Abstract
Increasingly, the global public relies on social media for information gathering and opinion formation but the spread of misinformation is a major global risk. Though this is not necessarily a new problem, the prevalence of misinformation and disinformation and their speed in dissemination via social media represents an important challenge for democracies. With the rapid rise of short-form video platforms, our study explores the fact-checking efforts by both fact-checking organizations and influential individuals on TikTok. Findings reveal the diversity of fact-checking content strategies applied by professionals and the public through the application of TikTok features and affordances. The study highlights how fact-checking content adapts to a platform's unique affordances and user culture highlighting the dynamic, interactive, and diverse efforts to address misinformation using short-form video.
Introduction
The global public relies on social media for information gathering and cognition (Kemp, 2023; Ortiz-Ospina, 2019), but the spread of misinformation is a major global risk (World Economic Forum, 2023). Fact-checking, media literacy, policy, and computational approaches are effective strategies to combat misinformation on social media (Haque et al., 2020). Fact-checking efforts have doubled globally since 2015, with around 400 teams tackling political lies and hoaxes in 105 countries (Stencel et al., 2022). This global fact-checking movement has contributed to the evolution of professional fact-checkers’ networks and modes of governance (Lauer & Graves, 2025). Meanwhile, social media is shifting from text and image-based content to short and long-form video. Regarding current social media platforms, TikTok is the fastest-growing social media platform ever with its vast networked public (boyd, 2010; Kemp, 2023; Pardeshi, 2024; Singh, 2026). Given the rise of fact-checking and TikTok, our study explores fact-checking practices on the platform. More specifically, it investigates the content differences between public-used #factcheck and content from professional fact-checker PolitiFact, focusing on themes, creator profiles, and narrative types, as well as the use of platform features.
Literature review
False information and fact-check
A historically common issue in communication practices is the prevalence of misinformation and disinformation or simply false information (Mejia et al., 2018; Vida, 2012; Wardle, 2019). Wardle and Derakhshan (2017) conceptualize false information through the framework of information disorder, which considers harm and falseness to identify misinformation, disinformation and malinformation. Within this framework, false information encompasses misinformation and disinformation. Misinformation, drawing on an unintentional spread, includes false connections and misleading content. Disinformation, indicating an intentional spread, emphasizes false context, imposter content, manipulated content, and fabricated content. To address false information, one may attempt to verify the accuracy of the information by performing a fact check. As a specialized occupation, a fact-checker implements internal or external fact-checking research practices (Graves & Amazeen, 2019) to evaluate the veracity of assertions made by institutions and public figures to determine whether they are true or false (Walter et al., 2020). A fact-checking report offers arguments and supporting evidence to help debunk false information (Naeem & Bhatti, 2020). This study focuses on verification practices related to pure factual statements. Accordingly, it aligns with fact-checking efforts that primarily target false information, specifically misinformation and disinformation.
Over the last decade, fact-checking efforts expanded significantly on social media (Brandtzaeg et al., 2018; Lauer & Graves, 2025), with platforms like Facebook, X (Twitter), and YouTube taking steps to curb the spread of false information by collaborating with fact-checking organizations (Lauer & Graves, 2025; Ma & Feldman, 2022; Meta, 2025). The launch of the International Fact-Checking Network (IFCN) in 2015 at Poynter further institutionalized global fact-checking efforts, empowering fact-checkers through networking and training. Notable contributions from organizations like PolitiFact, which won a Pulitzer Prize in 2008, highlighted the critical role of fact-checking in maintaining journalistic integrity in a “posttruth” era (Graves, 2018; Graves & Amazeen, 2019; Haque et al., 2020).
The emergence of the COVID-19 pandemic in 2020 marked the beginning of the “infodemic” era (WHO, 2020), with fact-checkers facing unprecedented challenges in combating the vast amount of misinformation surrounding the global pandemic (Suárez, 2020). This infodemic is characterized by an overabundance of both accurate and inaccurate information, and represents a fundamental crisis of information distillation, unfolding across four hierarchical levels: science, policy and practice, news media, and social media (Eysenbach, 2020). The infodemic intensified the need for fact-checking, especially on social media platforms, which faced significant pressure to manage false or misleading content. By filtering and analyzing information, professional fact-checkers aim to bridge the knowledge translation gap between the best scientific evidence and public belief. Meta and Google have primarily supported fact-checking efforts by funding initiatives and providing tools to label and demote misleading content (Bell, 2019). Professional fact-checkers and platforms cooperated to deploy intervention toolkits, such as warning labels, that function as digital “antibodies” or forms of informational immunization, designed to alert users to problematic content and disrupt the mindless consumption of misinformation commonly found on social media (Mende et al., 2024). However, the lack of transparency in how platforms determine which content to remove or flag for fact-checking remains a point of contention (Cotter et al., 2022).
Scholars have debated the effort to combat misinformation through fact-checking. Experimental studies showed the positive impact of fact-checking information on people's political knowledge (Fridkin et al., 2015). However, correcting inaccurate information still shapes politically biased beliefs (Thorson, 2016), as well as potential “backfire” effects (Jiang & Wilson, 2018). Furthermore, Van Aelst et al. (2017) pointed out that evidence about how political misconceptions might be corrected through fact verification is inconclusive.
Fact-checking and social media
The rise of social media platforms has fundamentally transformed modern journalism, shifting it from traditional one-way broadcasting to interactive two-way communication (Singer, 2010; Van Aelst et al., 2017). This shift has posed challenges to the authority and function of mainstream media, with social media becoming the primary channel through which public figures communicate directly with the public. As misinformation spreads rapidly through these platforms, the need for fact-checking has grown significantly, resulting in the establishment of fact-checking organizations across the globe (Brandtzaeg et al., 2018; Graves & Amazeen, 2019). These organizations have adapted to this new environment by cooperating with major platforms like Meta and Google. Since 2016, Meta has partnered with the IFCN to flag false content, while Google has implemented “fact-check tags” across its platforms to help users identify verified information (Ma & Feldman, 2022). These partnerships aim to curb the spread of misinformation by promoting truth in public discourse, a goal shared by fact-checking organizations (Graves & Cherubini, 2016).
As misinformation spreads faster and further than facts in many cases (Ball & Maxmen, 2020; Lewandowsky et al., 2017), fact-checking on social media platforms has evolved beyond simply verifying public figures’ claims to addressing viral misinformation (Graves et al., 2024). Platforms like Facebook, Twitter, and YouTube have employed various strategies to limit the visibility of false or misleading content, including the use of warning labels and post demotion. The use of fact-checking labels, such as “Rated false” or “Disputed,” has been shown to influence user behavior, such as discouraging the sharing of misinformation or altering users’ perceptions of content accuracy (Ardèvol-Abreu et al., 2020; Clayton et al., 2020). Studies indicate that these labels can have positive effects, particularly in the context of vaccine misinformation (Zhang et al., 2021). However, some researchers argue that the success of these labels depends on how they are presented (Schuetz et al., 2021). In this new digital media landscape, fact-checkers are continually adapting their strategies to combat the challenges posed by viral misinformation, technical coordination issues, and the balance between platform governance and independent fact-checking efforts (Bélair-Gagnon et al., 2023). Thus, the integration of fact-checking into social media governance remains a critical and evolving aspect of addressing misinformation in today's digital environment.
The integration of algorithms into fact-checking processes is a significant development in how platforms manage misinformation. For Shin et al. (2025), algorithmic curation has shifted the landscape from editorial gatekeeping to infrastructural governance. For example, algorithms now classify and prioritize verification claims, source quality and content amplification (p. 193). An infrastructural approach acknowledges the importance of journalistic corrective gatekeeping toward a form of public knowledge verification practice oriented around algorithmic mediation. Fact-checking organizations now face the challenge of investigating content that is often more complex than what algorithms can easily detect, including satire, opinion posts, and other subtleties (Stewart, 2021). While platforms have utilized AI to detect false information (Gillespie, 2018, 2020), human oversight remains crucial to effectively identify and address viral hoaxes and disinformation campaigns (Adair & Stencel, 2020). The expansion of automated fact-checking tools, machine-learning-based detectors, and multilingual approaches has been suggested as a way to further enhance the fact-checking infrastructure (Flew et al., 2012; Shahi & Nandini, 2020; Walter et al., 2020). In rising to the challenges of managing misinformation in real-time across global platforms, Google's grant programs have provided fact-checkers with the resources needed to scale up their efforts (Lelo, 2022). These technological advancements have evolved alongside social engineering tactics, which platforms have increasingly used to address new forms of media manipulation circulating on social media (Gehl & Lawson, 2022). One of the most recent examples is the 2024 Indian General Election. Roy et al. (2025) highlight that professional fact-checkers operated as retroactive gatekeepers, collaborating with audiences to intervene after disinformation had already circulated widely on social media.
Social media and affordances
The concept of “affordance” originally referred to a specific type of relationship between an animal and its environment in ecological psychology (Gibson, 1979), and was later applied to the field of Human–Computer Interaction. Conceptualized as a relational property (Bucher & Helmond, 2018), Gaver (1991) proposed “technological affordances” to emphasize user-centered interactions with technology and to highlight how environmental factors shape social interaction. As an emerging example, artificial intelligence (AI) and generative AI (GAI) tools for fact-checking illustrate how AI- and GAI-based technologies offer functional affordances designed for specific verifying purposes. However, professional fact-checkers remain critical and cautious in adopting these tools due to ongoing concerns about accuracy and reliability (Dierickx et al., 2024).
In social media research, affordance theory has been widely applied to unpack the infrastructural features of digital platforms. The high-level affordances refer to the broader sociotechnical dynamics and conditions enabled by technical devices, platforms, and media, shaping patterns of action and interaction. In contrast, the low-level affordances refer to feature-oriented interface elements, such as specific platform buttons, which structure users’ immediate possibilities for action (Bucher & Helmond, 2018). For instance, the practice of using independent audio and postproduction tools in video creation reflects low-level affordances at the feature level. In contrast, recurring and standardized patterns in how audio and postproduction are combined, captured by the concept of templatability (Leaver et al., 2020), demonstrate a high-level affordance that shapes broader practices of content creation.
TikTok
TikTok, launched globally in 2017 after ByteDance acquired Musical.ly, has rapidly become one of the most popular social media platforms worldwide in over 150 countries. The app allows users to create short-form videos, typically 15–60 seconds long, and offers various creative tools like filters, sound effects stickers, emoticons, and vibrant text. TikTok's appeal lies in its engaging features, which foster user creativity and collaboration. With over 2 billion downloads (Pardeshi, 2024) and 1.99 billion monthly active users (Kemp, 2023; Singh, 2026), TikTok's success shows that social media is evolving from text-based, image-based, and video-based platforms, to short-form video.
TikTok's success can be attributed to its personalized content feed and interactive design (Schellewald, 2023). Kaye et al. (2022) noted that socially creative features, such as Duet, Stitch, Video Reply to Comments, and Use this Sound, significantly accelerate video duplication on TikTok. Socially creative features refer to the affordances that structure and shape creative, collaborative, and cumulative behaviors that are enabled and enacted by platform activity (Lebuda & Glăveanu, 2019). In addition, socially creative features on TikTok encourage participation both through their intended use and through the numerous ways users have reappropriated and subverted their affordances (Kaye et al., 2022).
TikTok and community
According to Kaye et al. (2022), TikTok favors community formation through content discovery rather than network-centered ties such as Facebook or Twitter/X (p. 91). This creativity-driven sociality leads to different types of communities on the short-form video platform: mainstream pop-culture-centric communities and niche communities formed through the interaction of algorithms, affordances and creativity. Popular communities comprise Internet celebrities, offline celebrities, notable short-form video creators migrating to TikTok and the TikTok-famous who achieved success through viral content. As opposed to popular communities, TikTok also nurtures niche communities that are effectively forms of networked publics (boyd, 2010) or even imitation publics (Zulli & Zulli, 2020) built by memetic interaction of content and engagement. For example, the For You page (FYP) encourages users to engage and imitate the content of other users applying socially creative features to improvise their own unique contribution to these lively and dynamic communities. In this sense, TikTok users build community through content discovery where “TikTok's machine-learning models and recommender systems labeled the users’ content as similar” (Kaye et al., 2022, p. 110). For niche communities, features such as the recommendation algorithm become key sociotechnical elements that lead to discovering new communities. In addition, another important factor leading to community is participation. The socially creative features of TikTok such as Duet and Stitch lead to distributed creativity where groups of individuals come together to generate new and creative output. Through spontaneous collaborative emergence and creativity, TikTok users participate in the long tail of content and creators (Kaye et al., 2022).
Fact-checking on TikTok
TikTok, like other major platforms, faces challenges with misinformation. To address this, it launched a Fact-Checking Program in collaboration with industry-leading fact-checkers and implemented in-app reporting tools (Bettadapur, 2020; Sidorenko Bautista et al., 2021). During the 2021 German federal election, TikTok partnered with Germany's DPA to label election-related content using automated systems (Bösch & Ricks, 2021). Ruak (2023) noted TikTok's efforts in reducing hoax news spread, while Bhargava et al. (2023) found that debunking videos effectively improved users’ ability to recognize misinformation, though preexisting beliefs still influenced their judgment.
Studies on combating misinformation through fact-checking on TikTok highlight both challenges and innovations. Bösch and Ricks (2021) observed improper labeling of election content in Germany during the 2021 federal election and recommended early collaboration with fact-checking teams. Alonso-López et al. (2021) found that TikTok accelerates the spread of misinformation, but also serves as an effective tool for countering it. Sidorenko Bautista et al. (2021) examined fact-checking practices across thirteen organizations and noted that fact-checkers adapted TikTok's vertical video format and editing tools to capture user attention, particularly in humanitarian and health topics. Lu and Shen (2023) found that fact-checking videos on Douyin (Chinese TikTok) were more visually dynamic, using brighter scenes and faster pacing compared to nonfact-checking content. Grover (2023) examined crowdsourcing approaches to combating misinformation on TikTok, highlighting that features like Duet and Stitch may allow users to use lopsided material to promote their viewpoint.
However, research on how TikTok's features support fact-checking content remains limited, and TikTok currently lacks specific tools that encourage user-generated content for debunking misinformation. With the rapid development of the fact-checking movement and TikTok's evolving platform features, it is reasonable to expect differences in how fact-checking content is presented on the platform. Given the positive impact of fact-checking, understanding the landscapes associated with fact-checking on TikTok is crucial.
Professional versus public fact-checking
Previous studies analyzed content posting by fact-checking organizations and fact-checkers to understand the landscape of fact-checking content on social media (Bélair-Gagnon et al., 2023; Graves et al., 2024; Lu & Shen, 2023; Sidorenko Bautista et al., 2021). Fact-checking organizations have paid more attention to “debunking” types of fact-checking content (Cazzamatta & Santos, 2024) and mainly publish humanitarian and health field topics on TikTok (Sidorenko Bautista et al., 2021). However, much less is known about the type of topics and content that public fact-checkers address on social media platforms. For example, content produced by professional fact-checking organizations is expected to comply with IFCN standards, presenting information in a structured and authoritative manner on TikTok. In contrast, public-generated #factcheck content reflects a decentralized mode of participation, shaped by the platform's socially creative features and participatory culture. Therefore, TikTok fact-checkers are more influenced by creator culture than traditional journalist standards. Compared to professional fact-checking organizations, crowdsourced volunteers who create fact-checking content may be less authoritative but better aligned with the digital tools that comprise the media landscape for fact-checking (Liu et al., 2025). Accordingly, #factcheck content creators emerge from the crowd-sourcing community (Graves, 2016; Liu et al., 2025) and strive as influencers to build trust among their followers. These key distinctions underscore the gap between institutional expertise and grassroots engagement in algorithmic-mediated digital fact-checking practices. Understanding these differences is essential for developing more effective, platform-specific strategies to counter misinformation on TikTok.
In a study on TikTok profiles that misinform, Alonso-López et al. (2021) characterized various narrative features such as the selfie, TikTok challenge, music and text as resources the creator draws upon to develop their message. Their study determined that “digital disinformers” were individual users who typically posted selfies or reposted videos from other digital platforms. Some digital disinformers applied music and text to their videos and hardly ever used TikTok challenges. In so doing, digital disinformers “insert videos and photos from other digital platforms, taken out of context and reinforced with biased texts” (p. 80). Yet, little is known about how public or professional fact-checkers utilize TikTok's narrative features and apply its socially creative features to advance their message.
As noted previously, there is a significant shift in how fact-checking and verification practices occur as algorithms, protocols and platform logics replace traditional editorial oversight. Shin et al. (2025) argue that hybrid arrangements of human and machine labor operate as a three-level analytical framework. This framework conceptualizes the epistemic labor of individual fact-checkers as the micro level, the institutional norms of fact-checking organizations as the meso-level, and the broader architectures of platform governance and global information flows as the macro-level. The boundaries between these layers are permeable as infrastructures support and also shape verification practices. For TikTok, little is known about the epistemic labor of TikTok fact-checkers. In other words, how do TikTok fact-checkers use their discretion and interpretive agency to create content? Similarly, fact-checking organizations interact with TikTok infrastructure to create fact-check content. In doing so, these choices become part of their institutional practice. Although much of TikTok's infrastructure and recommendation algorithm lack transparency, we can examine how different actors apply TikTok's socially creative features and begin to gain some insight into how TikTok's infrastructure shapes fact-checking practices on its platform.
Research questions
With all the rapid development in the fact-checking movement and TikTok platform features, it stands to reason that there will be platform affordance differences among creators in fact-check content on TikTok. Given the beneficial effects of fact-checking, exploring content associated with fact-check use on TikTok is imperative. The main research question of this study is: What are the differences between the public-used #factcheck content and their creators compared to a professional fact-check organization's content on TikTok?
To unpack fact-checking practices on TikTok, this study adopts the analytical framework of epistemic infrastructure (Shin et al., 2025), examining epistemic labor, institutional practices, and their interactions with infrastructure. As an exploratory step toward understanding the platform's algorithmic operations, this study focuses on TikTok's creative affordances in the analysis.
For public #factcheck creators, it is important to examine their interpretive agency and discretionary choices through specific themes and content decisions that instantiate their epistemic labor. In addition, #factcheck creators apply TikTok's affordances and features to navigate the platform's recommendation algorithms providing insights into the interaction between epistemic labor and infrastructure. The following research questions are asked for popular #factcheck content creators. RQ1: What general themes, creator profiles, and narrative types are prevalent in popular #factcheck on TikTok? RQ2: What combinations of platform features are used in popular #factcheck on TikTok? RQ3: What general themes and narrative types are prevalent by the professional fact-checker PolitiFact on TikTok? RQ4: What combinations of platform features are used by the professional fact-checker Politifact on TikTok?
Methodology
To explore #factcheck on TikTok, our analysis incorporates aspects of social media data analytics and content analysis.
Data collection
This study aimed to explore the landscape of fact-checking content on the TikTok platform. In April 2023, the authors utilized a web-scraping service (Bright Data, 2025) to acquire TikTok content. The sampling frame covered a 2-year period and consisted of two data collection requests. Notable political events were the Russian invasion of Ukraine and the U.S. 2022 midterm elections (Hodge et al., 2022, Wolf & Cohen, 2022). All short-form videos were captured for data analysis.
The first data collection scraped public TikTok posts with the #factcheck hashtag yielding 108 TikTok posts dating back to May 2021 uploaded by 72 unique TikTok users. Sixteen posts were removed because they were not relevant or duplicates. Notably, although this study used an English hashtag, the sampled videos included content in other languages, such as Spanish and German, and posts from accounts like Bolivia Verifica.
The second data collection captured TikTok posts uploaded by PolitiFact, a fact-checking organization and an indicator member of the IFCN. 159 TikTok posts from PolitiFact (@politifact) were collected from May 2021 and matched the first data collection time frame. PolitiFact serves as a core organization within the professional fact-checkers’ network, functioning as a key community builder. It is directly operated by the Poynter Institute, which also leads the IFCN (Lauer & Graves, 2025). As one of the first fact-checking databases and the creator of the trademarked Truth-O-Meter (Graves, 2016), PolitiFact is considered a benchmark for studying professional fact-checking practices in this study.
Together, three PolitiFact TikTok videos were removed because they were in the #factcheck collection. The final data set consisted of 248 TikTok posts dating from May 2021 to March 2023 uploaded by 60 unique TikTok content creators. See Figure 1.
Inclusion of TikTok #factcheck and @politifact posts, and reasons for exclusion.
Data analytical approach
Publicly available TikTok video metadata included account username, timestamp, likes, comments, url, and so on. Although engagement on social media can be analyzed a variety of ways, our study uses a conventional measure defined as likes plus comments (Buente et al., 2020; Carah & Shaul, 2016; Chen et al., 2021; Chugh et al., 2019). Social media engagement can be conceptualized into three tiers: low-level manifestations, mid-level relational connections, and higher-level actions and impacts (Johnston & Taylor, 2018). Within TikTok's algorithmically curated environment, the primary unit of analysis is creators’ short-form videos rather than users’ social networks. Low-level engagement measurements, including views, likes, and visits, primarily reflect the popularity of the fact-checking content itself. Verification practices may further unfold through mid-level discursive interactions in comment spaces and higher-level downstream consequences. However, these dimensions are beyond the scope of this study, which focuses on examining the levels and interactions of fact-checking infrastructural governance.
For content analysis, our codebook followed prior studies in TikTok disinformation, fact-checking and influencers (Alonso-López et al., 2021; Campbell & Farrell, 2020; Kaye et al., 2022; Rauchfleisch et al., 2022; Sidorenko Bautista et al., 2021). To test codebook reliability, 60 posts were randomly extracted from the TikTok dataset (N = 248) and assigned to two college students trained on the five codebook categories discussed in the next section. The reliability test indicated an acceptable Krippendorff's alpha for all variables (Table 1).
Reliability Test of Coding Scheme.
Codebook categories
General themes (Table 2) are based on scholarly research on disinformation and fact-checking on TikTok (Rauchfleisch et al., 2022; Sidorenko Bautista et al., 2021). General themes reflect common misinformation domains such as health, environment, politics and others. The “others” category entailed the indirect fact-checking-related videos such as opinions about fact-checkers.
Coding Scheme of General Themes, Creator Profiles, and Narrative Types.
IFCN = International Fact-Checking Network.
Two additional categories examined the content creator and their use of narrative evidenced by scholars studying TikTok disinformation (Alonso-López et al., 2021) and social media influencers (Campbell & Farrell, 2020). Narrative types represent various presentation content strategies to tell a story. Creator profile distinguishes between individuals, influencers, fact-checking organizations and fake content creators. See Table 2.
The coding scheme of platform features (Table 3) contained audio use and postproduction. The platform feature categories were inspired by TikTok's socially creative features. Audio use indicated the music type, original audio or multiple soundtracks in the short-form video. Postproduction referred to special effects or creative affordances applied in the short-form video. Postproduction can be placed in multiple content categories. All other codebook categories are mutually exclusive.
Coding Scheme of Platform Features.
Results
The 92 TikTok posts with #factcheck were from 60 unique users. Within the 60 unique TikTok creators, the average follower number was 3,015,59.50 (SD = 5,396,14.48). These creators had an average 6,586,487.58 profile likes count (SD = 16,722,947.86). The average footage was 76.72 s (SD = 47.34). These posts had 8,150,550 likes count (M = 88,592.93, SD = 492,265.01) and 99,898 comments count (M = 1085.85, SD = 2401.18) in total. For social engagement in #factcheck, the engagement score (likes + comments) was 8,250,448 (M = 89,678.78, SD = 494,321.10).
PolitiFact (@politifact) had 3.1 million profile likes and 156,400 followers in total. Within the 159 TikTok posts from PolitiFact, the average footage was 53.23 s (SD = 32.22). These posts had 3,100,381 likes count (M = 19,499.25, SD = 83,980.97) and 38,737 comments count (M = 243.63, SD = 758.38) in total. For the social engagement, the engagement score (likes + comments) was 3,139,118 (M = 19,742.88, SD = 84,571.35).
The results of the content analysis are presented in Table 4, which details the distribution of general themes, creator profiles, narrative types, and platform features.
Distribution and Comparison of Content Analysis.
Notes. ^Four videos used visual scenes without any audio. In this case, there were 155 instances for @politfact.
^^Postproductions could have multiple codes per video. In this case, there were 125 instances of platform features for #factcheck and 162 instances for @politfact.
^^^For the chi-square comparison, three videos were in both #factcheck and @politfact datasets resulting in N = 248 instead of N = 251.
*p < .05, ** p < .01, *** p < .001, two-tailed test.
#factcheck on TikTok
As shown in Table 4, political issues (32/92, 34.8%), lifestyle (28/92, 30.4%), and health issues (12/92, 13.0%) were the top three themes on #factcheck TikTok. In addition, almost 80% of Fact-check content creators preferred to use selfies for telling stories on TikTok (72/92, 78.3%) while the remaining 20% prefer not to be seen in their fact-check content (19/92, 20.7%).
A large majority of fact-check content creators prefer either the use of original audio (56/92, 60.8%) or to apply multiple soundtracks (30/92, 30.7%) in their videos. This study also recorded song names with hashtags in the “Use This Sound” in-app function to understand the occurrence frequency of specific audio. The result showed that Monkeys Spinning Monkeys by Kevin MacLeod and Kevin The Monkey was used in three different videos. This lilting melody had music without lyrics and became background music for multiple soundtracks. Our results suggest that no significant “sound” imitation content was unique. Accordingly, fact-checking content often requires a detailed explanation, leading to more complex audio being used.
Over 120 codebook references relate to postproduction features. The subtitle feature (75/125, 60.0%) on #factcheck was widely the most commonly used TikTok affordance. In addition, #factcheck content creators were either an influencer (56/92, 60.9%) or a fact-checking organization (27/92, 29.4%). Notably, these fact-checking organizations included VERIFY, owned by a U.S. media group, TEGNA, IFCN members, Snopes.com, Bolivia Verifica, PolitiFact, and MediaWise.
@politifact on TikTok
Political (77/159, 48.4%) and health (30/159, 18.9%) issues were the most prevalent themes for PolitiFact with a high percentage utilizing selfies (148/159, 93.1%) to construct their stories. Aside from four TikTok videos which did not use audio, PolitiFact preferred the use of multiple soundtracks (118/155, 76.13%) in much of their content. One in five TikTok posts contained original audio and a very small percentage (6/155, 3.9%) utilized music.
There were 162 codebook references to postproduction features in PolitiFact content. The subtitle (67/162, 41.4%), Stitch (35/162, 21.6%), and Video Reply to Comments (30/162, 18.5%) were the top three leading features. Results echoed #factcheck in that the socially creative feature “Use This Sound” (Kaye et al., 2022) did not demonstrate unique character in #@politifact videos. Overall, “political issues” themes, the “selfie” narrative types and the “subtitle” in postproduction predominated PolitiFact content on TikTok.
#factcheck versus @politfact
In comparison, both #factcheck and PolitiFact drew a majority of their content themes on political topics. As shown in Table 4, political issues were ranked the most frequent content theme for both datasets. However, #factcheck featured more content on lifestyle themes (+21% greater) than TikTok videos produced by PolitiFact. Multiple themes were more likely to be encountered in #factcheck than PolitiFact content (+6.5%). In contrast, narrative-type content ranked similarly among both datasets.
Both #factcheck and PolitiFact content creators favored applying the subtitle postproduction feature over other affordances with #factcheck creators applying it far more often than PolitiFact (+18.6%). However, stitch and video reply to comments features (−12%) were less observable in #factcheck content yet #factcheck creators applied the Duet feature slightly more often (+3.6%). Among content creators, influencers outnumber fact-checking organizations almost 2 to 1 in #factcheck (Table 4).
This study also determined that TikTok posts with #factcheck (89,678.78, SD = 494,321.10) had larger engagement compared to @politifact (19,742.88, SD = 84,571.35 (see Table 4). The average engagement of the #factcheck video was about 4.5 times the @politifact average engagement. Remarkably, there were only 14.5% of videos with the #factcheck hashtag within the TikTok posts from PolitiFact. Furthermore, there was no significant effect between TikTok posts with and without the #factcheck hashtag on engagement gain, t(247) = 1.37, p = .17.
Discussion
Prior research has observed that humanitarian and health videos are popular topics among fact-checkers (Sidorenko Bautista et al., 2021). In addition, lifestyle videos have become standard practice among TikTok creators to document their lives (Zeng, 2023). Accordingly, our study demonstrates that #factcheck content creators adapted their content strategies to TikTok audiences.
Important characteristics of fact-checking content
The fact-checking content of #factcheck and @politifact shared some common features, such as political theme, selfie narrative type, and subtitles in postproduction. Political fact-checking is a highly relevant concern regardless of the medium. Political issues, such as the Ukraine War and claims made by public figures, were two prevalent topics in both #factcheck and @Politifact content. Topics ranged from fact-checking a fake war scene during the Ukraine War to scrutinizing claims by political elites. Other politically related topics include abortion policy, the Flores settlement, the Health Insurance Portability and Accountability Act, the Declaration of North America, and gun control. Our findings align with current research noting that political topics are a consistent source of misinformation and are longstanding fact-checking targets (Bösch & Ricks, 2021; Graves & Amazeen, 2019; Graves & Cherubini, 2016; McNair, 2017; Nyhan et al., 2020).
In our study, fact-check content creators, either #factcheck or @Politifact, appear to require the content creator be shown in the video. Selfies can be understood as a form of online corporeal sociability that not only gestures toward authenticity but also functions as a mode of witnessing, offering a personal point of view within broader sociopolitical contexts (Senft & Baym, 2015). Influencers on social media platforms employ authenticity as a strategic tool to enhance relatability, reinforce their credibility as celebrities, and engage in the attention economy (Abidin, 2020; Jerslev, 2016; Jerslev & Mortensen, 2016). Stein et al. (2022) analyzed TikTok videos and revealed that creators’ relatability—through self-expression and self-disclosure—played a crucial role in communicating sexual health information. Furthermore, Lu and Shen's study on Douyin (2023) demonstrated that face presence is not an effective engagement strategy. Thus, selfies in fact-check TikTok content can be seen as improving the reliability and credibility of fact-checking claims.
Fact-checking content typically provides arguments and supporting evidence to help combat misinformation (Naeem and Bhatti, 2020). The debunking information ordinarily includes the claims that must be fact-tested, some verifiable questions, and supporting evidence to reach a verdict (Vlachos & Riedel, 2014). Therefore, it is difficult to explain the above information only in an oral statement. Hence, subtitles and picture-in-picture can deliver a large body of knowledge in a short time. However, subtitles can be a powerful tool for disseminating false information, enabling the manipulation of audio and verbatim text to severely misrepresent a political figure's original speech (Martínez-Carrillo & Tamul, 2019). In either case, fact-checking content often requires background information, which is why subtitles were widely used. They help clarify complex points and generally ensure that important details are accessible, aiding viewer comprehension and supporting content accuracy.
Role of TikTok affordances on fact-checking content
The #LearnonTiktok and #FYP appeared frequently in TikTok fact-check content. Therefore, hashtags do appear to play a crucial role in promoting views or educating users. The use of TikTok's FYP uniquely distinguishes the platform from other profile-based social media networks (Schellewald, 2023) and #LearnonTiktok suggests that fact-checkers intend to educate TikTok users. Yet, the relationship between #LearnonTiktok and #FYP and its application by fact-check content creators is an area of further study.
Applying relevant hashtags is an important aspect of creating fact-check content. It could be argued that TikTok operates as a hashtag public previously observed on Twitter (Bruns & Burgess, 2015). Yet, if hashtags are understood as filters, TikTok appears to employ a more complex and multi-layered filtering system to generate communities (Vizcaíno-Verdú & Abidin, 2022). Therefore, the platform's algorithms not only group users by hashtagged content but also by original audio and certain postproduction features. These algorithmic influences demonstrate that TikTok's fact-checking creators foster both imitation and hashtag publics.
Discovery and participation in the TikTok factcheck community
Stich, Duet, Video Reply to Comments and the use of audio on TikTok are considered socially creative features to help build community on the platform (Kaye et al., 2022). These socially creative features are both designed to increase content diversity and to create social networks. It echoes the concept of “imitation publics” (Zulli & Zulli, 2020), which indicates that memetic features on TikTok modify sociality and present an innovative form of a networked public. In our study, @PolitiFact applied socially creative features, such as Stitch and Video Reply to Comments to build community through participation. As a professional organization, it could be argued that PolitiFact is addressing problematic content on TikTok by recontextualizing it indirectly through the Stitch feature or directly by video replying to comments. By engaging in distributed creativity, Politifact becomes an important participant in creative chains active in the fact-check community. On the other hand, #factcheck content creators were more likely to use sound (original audio) and subtitles to be discovered in their relevant communities. It is plausible that #factcheck influencers and organizations use TikTok to find audiences hoping for the TikTok recommender algorithm to form credible content-based connections. As a result, one would expect these #factcheck content creators to be adept at understanding algorithms and features on TikTok to land on FYPs. Research indicates that fact-check content creators are very knowledgeable about platform infrastructure and dynamics (Dierickx et al., 2024; Roy et al., 2025). This could explain why #factcheck content had much higher levels of engagement than Politifact content. However, what is not clear from our findings is to the extent to which careful sourcing, contextual precision and procedural transparency are vital to these content creator fact-checking processes. As argued by Shin et al. (2025), fact-checking is not merely a technical task but a socially situated process involving judgment, ethics, and accountability. This remains one of the key epistemic challenges in the current media environment (Shibuya et al., 2025; Shin et al., 2025). In summary, it appears #factcheck content creators favor the discovery affordance over participation while Politifact favors the participation affordance over discovery (Kaye et al., 2022). In either case, a niche community of factcheckers is utilizing the short video platform but the material and institutional conditions for verification are arguably reconfigured.
Our study highlights how fact-checking content adapts to a platform's unique affordances and user culture. This aligns with prior studies (Grover, 2023; Lu & Shen, 2023; Sidorenko Bautista et al., 2021), indicating that fact-checking content is a dynamic, interactive, and diverse place on social media. Additionally, professional fact-checkers have adjusted their fact-checking presentation, moving away from formal report templates to better fit TikTok's platform. However, despite TikTok's innovations in fact-checking content, it remains limited in how platform-specific tools systematically support fact-checking. As discussed previously, the average engagement of user-generated #factcheck content was significantly higher than that of professional fact-checker PolitiFact. This finding echoes previous research on #CovidVaccine TikTok videos, which revealed that the majority of trending content was created by individuals who did not self-identify as professionals or experts (Lewis & Grantham, 2022). TikTok's socially creative features may boost grassroots engagement, and its unique technological affordances allow content creators to employ various communicative forms. However, the platform still lacks dedicated mechanisms to encourage widespread user participation in combating misinformation through fact-checking. Stein et al. (2022) pointed out that TikTok is a platform centered on interactivity, where its technological affordances enable content creators to employ diverse communicative forms, often blurring the boundaries between health education and entertainment. Although the results underscore the evolving nature of fact-checking on TikTok, both professional fact-checkers and public actors engage in ways that are shaped by the platform's design.
Conclusion
This study draws on the epistemic infrastructure framework (Shin et al., 2025) and identifies three interrelated domains (epistemic labor, institutional practices, and infrastructural systems) that collectively reshape fact-checking on TikTok. The infrastructural systems, comprising platform algorithms and affordances, exert a substantial influence by shifting verification from traditional editorial gatekeeping toward forms of infrastructural governance. Professional fact-checking institutions such as PolitiFact adapt their human-centered workflows to the platform by prioritizing participatory affordances and directly addressing false information or media literacy content through TikTok's socially creative features, particularly Stitch and video replies to comments. Meanwhile, #factcheck content creators engage in epistemic labor by leveraging TikTok's recommendation algorithm to optimize discovery affordances and form niche content communities characterized by higher engagement. As fact-checkers increasingly operate within a short-video ecosystem, TikTok's infrastructure reconfigures both the material and institutional conditions of verification.
There are still some arguments about the effect of fact-checking content in combating false information. Widely spreading fact-checked information is often unable to counter the viral presence of misinformation. The corrected inaccurate information still shapes politically biased beliefs (Thorson, 2016), and the evidence remains inconclusive regarding the effectiveness of fact verification in correcting political misconceptions (Van Aelst et al., 2017). It is plausible that the fact-checking information possibly lacked consistency within political issues (Lelo, 2022). The continued influence effect argues that even when fact-checking content is provided, misinformation frequently persists and sways inferential thinking (Ecker et al., 2020). Furthermore, the backfire effect argues that a correction serves to reinforce the very misunderstanding that it is meant to address (Jiang & Wilson, 2018). Nonetheless, some studies indicated the positive impact of fact-checking information on people's political knowledge (Fridkin et al., 2015), as well as the safety of repeating misinformation when correcting it (Ecker et al., 2020; Swire-Thompson et al., 2022). This study revealed that fact-checking by professional fact-checkers on TikTok tends to produce more consistent content compared to nonprofessionals. Regardless of where they are found, it is essential to have fact-checkers on TikTok to foster meaningful public discourse and address misinformation.
This study has limitations which restrict the generalizability of the findings. PolitiFact serves as an indicator of the global fact-checking movement and a leader in professional fact-checking practices (Lauer & Graves, 2025), though it has historically focused on debunking political claims (Graves, 2016). Although the results of this study did not reveal significant differences from the broader public fact-checking landscape, the growing expertise differentiation within the fact-checking community suggests directions for future research. Future studies could consider a wider range of professional fact-checking organizations with more diverse institutional backgrounds. Notably, relying on the hashtag #factcheck may not capture the full scope of fact-checking content on TikTok when sampling fact-checking videos. For instance, only 14.47% of PolitiFact's TikTok posts included the #factcheck hashtag. Although the two samples in this study are comparable in size and provide a general fact-checking content landscape on TikTok, future research could focus on specific time periods or event-driven fact-checking practices using targeted hashtag strategies. Such an approach would enable a more nuanced understanding of how creators engage with and adapt to platform algorithms. TikTok is a fast-evolving mobile platform, making the acquisition of TikTok data difficult and limiting the number of TikTok videos captured using third-party tools. This study is limited by a 2-year sampling period, and the dataset samples do not capture all TikTok content tagged with #factcheck. In addition, TikTok's algorithmic structure does not allow users to directly search for fact-checking content in a comprehensive way. Future studies should examine larger fact-check content collections on TikTok. It could also explore linguistic and geographic differences and how these factors shape fact-checking practices on the platform. In addition, TikTok created new affordances which may have a significant effect on fact-checking efforts. Since February 2022, TikTok has tested in-app features allowing the upload of 15-min and 10-min videos. Therefore, future work could compare the difference between short-form and longer-form fact-checking content on TikTok. In September 2023, TikTok launched a new tool to help creators label their AI-generated content and it also began testing ways to label AI-generated content automatically (TikTok, 2023). These developments indicate that the TikTok algorithm lacks transparency, and its in-app functions and affordances are constantly changing. Moreover, since the TikTok algorithm and in-app features may vary across countries, future research could use the TikTok research API to access TikTok data and examine regional differences.
Overall, this study provides robust evidence for a shift in fact-checking on TikTok from traditional human-centered gatekeeping to infrastructural governance, aligning with the epistemic infrastructure framework (Shin et al., 2025). Practically, it applies content analysis to complement and extend prior qualitative-oriented fact-checking research, while unpacking how TikTok's affordances shape community formation and generate networked publics (boyd, 2010) or even imitation publics (Zulli & Zulli, 2020) through its algorithm.
Footnotes
Acknowledgment
The authors are sincerely grateful to the journal editors and the two anonymous reviewers for their valuable suggestions. This research is based on the first author's thesis. She wishes to extend her heartfelt thanks to the thesis committee: Dr. Adrian Rauchfleisch (Chair), Dr. Ji-Lung Hsieh, and Dr. Shih-Hsien Hsu.
Ethical considerations
This study utilized a web-scraping service to acquire TikTok post content with the #factcheck hashtag. The study also captured TikTok posts uploaded by PolitiFact, a fact-checking organization. No human subjects are involved and participant consent is not needed for the content analysis in this study.
Funding
The authors received no financial support for the research, authorship, and/or publication of this article.
Declaration of conflicting interests
The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.