Understanding screenshot collection and sharing on messaging platforms: a privacy perspective

Abstract

Individuals rely on messaging platforms to form and maintain intimate relationships, trusting shared information will remain within intended digital confines. However, the screenshot feature allows people to capture and store pieces of private conversations as a separate file on their device, rendering them shareable with third parties. While usage of this feature can be benign, this study focuses on its ability to breach privacy expectations within messaging platforms, termed within communication privacy management theory as privacy turbulence. This study recognizes the power of both interpersonal dynamics and platform affordances in constraining existing norms around screenshot collection and sharing others’ private messages. Experimental results (n = 302) suggest obscuring received messages upon use of the screenshot feature and stating an explicit privacy rule significantly reduce screenshot collection and sharing, respectively. Implications for communication theory and the future of messaging platform design will be discussed.

Individuals rely on messaging platforms to form and maintain intimate relationships, trusting shared information will remain within intended digital confines. However, these conceivably controlled environments are not impermeable due to the screenshot feature, which is integrated into most smartphones (Coluccini, 2020). The screenshot feature enables users to capture and store pieces of private conversations as a separate file on their device, making them readily accessible and shareable with third parties. While the screenshot feature serves utilitarian purposes (Ali et al., 2022; Mottelson, 2023), its ability to capture and share intimate messages has created a frictionless tool for blackmail or public shaming (Corry, 2021) and has invoked harms that have landed in courtrooms (Ingber, 2024). Building on Shore and Prena (2023), this study leverages communication privacy management (CPM) theory to understand individuals’ usage of the screenshot feature to breach others’ expectations of privacy on messaging platforms, which Petronio (2002) would define as privacy turbulence. Privacy turbulence occurs when control is lost over shared information (Petronio, 2002). The screenshot feature enables privacy turbulence (Shore & Prena, 2023), as it authorizes the permanent storage and dispersion of private conversations without notice or consent from implicated individuals.

Given its threat to digital communications, empirical attention to the screenshot feature is essential. Although Shore and Prena (2023) introduce the screenshot feature as an enabler of privacy turbulence, their study is limited to sharing screenshots of highly intimate messages received from a hypothetical “close friend” and does not consider other information types or gauge real-time behaviors. The current study addresses these limitations by providing a more generalizable and ecologically valid framework and method, respectively, contributing to the emerging literature on screenshot collection and sharing from a privacy perspective.

This study also builds on Shore and Prena’s (2023) extension of CPM by highlighting the platform's role in how people manage others’ information (i.e., platform privacy rules). In addition to recognizing the power of interpersonal dynamics (Petronio, 2002), this article proposes affordance-based privacy rules as a type of platform privacy rule that has the potential to constrain screenshot collection and sharing of digital messages. Results of this experiment provide evidence of the normality surrounding screenshot collection and sharing of digital messages—in addition to qualitative rationale for why people do this—and reveal how these behaviors can be influenced by interpersonal dynamics, affordance-based privacy rules, and information context. Implications for communication theory and the design of messaging platforms will be discussed.

Defining screenshot collection and sharing as distinct turbulent behaviors

Use of the screenshot feature has become incredibly common among young adults for purposes of both collecting and sharing information (Cramer et al., 2022; Mottelson, 2023). Screenshot collection and screenshot sharing evoke independent privacy harms, warranting their conceptual separation (Ingber, 2024). Cramer et al. (2022) define screenshot collection as the capturing of content from a device screen to store as a new image file on the same device. Harms that arise from screenshot collection stem from the threat of usage and sharing. For example, collected screenshots of intimate messages could be leveraged as blackmail (Corry, 2021). Notably, screenshot collection of digital messages may be benign. Mottelson (2023) highlights many people collect screenshots when they just want to remember something. Nevertheless, screenshot collection can create privacy turbulence through enabling record persistence in a space unknown to others contributing to the conversation. The threat of screenshot collection without proper accountability may discourage disclosure on messaging platforms (Shore & Prena, 2023).

Screenshot sharing is a separate action following—and often dependent on—screenshot collection, where the collected screenshot is disseminated. Of note, screenshot sharing is not always dependent on collection; for example, when individuals share screenshots received from others, no initial collection is required. Cobb and Kohno (2017) provide evidence that nearly half of their respondents periodically took screenshots of profiles on dating applications to share them with friends. Corry (2021) discusses instances where screenshots were shared with others as a means of public shaming. Similar to screenshot collection, screenshot sharing does not always have harmful intent. For example, youth commonly send screenshots of messages to their friends when they need advice (Ali et al., 2022).

The state of screenshot collection and sharing research

Despite integration of the screenshot feature into the iPhone’s software in 2007—with Android following in 2011 (Coluccini, 2020)—it has gained little attention from researchers beyond a discussion around why people take and use screenshots. Among these findings, researchers have demonstrated screenshots are often leveraged as socialization tools (Cramer et al., 2019), helping to develop trust and negotiate hierarchies of power (Jaynes, 2020). Individuals have also reported using the screenshot feature for bookmarking digital information, containing and disclosing private interactions, or reframing content for social media (Cramer et al., 2022; Mottelson, 2023). While archival may have been the intent of integrating the screenshot feature into mobile devices, individuals often use technological features beyond their original purpose (Scharlach & Hallinan, 2023). Use of the screenshot feature to infringe upon expected informational boundaries within messaging platforms has received minimal attention from communication scholars (Ingber, 2024; Shore & Prena, 2023).

That being said, the harms of the screenshot feature have not gone unnoticed by industry professionals. Snapchat has been most notably acknowledged for implementing design cues which hold co-owners accountable for screenshot collection (Bayer et al., 2015; Kofoed & Larsen, 2016). A unique response to the screenshot feature has been implemented by Signal, which allows users to block people who have physical access to their phone from taking screenshots of their messages (Signal, 2021). However, this does not stop people from taking screenshots of received Signal messages on their own device.

Sexting researchers have suggested anonymizing disseminated screenshots unless there is a strong reason not to (e.g., evidence of harassment) (Stoicescu & Rughiniş, 2021). Geeng et al. (2020) contend screenshot notifications and blurring faces in disclosed imagery would aid in keeping intimate messages secure. Although this research focuses on sexting, it grounds an argument inclusive of all digital communications. In fact, Hasan et al. (2023) recognize the privacy harm invoked by the screenshot feature by including its usage in their scale measuring perceptions about others’ privacy.

Communication privacy management theory

CPM provides a useful framework to understand screenshot collection and sharing of digital messages (Petronio, 2002). This theory has been used to understand privacy management across communication contexts including e-commerce (Metzger, 2007), social networking sites (Waters & Ackerman, 2011; Zhang & Fu, 2020), and online messaging (Venetis et al., 2012). Petronio (2002) conceptualizes privacy boundaries as metaphorical lines to distinguish between public and private information. CPM asserts people form these boundaries as a result of their recognition of ownership and desire for control over private information. Receivers of shared information are designated as information “co-owners.”

Once a collective boundary is formed with a co-owner, a rule-based management system governs the co-responsibility of shared information. The co-owner is responsible for following privacy rules regarding who else can know the information (boundary linkage rules) and how much of it can be shared (boundary permeability rules). A breakdown in this privacy management effort—or disruption to established boundaries, rules, and dynamics of information ownership and control—defines privacy turbulence (Petronio, 2002). CPM frames the breaking of established privacy rules as a natural component of the privacy management life cycle (Petronio, 2002), after which information owners recalibrate their privacy management strategy (Steuber & McLaren, 2015).

Engagement in screenshot collection and sharing demonstrates a false sense of ownership over co-owned information. Petronio (2002) refers to this as the “pretend factor,” where co-owners feel they have a right to share information that is not theirs. Although the pretend factor was developed in the context of interpersonal gossip (i.e., quoting a previous conversation), the validity attributed to screenshots of messages gives these records more weight than information recall (Yadav et al., 2023). The pretend factor is amplified when a secondary tool (e.g., the screenshot feature; audio recorder) surreptitiously captures records without permission, thereby creating a new record under the newfound ownership of the co-owner. When people sense a higher level of ownership over co-owned information, they are more likely to share it in the future (McNealy & Mullis, 2019). Use of the screenshot feature to capture and share records maintaining high perceived validity (Yadav et al., 2023) raises new questions about the consequences of misperceived data ownership.

Explicit privacy rules

Petronio (2002) defends that privacy turbulence may occur if rules are misinterpreted by the co-owner, creating “fuzzy” boundaries (Petronio, 2002). She argues when rules are explicit, co-owners are less likely to create privacy turbulence. This may include specifically telling a conversation partner not to spread information beyond the confines of the communication platform. Venetis et al. (2012) found more than half of their participants who had previously experienced privacy turbulence failed to provide an explicit privacy rule. As such, co-owners may be less likely to use the screenshot feature to collect and share a digital message when an explicit privacy rule is conveyed by the information owner.

H1. The presence of an explicit privacy rule will discourage co-owners from a) collecting and b) sharing screenshots.

Platform privacy rules: the role of affordances in privacy management

Of note, Venetis et al. (2012) argue explicitly stating privacy rules may come off as unnatural or offensive to conversation partners. Individuals should not bear the burden of managing their own privacy in an ever-changing, complicated digital world. Long gone are the days where privacy turbulence can be discussed without integrating the impact of platforms. Platforms host networks of individuals and businesses in their quest to communicate, organize, and interact (Gillespie, 2010). Their power “interfere[s] with social and democratic functions in society” (Van Dijck et al., 2019). Through their behind-the-scenes business practices and design (e.g., affordances, stagnant features), platforms create non-negotiable privacy rules that override individual desires for privacy. For instance, by instituting preventative features (Bhandari et al., 2021) and guidelines (Chandrasekharan et al., 2017), platforms can constrain harmful behaviors. Shore and Prena (2023) argue platforms have the power to foster rules that prioritize user privacy, proposing “platform rules” as an extension of CPM.

One way platform rules can be deployed is through affordances. This article will refer to this type of platform rule as affordance-based privacy rules. Technological affordances have been defined as perceived “possibilities for action” (p. 36) situated between the user and technological interface (Evans et al., 2017) that can guide social behaviors (Fox & McEwan, 2017). Communication scholars have supported the role of affordances on privacy management, thereby recognizing the presence of affordance-based privacy rules (Trepte et al., 2020). Specific to messaging, Coduto (2024) found perceived affordances such as editability and social presence contributed to sexting patterns.

Affordance-based privacy rules arise as a result of affordances which promote possibilities for information management on a platform (Fox & McEwan, 2017) or in relation to a specific element of the platform (Bayer et al., 2020). Critically, these rules are not techno-deterministic in that their presence does not determine a specific outcome. Rather, echoing Evans et al. (2017), the impact of affordance-based privacy rules varies. Nevertheless, this study contends affordance-based privacy rules can discourage the privacy turbulence provoked by the screenshot feature.

Identifiable accountability for screenshot collection

Popular messaging platforms (e.g., iMessage, WhatsApp) allow for screenshot collection and sharing to occur undetected, meaning the behavior is unknown to the information owner. Further, if screenshots of digital messages were to be dispersed, there would be no way to identify who collected the original screenshot without further context. Although messaging platforms generally do not afford anonymity, as message receivers are typically aware of those who they are communicating with (Broeker, 2023), anonymity is afforded as they interact with the screenshot feature. Anonymity has been defined as a key affordance of communication channels (Evans et al., 2017; Fox & McEwan, 2017). When people feel their behavior is anonymous and not likely to provoke identifiable accountability, they become particularly disinhibited (Suler, 2004) and may engage in vitriolic behavior (Theocharis et al., 2016). Screenshot collection and sharing on most messaging platforms leave no traces behind, eliminating any risk in violating others’ privacy. This is distinct from Snapchat’s approach which makes screenshot collection of shared images identifiable. It is possible that the absence of anonymity—or personally identifiable accountability for actions—may lessen co-owners’ willingness to collect and share screenshots. This article conceptualizes accountability as the inverse of anonymity. Consistent with Evans et al.’s (2017) conceptualization of affordances, accountability can be established through various mechanisms and lead to different outcomes. This may include strategies similar to Snapchat’s screenshot notification or stricter measures that more publicly punish users for use of the screenshot feature.

Individuals are less likely to share the personal data of others when they perceive it would be identifiable to the information owner (Pu & Grossklags, 2017). Shore and Prena (2023) observed young adults were less willing to take screenshots on platforms that alerted their conversation partner of their behavior (e.g., Snapchat). Heightening perceptions of accountability for screenshot collection may not only lower perceptions of ownership over co-owned information shared within messaging platforms, but also yield positive effects for information owners. In fact, research has demonstrated accountability for screenshot collection makes people feel more in control of their information shared on messaging platforms (Shore & Prena, 2023; Velten et al., 2017). The following is hypothesized:

H2. Accountability will discourage co-owners from a) collecting and b) sharing screenshots.

Addressing persistence through embedding obscurity in design

While some platforms have started to promote accountability by providing a notification to users when their content has been subject to the screenshot feature (Kofoed & Larsen, 2016), this does not eliminate the possibility of privacy turbulence as individuals would still have the record to permanently store and share. In their discussion of platform power, Nieborg and Helmond (2019) note that even if users opt out of platforms, this does not reverse or delete any previously shared data. This exemplifies an interaction between privacy turbulence and data persistence.

Persistence is an affordance of messaging platforms that is amplified in light of the screenshot feature. Treem and Leonardi (2013) define persistence as accessibility to information in the same form as the original display after it had been revealed. Messaging platforms allow the screenshot feature to extend the lifetime of already persistent conversations by creating and storing a replicated version of the record. The screenshot feature exacerbates persistence on messaging platforms by capturing an out-of-context snippet of a conversation for storage in one’s camera roll and beyond. Due to the screenshot feature, users may not fully trust the privacy supposed to be afforded by ephemeral messaging (Waddell, 2016).

One way to rid screenshots of their persistence is to embed obscurity in design. Obscurity directly targets persistence by constraining messages received by the co-owner from persisting either (a) within their private camera roll or (b) through further dispersion. Hartzog (2018) promotes obscurity to enhance privacy by allowing less transparent access to specific, sensitive information. Hartzog and Stutzman (2013) define obscurity as “a context in which information is relatively difficult to find or understand” (p. 388). Of note, obscurity of information does not imply anonymity. One may be able to deduce who information belongs to while not knowing the specifics of what such information is. For example, Shore and Cummings (2022) discuss location data obscurity which would allow people to have a general indication of where an identifiable user in their network is located without exposure to their exact location. Results from Shore and Cummings (2022) indicate a desire for personal obscurity is influenced by other embedded technology features. However, no research to date supports obscurity as hindering co-owners from engaging in privacy turbulence. Evidence of this would support Hartzog’s (2018) proposal for embedding obscurity in design to enhance individual privacy. Adhering to Evans et al.’s (2017) definition of affordances, obscurity can have variability in its implementation. For example, data collected via screenshot may be completely obscured by producing a black screen in response to screenshot collection (e.g., Netflix) or temporarily obscured through limiting access to private content using password protection (e.g., Snapchat’s “My Eyes Only”). Within the context of this study, obscurity will diminish the persistence of digital messages by blurring content sent by the information owner upon screenshot collection. If such content was obscured, screenshots of digital messages would lose their persistence, creating a barrier to privacy turbulence.

H3. Obscurity will discourage co-owners from a) collecting and b) sharing screenshots.

This study takes interest in an interaction effect between explicit and affordance-based privacy rules. According to Trepte’s (2021) social media privacy model, users’ perceived affordances dynamically interact with their initial assessment of the information owner’s ideal level of access. Although researchers have considered both interpersonal factors and platform affordances in their study of privacy management (Vitak & Kim, 2014), an interaction has not been quantitatively demonstrated.

RQ1. How does an explicit privacy rule interact with affordance-based privacy rules (i.e., accountability and obscurity) on screenshot a) collection and b) sharing?

This study will measure several control variables including information sensitivity (Wirth et al., 2019) and perceptions of control and ownership over the information received in text messages (Petronio, 2002). Additionally, Petronio (2002) mentions privacy turbulence may be created depending on the information context (i.e., motivational criteria), which echoes the unique privacy management decisions made about specific pieces of information (Bute et al., 2015; Nissenbaum, 2009). Finally, this study controls the norms and habits of collecting and sharing screenshots of text conversations, as supported by Trepte (2021). The theoretical framework underlying this study can be found in Figure 1.

Methodology

Participants

Participants from the USA were recruited from Prolific, which has been deemed to produce higher data quality than alternative recruitment platforms (Douglas et al., 2023). A power analysis was conducted a priori using G*Power with a desired effect size of .25 (Cohen, 2013), power of .99, error of probability <.05, and 9 predictor variables. The power analysis indicated this experiment would require approximately 260 participants. Given anticipated unusable participant data, 318 participants were recruited to ensure the minimum participant sample. Participants were removed from the final sample if they failed the manipulation check, as described in the next section (n = 302). All participants were between 18 and 30 years old, as previous literature supports the use of the screenshot feature within this age demographic (Shore & Prena, 2023) (M = 24.57, SD = 3.36). Participants identified as male (n = 144), female (n = 139), non-binary (n = 17), or preferred not to say (n = 2). In terms of race, participants reported to be Asian or Pacific Islander (n = 47), Caucasian (n = 165), Hispanic (n = 31), Black/African American (n = 27), Middle Eastern (n = 2), mixed race (n = 29) or preferred not to say (n = 1). Participants indicated their highest level of completed education was high school/GED (n = 159), Bachelor’s degree (n = 117), Master’s degree (n = 17) or Advanced Graduate Degree (n = 9).

Procedure

This experiment employed a 2(Explicit privacy rule: present or absent) × 2(Accountability: present or absent) × 2(Obscurity: present or absent) between-subjects design. Participants were first asked about the extent to which they collect screenshots of text conversations and what messaging platform they most dominantly use to communicate with friends and family. They were immediately informed that because of their responses to these questions, they qualified for participation in the study. While these appeared to be screener questions, the purpose of these preliminary questions was to contribute to a convincing deception.

Participants were deceived to believe the researchers were developing a new mobile plugin for their most frequently used messaging service (custom text was piped in from their previous response). The plugin was said to help people organize screenshots of the messages in their camera roll based on their context. Adapting Chang et al.’s (2016) design, participants believed the purpose of this study was to help the researchers train an algorithm for the plugin.

Once participants understood the purpose, they were presented with four screenshots to code for context and information sensitivity. Informed by Shore and Prena (2023), participants categorized the screenshots as “funny,” “drama,” “information,” or “other” (accompanied by a space to write-in a new category). Half of the screenshots contained an explicit privacy rule (e.g., “Don’t tell anyone”). A manipulation check ensured participants recognized the presence of an explicit privacy rule by asking, “In the text above, did the sender (GREY bubbles) request that the receiver (BLUE bubbles) keep the information private?”

After coding the screenshots, the purpose for which was solely to enhance the algorithm training deception, participants began the second part of the study. Participants were told “in order to enhance our user experience,” we wanted to know about their “own screenshot collection and sharing intentions of particular types of screenshots.” In this part of the study, they were informed of an additional feature of the plugin (accountability, obscurity, both, or none). Following a textual and visual description of the additional plugin feature (see Manipulations section), all participants viewed identical instructions:

When reviewing the following text message, imagine that you are playing the role of the BLUE bubbles. You will be making decisions about collecting and sharing information sent by the other user (GREY bubbles).

Note that we are not interested to know if you would collect and share these specific text messages; rather, we raise these texts as examples of TYPES of information that could be useful to screenshot and share.

Participants’ subsequent responses were based on one randomly selected screenshot from the previously coded set. These screenshots were pre-categorized by the researcher as representing content that is funny, gossip, infidelity, or a list of items (see Figures 2–5). Under the assumption that the plugin had been installed on their phone, participants then used their randomly selected screenshot to answer the screenshot collection and sharing questions. The final part of the experiment included control variables (see Measures section) and open-ended questions to contextualize rationale decisions. At the end of the study, participants were debriefed on the true purpose of the experiment (see Debrief Message in the Supplementary materials).

Manipulations

Explicit privacy rule was manipulated within the presentation of the text message stimuli. When a privacy rule was explicit, the information owner indicated the co-owner should not share this information with anyone. When the explicit privacy rule was absent, this part of the conversation was not depicted (see Figure 6).

Accountability was manipulated to reflect screenshot collection either being known or not known by the information owner. When accountability was present, participants were told: “A feature of this new plugin includes informing conversation partners that you took a screenshot of the conversation (just like Snapchat).” Participants were then shown a visual of the notification that the information owner would receive if a screenshot was taken (see Figure 6).

Obscurity was manipulated in a way that rid the co-owner of the persistence enabled by the screenshot feature. Participants were told, “A feature of this new plugin is blurring of the sender's (GREY bubbles) texts upon screenshot collection so that their content is not readable when saved in your camera roll.” Participants were then given a visual of this feature, showing the difference between the initial conversation and what the messages would look like if a screenshot was collected (see Figure 6).

Figure 6 provides an example of the stimuli when an explicit privacy rule, accountability and obscurity were present. For other versions of the stimuli, see the Supplementary materials.

Measures

The following variables were measured on a 7-point Likert scale ranging from Strongly disagree (1) to Strongly agree (7). A full list of all items used in this study can be found in the Supplementary materials.

Screenshot collection was measured “assuming that our screenshot plugin had been installed on your phone [with the accountability/blurring feature].” Participants were then asked to rate their level of agreement with three statements including “I would take a screenshot of this kind of conversation,” “I would absolutely not collect the contents of this kind of conversation using a screenshot,” and “It would be useful for me to take a screenshot of a conversation like this” (M = 3.56, SD = 1.65, a = .79).

Screenshot sharing was measured under the assumption that the plugin had been installed on the participants’ phones. Participants in the obscurity condition were instructed to remember “the captured screenshot would be blurred” to ensure manipulation salience. Participants were then prompted with the language “I would feel okay sharing a screenshot of this kind of conversation with…” followed by actor-specific items (i.e., a close friend, a family member, a romantic partner, an acquaintance, a stranger, I would not share a screenshot of this conversation with anyone). The averages across items were combined to create a comprehensive screenshot sharing score (M = 3.82, SD = 1.53, a = .88).

Perceived ownership was measured using three items adopted from Peck et al. (2013). Participants were prompted with the statement, “When I receive a text message from someone…”. Participants rated their level of agreement with the following statements: “I feel personal ownership of the information sent to me,” “I feel like information sent to me is mine,” and “I cannot claim ownership over someone else's information, even if it is shared with me” (M = 3.54, SD = 1.41, a = .83).

Perceived control was adapted from Krasnova et al. (2010). Items included “I feel in control of the information provided to me through text messages,” “Privacy settings allow me to have full control over the information provided to me through text messages,” and “I feel in control of who can view the information communicated within my text message.” Only the last two items were used in data analysis given the lack of reliability of the first item with the others (M = 4.09, SD = 1.38, a = .71).

Information sensitivity was adapted from Wirth et al. (2019). Participants ranked their level of agreement surrounding consequences based on the screenshot that they were given to reflect on such as if “…this screenshot was made publicly available” (M = 4.26, SD = 1.77, a = .84).

Screenshot collection habits were measured using seven items adapted from Verplanken and Orbell (2003). Participants were prompted with “Taking screenshots of text messages is something…” before responding to the items. An example item included, “… I do frequently.” (M = 2.77, SD = 1.35, a = .91).

Screenshot norms were operationalized through an adapted scale (Merhi & Ahluwalia, 2019). Items included, “It is common to find others collecting screenshots of text messages”; “Most people I know generally collect screenshots of text messages,” “It is likely that most people collect screenshots of text messages,” “I believe others collect screenshots of text messages” (M = 4.97, SD = 1.29, a = .87).

Results

To prevent selection bias, experimental conditions were compared across important demographic information using one-way analysis of variance (ANOVAs). Results of these analyses suggest no differences across the conditions on age (F(1, 300) = 1.22, p = .27), gender (F(1, 300) = 0.26, p = .61), education (F(1, 300) = 0.10, p = .75), and screenshot habits (F(1, 300) = 0.00, p = .99).

Hypothesis testing

Given the dependence that screenshot sharing can have on screenshot collection, a Pearson’s correlation test was conducted to determine if these variables could be evaluated in the same model. The positive correlation between screenshot collection and sharing was significant r (300) = .22, p < .001, but not strong enough to warrant exclusion of screenshot collection as a covariate in the screenshot sharing model. Thus, to analyze these hypotheses, a multiple regression was run. All assumptions for multiple regression were tested and met for both the screenshot collection and sharing models.

To decide the extent to which interaction effects would be included in the model, Akaike Information Criteria (AIC) was used to compare model fit across those which had no interaction effects, all possible interaction effects, and significant interaction effects. No interaction effects were included in the screenshot collection model (AIC = 1117.95) and only significant interaction effects were included in the sharing model (AIC = 1068.53) as these were the lowest AICs as compared to the other models.

With the exception of screenshot collection in the screenshot sharing model, both final linear regression models included the same covariates: information sensitivity, screenshot category, screenshot habits, information ownership, information control, and perceived screenshot norms (see Table 1). A moderate proportion of the variance in screenshot collection (adjusted R-squared = 0.167) and sharing (adjusted R-squared = 0.213) was explained by the predictor variables.

Screenshot collection

H1a predicted an explicit privacy rule would discourage co-owners from collecting screenshots. This hypothesis was not supported (B = −0.01, SE = 0.18, p = .95, partial η² < .001). Explicit privacy rules did not discourage information co-owners from collecting screenshots.

H2a predicted the presence of accountability would discourage screenshot collection. This hypothesis was not supported. Accountability was a positive predictor of screenshot collection, opposing H2a (B = 0.37, SE = 0.18, p = .04, partial η² = .03).

H3a predicted the presence of obscurity would discourage screenshot collection. This hypothesis was supported (B = −0.60, SE = 0.17, p < .001, partial η² = .04). Obscuring received text messages discouraged co-owners from collecting screenshots of digital messages.

Within this model, more frequent habits (B = 0.27, SE = 0.07, p < .001, partial η² = .09) and higher normative perceptions of screenshot collection (B = 0.15, SE = 0.07, p = .04, partial η² = .01) were positive, significant predictors of screenshot collection.

Screenshot sharing

H1b predicted an explicit privacy rule would discourage screenshot sharing. This hypothesis was supported, (B = −0.33, SE = 0.16, p = .04, partial η² = .02). When the explicit privacy rule was present, co-owners were discouraged from sharing screenshots of digital messages.

H2b predicted accountability would discourage screenshot sharing. This hypothesis was not supported, (B = 0.18, SE = 0.22, p = .42, partial η² < .001). Accountability did not discourage participants from sharing screenshots of private digital messages.

H3b predicted obscurity would discourage screenshot sharing. Results indicated obscurity as positive predictor of screenshot sharing, opposing the hypothesis (B = 0.89, SE = 0.23, p ≤ .001, partial η² = .02). Obscurity did not discourage screenshot sharing when presented on its own. However, when accountability and obscurity were both present, participants were less likely to share screenshots (B = −0.67, SE = 0.32, p = 0.04, partial η² = .01).

Similar to screenshot collection, several control variables contributed significant variance to screenshot sharing. Perceived norms were a positive predictor of screenshot sharing (B = 0.14, SE = 0.07, p = .02, partial η² = .03). Additionally, information ownership was a significant, positive predictor of screenshot sharing (B = 0.21, SE = 0.06, p < .001, partial η² = .05). While not emerging as a significant predictor of screenshot collection, screenshot category had a significant impact on screenshot sharing. As compared to the “funny” screenshot, participants were significantly less likely to share the “gossip” (B = −0.75, SE = 0.29, p = .009, partial η² = .04) or “infidelity” (B = −0.95, SE = 0.31, p = .002, partial η² = .04) screenshots. Finally, this model demonstrated a positive relationship between screenshot collection and screenshot sharing (B = 0.17, SE = 0.05, p = .04, partial η² = .03).

RQ1 concerned the interaction between explicit and affordance-based privacy rules. There was no interaction effect between the explicit privacy rule and accountability (B = −0.47, SE = 0.50, p = .35, partial η² < .001) or obscurity (B = 0.04, SE = 0.50, p = .93, partial η² < .001) on screenshot collection. The same results were revealed regarding the interaction effect between the explicit privacy rule and accountability (B = −0.41, SE = 0.45, p = .37, partial η² < .001) or obscurity (B = 0.09, SE = 0.45, p = .85, partial η² < .001) on screenshot sharing. The only significant interaction effect was between the two platform rules, accountability and obscurity, on screenshot sharing (B = −0.65, SE = 0.33, p = 0.04, partial η² = .01).

Qualitative rationale for screenshot collection and sharing

Quotes from participants detailing their rationale for screenshot collection and sharing provided additional insight into the quantitative findings. All quotes were iteratively coded based on theoretical relevance and issues that arose in the texts (Attride-Stirling, 2001). Supporting H1a, participants detailed the efficacy of the explicit privacy rule on preventing screenshot sharing. For example, one participant noted, “The sender said not to share it with anyone, so I would not share it with anyone at all.” Similarly, another participant voiced, “the information was for ME only to remember and not share due to the messenger asking me not to.”

Several participants remarked on the influence of the contextual category on their decision-making. For instance, a participant who was reflecting on the “funny” screenshot noted the following, “It's harmless. Everyone has moments like that, and it’s just for fun.” This rationale differed when reflecting on the “infidelity” screenshot. One participant stated, “Sometimes, if the other person is in the wrong, we should share information if we feel it's justified to share. It's just our matters concerning our interpersonal relationships with other people.” This illustrates a sense of empowerment felt by some individuals when collecting screenshots of digital messages.

Qualitative responses also provided insight into obscurity’s influence, both on screenshot collection and sharing. One participant touched on what was a common theme regarding screenshot collection: “I think that it would be pointless and strange to collect a screenshot of this text thread since the useful information (what to bring to the party) is blurred out.” In terms of screenshot sharing, however, this sentiment was different. One participant noted, “With the blurring feature, there is no risk to sharing it, so I would be willing to with just about everyone.

Discussion

The purpose of this experiment was to understand the role of explicit and affordance-based privacy rules on screenshot collection and sharing of digital messages, applying a modernized version of CPM which recognizes the power platforms yield over privacy turbulence. Results not only suggest the normalization of these behaviors, but also that obscurity and an explicit privacy rule can prevent screenshot collection and sharing of digital messages, respectively. This experiment provides implications for future research on screenshots and privacy management of others’ information more broadly while also provoking insights for the design of messaging platforms.

Expanding beyond (but still including) interpersonal privacy rules

Participants generally followed the explicit rule set by their conversation partner when it instructed them not to share the information beyond the confines of the conversation. These findings align with Venetis et al. (2012), who found individuals were more likely to experience privacy turbulence when an explicit privacy rule was not clearly asserted.

Although this result suggests individuals can autonomously prevent screenshot sharing through voicing explicit rules, they should not have to carry this burden. Rather, messaging platforms could embrace explicit privacy rules to discourage users from screenshot collection and sharing within design. For example, platforms can communicate an explicit privacy rule through creating a “Confidentiality Mode,” which Ingber (2024) proposes as being able to nudge users against collecting screenshots of private communications ex-ante. Through this design enhancement, users would be reminded by the platform—as opposed to their conversation partner—that certain types of communication are meant to be private. Moreover, it would clarify any “fuzzy” boundaries that spur privacy turbulence (McNealy & Mullis, 2019) and serve as a reminder against the pretend factor. “Confidentiality Mode” would be particularly useful in group messaging, where it is more difficult to ensure privacy rules are understood by the entire group (Mansour & Francke, 2021). Explicit cues indicating antinormative behavior have proven successful when used to combat harassment (Bhandari et al., 2021).

Another way messaging platforms can set expectations for user behaviors is through community guidelines. Chandrasekharan et al. (2017) found Reddit successfully diminished hate speech from subreddits ridden with hate through instating an anti-harassment policy. Reddit’s ban had direct consequences, deleting existing hateful subreddits and those that tried to spring up afterwards (Chandrasekharan et al., 2017). Of note, updating community guidelines is not always effective. For example, while Facebook removed some antivaccine content through its misinformation policy, Broniatowski et al. (2023) found this did not decrease overall engagement with the remaining antivaccine content.

Messaging platforms should remain in tune with the changing values and norms of its users. Recent research found TikTok updated its community guidelines to reflect the public’s outcry for privacy and safety protections (Chan et al., 2023). However, in implementing platform restrictions, decision-makers must avoid unnecessarily limiting benign behavior which may disproportionately chill marginalized individuals (Haimson et al., 2021; Roth, 2015). Governing screenshot collection and sharing of private communications is particularly important for protecting marginalized groups, as these individuals already sense a heightened risk of identity disclosure online (Duguay, 2016). When people feel heightened privacy concerns, they are less likely to be intimate and honest online (Zhang & Fu, 2020).

The weakness of an accountability cue

The presence of accountability was not independently predictive of screenshot collection or sharing of digital messages. These results were surprising given the suggested effectiveness of an identical cue in preventing screenshot collection on Snapchat (Shore & Prena, 2023). However, one study found despite Snapchat’s screenshot accountability cue, college students still felt it was socially acceptable to collect screenshots on the application (Bayer et al., 2015). This competing evidence around the efficacy of Snapchat’s screenshot accountability cue raises questions about the strength of ephemeral communication. Research has shown the appeal of platforms such as Snapchat given its ephemerality, which diminishes self-consciousness and encourages playfulness within social interactions (Xu et al., 2016). The screenshot feature challenges ephemeral communication through making content persistent. In line with Bayer et al. (2015), this study showcases that accountability for violating ephemerality norms is not sufficient. Future research should continue to explore strategies to encourage ephemeral digital channels akin to face-to-face communication which allow individuals to discuss intimate topics at zero cost (Hollan & Stornetta, 1992).

Context-dependent privacy turbulence

While accountability did not significantly impact screenshot sharing, the presence of an accountability cue predicted higher levels of screenshot collection. Qualitative rationale illustrates the reasoning behind this finding may be context-dependent. In the case of the “infidelity” message, some participants believed a third party had a right to know about the contents of their conversation; thus, worries about being held accountable for collecting the screenshot were dismissed. Rationale for collecting the “funny” screenshot was distinct from this, with participants expressing screenshot collection would be harmless. Context-dependence was quantitatively demonstrated within the screenshot sharing model. Compared to the “funny” screenshot, participants were significantly less likely to share screenshots resembling those concerning infidelity or gossip.

These findings exemplify the consideration of contextual criteria when deciding whether to reveal private information (Petronio, 2002). For example, females develop privacy rules around their sexual identity in the workplace through evaluating the risk of sharing (Helens-Hart, 2017). Rules around more sensitive information (i.e., health data) may be taken more seriously than other types of information (Venetis et al., 2012). Outside of CPM, findings of this study illustrate the goal-oriented nature of information sharing (Bazarova & Choi, 2014). For instance, Facebook users perceive higher costs and reputational concerns for emotional and controversial topics than those that are informational or trending (Lin et al., 2024). In contexts such as sensitive health data disclosures, individuals prefer anonymous online communities to ensure privacy (Frost et al., 2014). Future research assessing privacy management of others’ information should quantitatively control for the perceived risk and benefits of sharing. Additionally, future research should take a focused approach to understand how screenshot collection and sharing differ based on the context of the conversation.

Obscurity: making screenshots useless

Unlike accountability, obscurity prevented co-owners from engaging in privacy turbulence. This result provides evidence supporting Hartzog’s (2018) argument for embedding obscurity within design, showcasing privacy turbulence—specifically screenshot collection—can be prevented through obscurity. Obscurity successfully removed the functionality of the screenshot feature, which was no longer able to capture private conversations to share with others. As screenshot sharing often does not occur without collection, the aforementioned finding demonstrates obscurity can stop the privacy turbulence created by the screenshot feature in its tracks.

That being said, obscurity was a positive predictor of screenshot sharing. While this opposes H3b, qualitative rationale indicates recognition that obscurity removes risk associated with sharing screenshotted digital messages. It is possible that individuals could find obscurity useful because they would be able to share screenshots of digital messages without violating someone else’s privacy, increasing perceptions of ownership. However, when both accountability and obscurity were present, people were significantly less likely to share screenshots. Participants may have felt screenshot sharing was pointless since the information owner could falsely perceive a violation to their privacy.

A behavioral echo: norms sustaining screenshot collection and sharing

Results from this study provide evidence that collecting screenshots of digital messages is perceived as a normal behavior among young adults. In fact, these normative perceptions were predictive of both screenshot collection and sharing. Platforms have the power to shape their own norms through design (Scharlach & Hallinan, 2023). Fiesler and Bruckman (2019) highlight the success of social norms is often dependent on the strength of the community within a particular platform; further, they argue these norms can do a better job at enforcing rules than the law. One way messaging platforms may be able to target norms working against the usage of the screenshot feature is through decreasing perceptions of ownership over shared information (Zhang et al., 2022). In this way, platforms can cultivate community norms that discourage individuals from sharing information which is not theirs. Future research should explore how affordance-based privacy rules may interact with perceptions of ownership of others’ information.

Practical implications

The results of this study provide implications for the design of messaging platforms. As previously noted, accountability may not be universally applicable to the prevention of screenshot collection and sharing. Messaging platforms seeking to implement accountability, as Snapchat has done, must do more than simply embed a new design feature. Additionally, this experiment showcased the efficacy of obscurity to prevent screenshot collection. Platforms should take this finding, and Hartzog’s (2018) push for obscurity in design, seriously. Geeng et al. (2020) similarly suggest blurring as a privacy-enhancing feature to keep intimate messaging secure. However, platforms should consider situations in which people need access to unobscured screenshots of messages and provide relatively frictionless routes to achieve this. These considerations should be explored by future research to better inform restrictions of the screenshot feature within messaging platforms.

Given the normalized usage of the screenshot feature within messaging platforms (Bayer et al., 2015; Shore & Prena, 2023), it is possible that young adults enjoy messaging platforms that allow for frictionless screenshot collection and sharing. Companies will need to decide if they want to vitalize private communication spaces, thereby shifting screenshot norms, or turn a blind eye to their enabling of privacy turbulence.

Theoretical implications

This study presents several implications for communication privacy scholars. First, it demonstrates the power of both interpersonal dynamics and platform design in the management of others’ information, supporting Trepte’s (2021) theory. Results also support the power of affordance-based privacy rules, a novel contribution of this article. As such, CPM cannot continue to be empirically tested in its traditional form. While the interpersonal rules are still relevant to decision-making, future work should continue to test platform rules. Platform rules are not limited to those tested in this study nor those that are affordance-based; rather, future applications of CPM can further operationalize platform rules as stagnant features, or behind-the-scenes institutional practices. As privacy research continues to expand within the field of communication, it is integral to develop theories that keep up with modern technological developments and cultural norms.

This study also recognized screenshot collection and screenshot sharing as distinct examples of privacy turbulence. Petronio’s (2002) original definition of turbulence is limited to unintended sharing of information. Future research should continue to provide evidence for the harms of interpersonal data collection.

Finally, results of this study showcase the normalization of screenshot collection and sharing on messaging platforms, in alignment with Shore and Prena (2023). This contrasts with previous research, which indicated that individuals perceive significant privacy concerns when screenshots of their private messages are taken (Cobb & Kohno, 2017). Future research should investigate whether people treat others’ data the way they want their own data to be treated. CPM has yet to consider the privacy expectations of the creators of privacy turbulence when it concerns their own personal information. These may be referred to as inherent privacy rules.

Limitations

This study has limitations. For instance, the text messages used in this study were artificially created. This obfuscated consideration of differences that would be dependent on the personal connection one may have with their conversation partner. Future research should give participants the opportunity to reflect upon their own screenshots of digital messages, noting context and conversation partner. Related to this, the current study operationalized screenshot sharing across a variety of social actors. Although this scale was deemed internally consistent, it would be interesting for future work to explore differences in how people share screenshots across receivers. Another limitation of this study was that manipulation checks were not put in place for each independent variable in an effort to maintain an effective deception (Grady et al., 2024). Although explicit and repetitive signals of each manipulation were presented during the procedure, recognition of accountability and obscurity was not confirmed. Further, a screenshot prevention plugin that manipulates design features—akin to the one presented to participants in this study—would be better assessed through a plugin installed on participants’ phones for a prolonged period of time.

A conceptual limitation of this study was defining screenshot sharing as the transport of the message from a co-owner to a single third party. In reality, third parties can continue to share screenshots with additional people without a connection to the initial information owner. In fact, this behavior is likely because a third party has no connection to the original conversation and thus lacks knowledge of the information owner’s privacy rules (Petronio, 2002). The domino effect of privacy turbulence, including the weakened impact of privacy rules as information spreads, should be examined in future research.

Of note, screenshot prevention efforts are limited in that people can capture private messages using a second device or by installing third party applications. These workarounds should not be targeted, as there are instances where private messages are needed as evidence of harassment (Citron, 2022). Rather, messaging platforms should remind users that communication is supposed to remain private. Future research should explore the use of workarounds to screenshot prevention efforts and how this may alter a screenshot’s authenticity (Yadav et al., 2023).

Conclusion

This study provides evidence for the power of interpersonal and affordance-based privacy rules on screenshot collection and sharing. It contributes novel practical interventions for messaging platforms and advances an existing theoretical framework on privacy management. In order to have truly private messaging platforms, platform designers should acknowledge how features integrated in the mobile hardware may threaten their intended affordances (e.g., ephemerality and privacy). The desire to maintain users’ expected privacy boundaries should give companies the incentive to do this. By mitigating use of the screenshot feature to capture private conversations, platforms can start shifting perceptions of ownership and entitlement over others’ information.

Supplementary material

Supplementary material is available online at Journal of Computer-Mediated Communication.

Data availability

The data underlying this article will be shared on reasonable request to the author.

Conflicts of interest: The author declares that there is no conflict of interest.

Acknowledgments

I would like to extend my deepest gratitude to my dissertation committee who worked with me to develop and believed in this project from start to finish: Kelsey Prena, James J. Cummings, Woodrow Hartzog, and Chris Chao Su. I also want to thank the Michaels Hoefges Graduate Student Research Fund, awarded by the Law & Policy Division of the Association for Education in Journalism in Mass Communication (AEJMC), and Boston University's Division of Emerging Media Studies for providing funding for this project. Finally, I would also like to thank the anonymous reviewers and Editors for their incredibly thoughtful feedback on this manuscript.

References