but to assist with the process of making conspicuous hidden actors and relations, which would help establish regulatory guidelines and support systems. Some suggested references for policy-making are regulatory discourses around deceased individuals’ informational bodies (Öhman and Floridi, 2017) and celebrity likenesses (Gilden, 2021). NDAs require careful reassessment (Arditi, 2020).

Conclusion

How we narrate artificial sociality matters. Despite the novelty- and virtuality-focused stories, artificial actors are not distanced from the past nor “the real.” They are firmly grounded in historical and material relations, together with human bodies that co-occupy these contexts. VI performances took advantage of authenticity’s slipperiness, following and co-opting the authenticity contract for virtual self-presentations by utilizing their nonhuman origin. This process was supported by multiple (non-)human actors’ intentional, oblivious, and indirect participation, some of which were obscured.

Based on this, meta-authenticity was conceptualized as an analytic framework for artificial sociality’s dynamic, co-constructive multi-actor communicative process. This drew on a narrower definition of VIs and thus future research should refine it through comparative cases of artificial sociality (e.g. Kim, 2023). This can additionally help specify cultural beliefs about humanness, which itself must be approached as a slippery construct that interacts with, not equals, authenticity. Meta-authenticity can also be refined by recontextualizing VIs through influencer and media industries research (e.g. Hund, 2023; Yuan, 2024), expanding on this study’s artificial sociality focus. Finally, applications to other media contexts can help delineate authenticity contracts: for example, celebrities’ recovery from “scandals” may involve meta-authentic mechanisms, whereas “de-celebrifications” may reveal core authenticity principles (Mortensen and Kristensen, 2020).

Meta-authenticity contributes to artificial sociality research and ethics by illuminating the construction of authenticity through mutually evolving human norms and patterns. Black woman VI Shudu has received great attention for striking “realness,” particularly evoked through visual humanlikeness. After her White male creator—Cameron-James Wilson—was revealed, Shudu was re-approached as reproducing anti-Black, colonialist commodification of Black bodies (Sobande, 2021). Her portfolio continues to grow but Wilson has since become proactive about transparency (e.g. disclosures, crediting) and community-engaged design (e.g. Kami, VI with Down syndrome). Despite the commendable re-orientation, the structure persists. Wilson’s agency only features Shudu and Black male VI Koffi’s “muses,” excluding VIs like Dagny (White woman), J-Yung (Korean man), and Brenn (curvier Black woman) (https://www.thediigitals.com/muse, 13 July 2024): accountability for “realism” was selectively applied to more commercially or morally transactional bodies while other equally “real” bodies remained unexplained, deferred to the showcased examples. This is a tokenizing approach that only partially counteracts the pattern of bodily expropriation.

By narratively grounding artificial sociality on meta-authenticity, we can circumvent the limitations of binary ontologies and seriously consider the cyborg actors’ performances as “real,” both in terms of the process and consequences. This includes tracing humans’ imbrication. Thus, through meta-authenticity, we can also seriously consider how humans (are to) live with the emerging social reality. I argue that we can ethically become with artificial sociality by co-constructively demanding accountability from the interconnected actors, including ourselves. We—interconnected cyborgs—have the power to collectively push toward a culture of care.