Symposium

Synthetic kayfabe: Generative AI, provenance, and authenticity work in wrestling fandom

Andrew Firr

University of Chester, Chester, United Kingdom

[0.1] Abstract—Professional wrestling relies on negotiated authenticity, in which communities manage the boundaries between performance, documentation, and real encounters. Generative AI destabilizes these boundaries by making persona-consistent and archive-consistent media cheap to produce, increasing provenance uncertainty, and prompting visible authenticity work in wrestling fan spaces.

[0.2] Keyword—Deepfakes; Fandom governance; Online communities; Persona; Voice cloning

Firr, Andrew. 2026. "Synthetic Kayfabe: Generative AI, Provenance, and Authenticity Work in Wrestling Fandom." Transformative Works and Cultures, no. 49. https://doi.org/10.3983/twc.2026.3233.

1. Introduction

[1.1] Professional wrestling is built on boundary management, and audiences and communities negotiate what counts as performance, documentation, and credible evidence rather than simply following storyline content (Jenkins 1992; Hills 2002). In wrestling, a promo is a spoken performance segment, usually delivered by a wrestler or manager, that builds character, rivalry, storyline stakes, and audience investment. These negotiations are not a side effect of wrestling; they are part of its cultural machinery, shaped by recurring tensions between staged spectacle and claims of embodied risk and athletic legitimacy (Mazer 1998). Traditionally, this negotiated performance logic has been described through kayfabe: the maintenance of wrestling's shared fiction, in which performers, promoters, and audiences help sustain what should be treated as part of the show and what may count as reality. Wrestling fandom also has a long history of metacriticism; fans circulate interpretations, insider knowledge, and close readings that sit alongside the text (Jenkins 1992; Gray et al. 2007). In this setting, arguments about authenticity are not about whether something is scripted; they often become arguments about how to treat artifacts and framings as more or less evidentiary and what kinds of disclosure are owed to others in a shared space.

[1.2] Generative AI changes the conditions under which boundaries can be maintained. Wrestling has always staged conflict, but AI lowers the cost of producing evidence-like artifacts that borrow the form's established cues of authenticity: In wrestling terms, this includes promos in a recognizable voice, lost segments that match period texture, images that read as candid documentation, and clips framed as backstage access (Chesney and Citron 2018; Farid 2018). As these materials often circulate in short, decontextualized formats, the audience's task can turn from interpreting storyline claims to assessing provenance and intent across platforms. This creates a provenance problem (Farid 2018): When an artifact looks like evidence, communities must decide whether to treat it as evidence, disclosed play, or harm. Synthetic kayfabe names that emerging condition: a change from negotiated ambiguity toward contested evidence, where provenance becomes central and new norms and sanctions appear in response.

[1.3] The scope cuts across promotions because no single organization owns this mechanism. World Wrestling Entertainment (WWE) is the dominant mainstream reference point for persona recognition and archival nostalgia, while All Elite Wrestling (AEW) has a digitally engaged audience, making it a frequent site for remix circulation. Independent wrestling and convention culture are also important because claims of real encounter can be simulated and can cause direct personal harm. More broadly, wrestling fandom is already organized around voice, character continuity, nostalgia, speculative interpretation, and remix practices; therefore, generative AI enters a culture that is already highly participatory rather than creating that impulse from scratch. What AI adds is a low-cost way of producing things that official wrestling cannot, will not, or should not provide: impossible promos in familiar voices, speculative continuations of past eras, and plausible extensions of characters or scenarios beyond the limits of sanctioned programming.

2. Kayfabe and boundary management

[2.1] Kayfabe is often described as the insistence that wrestling is real, historically policed by promoters and performers. Contemporary wrestling is better described as a flexible regime of boundary practices in which communities move between frames: in-ring performance, promotional media, interviews, behind-the-scenes content, and social media (Mazer 1998; Sammond 2005). Each frame carries different expectations about what can be treated as evidence; wrestling has rarely been a simple binary of truth and fiction. The form invites partial belief and continually negotiates how much of the work should appear as an authentic athletic contest versus a theatrical performance. Promotions have historically produced realness by staging legitimate bodily risk, cultivating persona continuity across years, and borrowing documentary signals associated with sports broadcasting (Mazer 1998). This shared fiction is carried not only through matches but also through promos, interviews, backstage segments, commentary, and promotional framing, with performers and promoters each helping sustain what the audience is asked to accept as part of the show.

[2.2] This regime works because wrestling supplies stable authenticity cues to its audience. Some cues are institutional (broadcast graphics, segment formats, and commentary authority), some are embodied (voice, cadence, and signature phrasing), and others are documentary (handheld backstage footage, candid travel vignettes, and informal social posts presented as access). Fans learn to read these cues and argue about them (Jenkins 1992; Hills 2002). They debate what was worked (part of the planned performance), what was a shoot (understood as drawing more directly on unscripted or off-script reality), and what should be taken as credible beyond the storyline (Mazer 1998; Sammond 2005). Importantly, these arguments are social, normative, and interpretive. Communities also form expectations about when backstage information can be shared, what kinds of rumors count as acceptable speculation, and when claims should be sourced, corrected, linked to evidence, or withdrawn. In other fandom contexts, this is often discussed as a form of informal governance and norm formation (Gray et al. 2007). Wrestling intensifies this dynamic because the text is designed to provoke boundary disputes.

[2.3] Wrestling therefore offers a useful case study for observing what happens when evidence cues become unreliable. In many scripted genres, documentary-looking artifacts are clearly outside the work; in wrestling, documentary framing is often part of the work's reality system, and promotions regularly exploit that porosity through metareference and real-life bleed (Laine 2020; see also Booth 2015 for broader fandom context). When communities are trained to treat certain formats as closer to the record, synthetic artifacts that mimic these formats increase the workload of classification. That workload is epistemic and ethical: Communities must determine whether an artifact should be treated as a record, disclosed play, or misrepresentation, and those classifications carry consequences for how performers and other fans are spoken about, believed, and treated. The key question is how the artifact positions the audience: as coparticipants in play, as witnesses to the alleged backstage truth, or as consumers of implied intimacy that has not been granted. These judgments become more difficult when the artifact inherits wrestling's established cues of authenticity.

3. Synthetic kayfabe

[3.1] Generative AI is important because it can replicate authenticity cues at a low cost. The artifacts most relevant to synthetic kayfabe are cue-consistent materials that borrow documentary authority and compete with the record, rather than straightforward falsehoods about storyline outcomes (Chesney and Citron 2018; Farid 2018). These cues can be understood in three overlapping categories: persona, archival, and documentary cues. Persona cues are central, and voice is a high-salience identity marker in wrestling, especially in promo culture (Mazer 1998; Sammond 2005). A promo can feel plausible even when unofficial if the cadence and vocal texture match what the audience associates with a wrestler. A visible ecosystem now exists for explicitly AI-labeled wrestling promo content in this mode, including short-form videos titled as "A.I. Wrestling Promo[s]" for legacy performers (Pro Wrestling Revitalized 2025a, 2025b, 2025c). These items show that persona-consistent promo production is a repeatable, public practice.

[3.2] This is important because persona is more than aesthetic in wrestling; it is an identity contract that can carry reputational and commercial value and is sustained through repeated public performances across many settings (Laine 2020). When synthetic media can reproduce persona cues, the boundary between homage, parody, and appropriation becomes more difficult to police, especially on fast-moving social feeds where context collapses. Archival cues are also important; wrestling is archive rich and nostalgia driven, and eras have recognizable production rules. Generative tools can approximate period feel, turning the archive from a record into a style that can be generated on demand. In wrestling culture, archival media are regularly used as evidence in debates about canon and legitimacy; if archive-consistent artifacts become easier to generate, the archive's privileged status as what happened becomes easier to contest (Sammond 2005). Some of this material is openly playful or fan made, some is packaged by small creators as repeatable content, and some circulates in more deceptive ways; all of it enters the same evidentiary environment. In practice, the producer field is mixed, including fans, small creators, commercial toolmakers, and potentially promotional or adjacent institutional actors, even though these groups do not use AI for the same purposes. Fans may use it for parody, homage, nostalgia, or speculative extension; small creators may use it for repeatable content production; and institutional or adjacent actors may be more likely to use similar tools for promotional experimentation, audience engagement, or low-cost content generation.

[3.3] Documentary cues are the third category. Backstage framing, exclusive interview formats, and candid access aesthetics carry documentary authority in wrestling spaces, and wrestling has long played with these cues through worked shoots, where performance borrows the look or language of unscripted reality for the audience and through metastorytelling (Mazer 1998; Sammond 2005). AI changes the cost structure; documentary-looking artifacts can be produced without access to any accountable production chain while still borrowing the credibility of the documentary form (Chesney and Citron 2018; Farid 2018). A further change is productization; promo generation is marketed as a template category, explicitly framed as an "AI wrestling video promo generator" rather than as a specialist craft (Hooked, n.d.). Once the format is packaged this way, the barrier to producing cue-consistent content drops again, and the volume of artifacts becomes less constrained by technical skills.

[3.4] Synthetic kayfabe names the point at which replicated cues begin to compete with boundary practices that normally stabilize wrestling realness. Classic kayfabe is managed fiction that invites partial belief, while synthetic kayfabe is managed ambiguity, in which provenance becomes central. What is the origin of this artifact, and what claims are being made regarding its aesthetic credibility? The consequence is both deception risk and the redistribution of interpretive labor and moral risk onto communities. Synthetic kayfabe also intersects with transformative practice, as wrestling fandom already sustains remixed cultures: parody promos, fantasy booking (fans imaginatively scripting their own match outcomes, storylines, or character directions), highlight edits, and meme formats that play with persona and continuity.

[3.5] Generative tools extend these practices by lowering production barriers and enabling rapid iterations of voice, era texture, and documentary framing. The same affordances that support playful transformations also increase the risk of misrecognition when artifacts are detached from disclosure and circulated as evidence. In this sense, AI is attractive not because it creates participatory desire but because it amplifies existing fan interests in voice, continuity, nostalgia, and speculative extension. It also offers a way to generate material that official wrestling leaves unavailable: alternative bookings, revived archival styles, and persona-consistent performances outside the formal production control. Historically, this is significant because wrestling has long depended on controlled ambiguity rather than transparent authorship. In a postkayfabe environment, audiences already know that performance and construction coexist; however, they still rely on promotions, performers, journalists, and fan communities to stabilize what counts as part of wrestling's shared realities.

[3.6] Synthetic kayfabe threatens this stabilizing function by allowing plausibly authentic materials to circulate without accountable authorship. The risk is not only that audiences are misled in individual cases but that wrestling's already fragile distinction between playful ambiguity and evidentiary confusion becomes more difficult to maintain.

[3.7] Therefore, the mechanism of synthetic kayfabe is best understood as a shift in the evidentiary environment. Wrestling communities already manage uncertainty about what is really happening behind the performance. AI expands this uncertainty by enabling the creation of plausible artifacts that borrow the cues that audiences use to assign credibility. This creates more frequent provenance disputes and increases the demand for governance responses that can traverse platforms.

4. Authenticity work and governance responses

[4.1] If synthetic kayfabe is a provenance problem, authenticity work is how communities contain it. These practices are common across online participatory cultures but take a distinctive shape in wrestling because the form already depends on negotiated boundaries. Several practices are especially visible, including disclosure demands, boundary policing, sanctions and escalation, and credibility repair. Disclosure demand is the baseline practice: When an artifact appears to be documentary, users ask whether it is real, who made it, and whether it is AI generated. Even when content is self-labeled, the label functions as a governance signal about how the artifact should be treated in circulation, separating disclosed play from implied evidence claims (Hetcher 2009). The "A.I. Wrestling Promo" format is useful here because it demonstrates cue replication while also showing that labeling is already part of how some creators manage reception (Pro Wrestling Revitalized 2025a). Where labeling is clear, responses may include curiosity, amusement, or technical interest as well as skepticism, because the issue is not AI in the abstract but how the artifact is being framed and what kind of claim it is asking others to accept. In other words, reactions are not best understood as simple acceptance or rejection of AI but as judgments about framing, disclosure, platform context, and whether the artifact is inviting play or asserting truth.

[4.2] Boundary policing is the next layer; this is not straight hostility toward AI but a decision about what belongs in a space and what should be removed, restricted, or quarantined. Norm talk appears in community discussions about AI-generated promo writing, including joking evaluation, comparative judgment, and signs that some users already treat these outputs as a distinct content category rather than as ordinary fan expression ("AI Now Can Write Wrestling Promos on Every Subject," https://www.reddit.com/r/Wrasslin/comments/11xgmot/ai_now_can_write_wrestling_promos_on_every_subject/). These debates are part of how spaces define acceptable play, and credibility repair becomes salient when synthetic claims are circulated as evidence. When false claims spread, communities post corrections, and performers sometimes intervene directly. The Zoey Stark retirement deepfake story is a useful example; coverage describes an AI-made clip using her likeness and voice and notes her efforts to shut down the claim (Upton 2025). In wrestling, repair often reasserts a boundary; storyline play is not the same as real-world claims about a performer's life. In these cases, the dominant affect is less playful fascination than irritation, skepticism, and corrective action because the artifact is being treated as an attempted evidentiary claim rather than a disclosed transformation. At the organizational level, the most plausible immediate pattern is selective management rather than blanket adoption: Promotions and event spaces may tolerate clearly disclosed playful remixing, but they have stronger incentives to resist or contain synthetic artifacts when performer likeness, event credibility, reputational harm, or false documentary claims are involved.

[4.3] Sanctions and escalation become most visible when synthetic media simulate real encounters. Conventions and meet and greets sit close to the realness boundary because they trade on proximity to the performers. When synthetic media fabricate intimacy or contact, they can shift from play to coercive misrepresentation. A strong anchor is the AJ Lee deepfake incident. Reporting describes an AI-generated kissing video featuring AJ Lee posted by a fan at a wrestling event; AJ Lee described it as harassment and requested its deletion, and the event organizer reportedly banned the fan from all future events (Powell 2025). This is the point at which synthetic wrestling media move decisively beyond remix or tribute and into nonconsensual persona use, prompting the strongest forms of sanctioning.

[4.4] This incident also appears in an external monitoring layer; the OECD AI Incidents and Hazards Monitor records the AJ Lee case as an AI incident involving nonconsensual synthetic media (OECD, n.d.). This is useful because it shows that the case traveled beyond wrestling-only media into a wider incident-tracking infrastructure. This also indicates how quickly synthetic media can be treated as a governance and safety problem rather than a creative practice. These responses point toward a normative-first governance pattern, in which communities tend to rely on disclosure requirements, removal, and sanctions before technical provenance solutions become widely usable, particularly because verification tools are uneven and slow to standardize (Farid 2018). This pattern is likely to persist even as detection improves, since the core issue concerns the acceptability of circulation as much as the synthetic status of an artifact, within a community that also recognizes performers as people alongside their on-screen roles. Based on the visible cases discussed, promotions, event organizers, and performers appear more likely to manage or resist synthetic artifacts when they threaten consent, performer image, or evidentiary trust than when they are clearly disclosed and low stakes.

[4.5] The strongest governance pressure is likely to cluster around artifacts that imply backstage access, sexualized intimacy, or endorsement because those artifacts exploit documentary cues to cross from interpretive play into coercive misrepresentation. This is where synthetic kayfabe becomes more than misinformation; it becomes a question of consent, harassment, and moral limits of persona replication.

5. A mechanism of synthetic kayfabe

[5.1] Synthetic kayfabe can be understood as a mechanism in which replicated authenticity cues increase provenance uncertainty and prompt authenticity work. Wrestling supplies a stable repertoire of cues that fans use to evaluate artifacts, including persona markers, archival textures, and documentary frames (Mazer 1998; Sammond 2005). Generative AI lowers the cost of reproducing those cues and increases the volume of cue-consistent artifacts that are not anchored to accountable production contexts (Chesney and Citron 2018; Farid 2018). The result is more than just additional false content; it is a shift in the evidentiary environment. Audiences and communities must decide how to classify an artifact whose style carries credibility, while its origin may remain unclear.

[5.2] The model is based on two linked questions: how strongly an artifact reproduces wrestling's authenticity cues and how clearly its synthetic status is disclosed. When the cue strength is high and disclosure is weak, the artifact is more likely to be treated as a possible evidentiary claim. This is especially likely when the artifact imitates backstage access, performer voice, archival footage, or intimate encounter. When disclosure is clear, the same technical affordances may be understood as a parody, homage, fantasy booking, or speculative play. Therefore, ethical and governance problems do not arise from synthetic production itself. They arise when synthetic production borrows evidentiary form without providing communities with sufficient information to judge origin, intent, and consent-related stakes.

[5.3] This mechanism helps distinguish lower-risk transformations from more harmful misrepresentations. A clearly labeled AI-generated wrestling promo may still raise questions about taste, ownership, or performer likeness, but it gives audiences a frame through which to understand the artifact as a transformation rather than as evidence. In contrast, an unlabeled or misleading clip that appears to show a performer making a real announcement, endorsing a claim, or participating in an intimate encounter places a heavier verification burden on fans, performers, journalists, moderators, and event organizers. In these cases, authenticity work becomes corrective work: Users ask for sources, challenge circulation, clarify disclosure, remove content, or impose sanctions.

[5.4] Therefore, the model can be stated as follows: The stronger the replicated authenticity cue and the weaker the disclosure, the greater the pressure for provenance dispute and governance response. Wrestling makes this pressure unusually visible because its culture depends on managed ambiguity, performer continuity, disputed evidence, and partial disclosure. Synthetic kayfabe emerges when that ambiguity is no longer stabilized mainly by shared performance conventions but by ongoing judgments about provenance, disclosure, consent-related risks, and harm.

6. Conclusion

[6.1] Wrestling fandom is a useful site for observing what happens when provenance becomes unstable in a culture already organized around managed ambiguity, persona continuity, and contested evidence (Mazer 1998; Sammond 2005). Generative AI does not just introduce false content into this environment; it reproduces the cues by which audiences ordinarily sort archive, performance, and documentary claims, increasing the amount of authenticity work required in circulation (Chesney and Citron 2018; Farid 2018). Synthetic kayfabe names that shift from negotiated ambiguity to contested evidence. In the near term, the containment of these artifacts is likely to depend less on fully reliable verification tools and more on disclosure norms, local moderation, and community sanctions (Gillespie 2018; Farid 2018). The strongest pressure will likely continue to fall on artifacts that imply backstage truth, intimate access, or nonconsensual persona use. Similar pressures may emerge in other media forms that trade on managed realness, including reality television.

7. Acknowledgment

[7.1] No human subjects were recruited for this study. This article relies only on secondary materials: publicly documented wrestling-related AI incidents and their public reception, a purposive sample of publicly accessible wrestling-related posts or videos explicitly labeled as AI generated, and publicly available governance texts from fan and event spaces. The sampling logic follows qualitative social media research practice, in which bounded, case-led collections are used to illuminate a mechanism rather than produce population estimates (Kozinets 2020). Ethical handling follows the Association of Internet Researchers' guidance for Internet research, including careful treatment of sensitive materials and avoidance of reproducing harmful artifacts (franzke et al. 2020). For this reason, this article relies on reporting and monitoring entries for harassment-related deepfake incidents rather than reproducing or linking to harmful media artifacts. The discussion of fan community norms uses publicly available posts and threads interpreted at the level of general governance language rather than individual profiling.

8. References

Booth, Paul. 2015. Playing Fans: Negotiating Fandom and Media in the Digital Age. University of Iowa Press.

Chesney, Robert, and Danielle Citron. 2018. "Deepfakes and the New Disinformation War: The Coming Age of Post-Truth Geopolitics." Foreign Affairs, December 11. https://www.foreignaffairs.com/articles/world/2018-12-11/deepfakes-and-new-disinformation-war.

Farid, Hany. 2018. "Digital Forensics in a Post-Truth Age." Forensic Science International 289:268–69. https://doi.org/10.1016/j.forsciint.2018.05.047.

franzke, aline shakti, Anja Bechmann, Michael Zimmer, Charles M. Ess, and the Association of Internet Researchers. 2020. "Internet Research: Ethical Guidelines 3.0." Association of Internet Researchers. https://aoir.org/reports/ethics3.pdf.

Gillespie, Tarleton. 2018. Custodians of the Internet: Platforms, Content Moderation, and the Hidden Decisions that Shape Social Media. Yale University Press.

Gray, Jonathan, Cornel Sandvoss, and C. Lee Harrington, eds. 2007. Fandom: Identities and Communities in a Mediated World. New York University Press.

Hetcher, Steven A. 2009. "Using Social Norms to Regulate Fan Fiction and Remix Culture." University of Pennsylvania Law Review 157 (6): 1869–935. https://repository.law.upenn.edu/Documents/Detail/using-social-norms-to-regulate-fan-fiction-and-remix-culture/154517.

Hills, Matt. 2002. Fan Cultures. Routledge.

Hooked. n.d. "AI Wrestling Video Promo Generator." Accessed March 16, 2026. https://www.tryhooked.ai/tools/ai-wrestling-promo-generator.

Jenkins, Henry. 1992. Textual Poachers: Television Fans and Participatory Culture. Routledge.

Kozinets, Robert V. 2020. Netnography: The Essential Guide to Qualitative Social Media Research. 3rd ed. Sage.

Laine, Eero. 2020. Professional Wrestling and the Commercial Stage. Routledge.

Mazer, Sharon. 1998. Professional Wrestling: Sport and Spectacle. University Press of Mississippi.

OECD (Organisation for Economic Co-operation and Development). n.d. "WWE Star AJ Lee Harassed by AI-Generated Deepfake Videos." AIM: AI Incidents and Hazards Monitor. Accessed November 20, 2025. https://oecd.ai/en/incidents/2025-11-20-5eb3.

Powell, John. 2025. "AI Deepfake of AJ Lee Leads to Lifetime Ban for Fan." Slam Wrestling, November 21. https://slamwrestling.net/news/ai-deepfake-of-aj-lee-leads-to-lifetime-ban-for-fan/.

Pro Wrestling Revitalized (@ProWrestlingRevitalized). 2025a. "Ox Baker A.I. Wrestling Promo." YouTube, October 1. Video, 2 min., 25 sec. https://www.youtube.com/shorts/5GBdypiYrQo.

Pro Wrestling Revitalized (@ProWrestlingRevitalized). 2025b. "Superstar Billy Graham A.I. Wrestling Promo." YouTube, October 1. Video, 2 min., 23 sec. https://www.youtube.com/shorts/31BnV1wKN1I.

Pro Wrestling Revitalized (@ProWrestlingRevitalized). 2025c. "Harley Race A.I. Wrestling Promo." YouTube, October 2. Video, 2 min., 47 sec. https://www.youtube.com/watch?v=CuWbaD8uQ3Q.

Sammond, Nicholas, ed. 2005. Steel Chair to the Head: The Pleasure and Pain of Professional Wrestling. Duke University Press.

Upton, Felix. 2025. "Zoey Stark Shuts Down Viral AI Deepfake Claiming She Retired from WWE." Ringside News, June 16. https://www.ringsidenews.com/zoey-stark-shuts-down-viral-ai-deepfake-claiming-she-retired-wwe/.