Privacy Beyond Generation: Education, Gender, and Social Media Surveillance in Türkiye

Abstract

Social media platforms make everyday visibility, interpersonal observation, and data collection increasingly difficult to separate. Drawing on semi-structured interviews conducted in Türkiye in 2022 with 21 participants from Generations X, Y, and Z, selected through stratified purposive sampling by generation, gender, and education, this study examines how participants defined privacy, managed visibility, and interpreted surveillance. The findings do not support a linear decline in privacy from older to younger cohorts. Generational differences were clearest in platform embeddedness, whereas education and gender cut across cohort lines: four of the six participants who acknowledged a privacy violation were university-educated, and the three primary-educated women, one in each cohort, did not see themselves as monitored. Across cohorts, a discourse of perceived control—“it is not violated as long as I do not allow it”—coexisted with continued participation, audience segmentation, and the observation of others. Because each gender–education combination was represented by one participant per cohort, these patterns are exploratory. The study conceptualizes privacy on platforms as situated boundary work by users who are simultaneously observers and observed, and treats generation as one social location among others rather than a stand-alone explanation.

1. Introduction

Social media platforms have transformed the conditions under which individuals become visible, communicate with others, and generate data about themselves. Activities that appear ordinary—posting photographs, following accounts, reacting to content, searching for people, or navigating a feed—can simultaneously function as social interaction and as data production. Privacy on social media is therefore not only a matter of what users deliberately disclose. It also concerns who can access information, how audiences are separated, how behavioural traces are interpreted, and how users understand the possibility of being monitored or classified. This article approaches these processes primarily through privacy boundary management, while using surveillance and generational technological socialization as complementary analytical lenses.

This makes the relationship between privacy and participation difficult to describe through a simple public/private distinction. Users may want social connection, professional visibility, entertainment, or personalized content while also attempting to limit access to particular aspects of their lives. Privacy-calculus research has long addressed this tension through the balance between perceived risks and benefits. Recent work continues to show that personalization and privacy concerns can coexist rather than cancel one another out (Cloarec et al. 2024). At the same time, the privacy-paradox literature has become more cautious about assuming a stable gap between concern and behaviour. A recent replication of three influential communication-privacy studies found that the relationship between privacy concerns and self-disclosure was sensitive to measurement and analytical decisions, underlining the need to avoid treating the privacy paradox as a universal mechanism (Masur and Ranzini 2026).

Research on privacy fatigue further complicates the picture. A study using data from 710 WeChat users found that privacy fatigue mediated the relationship between privacy concerns and disclosure intentions and was positively associated with these intentions and behaviour (Dong 2024). This is relevant to users who recognize privacy risks yet describe continued participation as normal, unavoidable, or necessary for social life. Such users may neither be unaware of privacy nor fully willing to surrender it; instead, they may combine concern, accommodation, and protective action.

Audience structure is another central dimension. Social media can bring otherwise separate relational contexts into the same communicative environment. Yet users do not merely submit to context collapse. Research on imagined audiences on Instagram shows that users differentiate audiences in relation to platform affordances and self-disclosure goals (Yao et al. 2024). Privacy management can therefore involve deciding not only what to share, but with whom, under what conditions, and through which account or platform.

Surveillance introduces an additional layer. Social media users can be observed by other users and by platform systems while also observing others themselves. They search profiles, inspect digital traces, and use publicly available information to learn about acquaintances or strangers. Simultaneously, personalized advertising and recommendation systems can create the impression that platforms know more about users than they explicitly disclosed. Research on algorithm awareness shows that awareness of personalized algorithms can shape privacy concerns and self-disclosure (Shin et al. 2022), while more recent work connects algorithm awareness, privacy fatigue, and communication privacy management in social media environments (Craig 2025).

A generational perspective can illuminate how these practices are embedded in different histories of technological socialization. Generations X, Y, and Z encountered digital media at different stages of the life course, but generational categories should not be treated as homogeneous psychological types. Recent research with 902 social media users in China found generational differences in the relationships among privacy concerns, self-disclosure, and relational outcomes, while also showing that frequency of social media use moderated these relationships (Qian and Zhang 2026). This suggests that generation matters, but not in isolation from patterns of use and social context.

The present study examines these questions through qualitative interviews conducted in Türkiye in 2022. The interviews capture a specific moment in the normalization of platform-based visibility and surveillance; the article does not treat the technical affordances of 2022 platforms as unchanged in 2026. Rather than asking which generation “values privacy most,” it asks how participants located in different generational cohorts defined privacy, managed visibility, observed others, and interpreted being observed. The empirical material points less to the disappearance of privacy than to changing definitions of privacy and different techniques for managing its boundaries.

The study addresses three research questions. RQ1: How do members of Generations X, Y, and Z conceptualize privacy in the context of social media? RQ2: How do privacy-related practices and perceptions of surveillance differ across generations? RQ3: How do social media practices shape the management of boundaries between visibility and privacy across generations?

The article’s central contribution is twofold. First, it shows how privacy boundary management is reorganized when users simultaneously participate in interpersonal observation and perceive themselves as objects of platform surveillance. Second, by comparing participants across generation, gender, and education, it shows that, within this sample, education and gender cut across generational differences in how privacy was defined and how its violation was judged. Generational comparison is therefore used to examine how boundary work is narrated from different technological and life-course locations, not to rank cohorts by the amount of privacy concern they possess. The qualitative material is especially useful for explaining practices that can appear contradictory in survey measures, such as denying that privacy is violated while suspecting that one’s phone is listening, or seeking protection from observation while observing others. In media and communication terms, these findings connect privacy boundary management to audience practices by showing how users negotiate visibility not only through disclosure choices but also through differentiated access, interpersonal observation, and perceptions of platform surveillance.

2. Literature Review and Theoretical Framework

2.1. Surveillance, Platforms, and the User as Observer and Observed

Surveillance scholarship has long examined how visibility and the possibility of observation shape conduct. Bentham’s Panopticon (Bentham 1995) and Foucault’s (1977) analysis of disciplinary power remain foundational for understanding the internalization of possible observation. Later accounts moved beyond a single centralized observer. Haggerty and Ericson’s (2000) surveillant assemblage, Lyon’s (2007) overview of surveillance studies, and work on social media surveillance (Fuchs 2011) emphasize distributed data flows, multiple observers, and the recombination of information across institutional and technological settings.

Platformized communication intensifies this distribution. Users voluntarily produce content and interaction data while platforms organize visibility through ranking, recommendation, and personalization. Surveillance-capitalism scholarship emphasizes the economic value of behavioural data and prediction (Zuboff 2019), whereas social media surveillance scholarship draws attention to the ways participation itself becomes entangled with monitoring (Fuchs and Trottier 2015). For the present study, the key analytical point is that users are not only objects of observation. They can also observe one another through profile searches, follower networks, public posts, and accumulated digital traces.

This dual position is important for understanding privacy. A user may object to unwanted monitoring while regarding the inspection of another person’s profile as ordinary curiosity. Social media therefore creates a recursive relationship between privacy and surveillance in which observing, being observed, anticipating observation, and regulating self-presentation can occur at the same time. This lateral, peer-to-peer monitoring has been described as social surveillance, in which users routinely watch one another and adjust their self-presentation in anticipation of being watched (Marwick 2012).

2.2. Privacy, Disclosure, Privacy Calculus, and Privacy Fatigue

Privacy is a multidimensional and context-dependent concept. Westin’s (1967) influential formulation treats privacy as a matter of control: individuals decide for themselves what others may learn about them and under what circumstances. Nissenbaum’s (2010) contextual-integrity framework shifts the emphasis from secrecy to the appropriateness of information flows within particular social contexts. In social media environments, control is further complicated by persistence, replicability, collapsed audiences, platform data collection, and inferences derived from behavioural traces.

The privacy paradox describes the apparent inconsistency between expressed privacy concerns and disclosure behaviour; behavioural research attributes it to uncertainty about consequences, context-dependence, and the malleability of privacy preferences (Acquisti et al. 2015). Yet recent replication work cautions against assuming that the relationship is stable across measures and contexts. Close replications of privacy-calculus, context-collapse, and privacy-paradox studies found substantial variation in replicated effects and showed that the privacy concern–self-disclosure relationship was particularly sensitive to analytical specifications (Masur and Ranzini 2026). This strengthens the case for qualitative research that examines how users themselves explain the coexistence of concern and participation.

Privacy calculus provides a related but distinct explanation: users may accept disclosure when perceived benefits outweigh perceived risks. This logic has been extended to the personalization–privacy tension, showing that the value of personalized experiences can coexist with concerns about information collection (Cloarec et al. 2024). Risk mitigation also matters. Protective strategies employed by social media users help explain why privacy concern and continued use can coexist (Gruzd and Hernández-García 2024).

Privacy fatigue adds another layer. Research has found that privacy fatigue mediated the relationship between privacy concern and disclosure intention among WeChat users (Dong 2024). Related work conceptualizes similar responses as privacy cynicism (Hoffmann et al. 2016) or as a digital resignation that is actively cultivated by corporate practices (Draper and Turow 2019), treating them as responses to perceived powerlessness rather than as evidence that privacy no longer matters. These concepts are useful for interpreting accounts of resignation, habituation, and the sense that users must overlook some privacy risks in order to remain socially connected.

2.3. Context Collapse, Imagined Audiences, and Boundary Management

Privacy on social media is also an audience problem. Content intended for friends can become visible to colleagues, relatives, strangers, or other audiences that were not salient at the moment of posting. Research on context collapse and the imagined audience addresses this convergence (Litt 2012), while communication privacy management theory, building on Altman’s (1975) conception of privacy as a dialectical process of boundary regulation, emphasizes that privacy is produced through rules governing access, disclosure, and co-ownership of information (Petronio 2002). Research shows that young people care about privacy but pursue it by managing audiences and meanings rather than by withdrawing from networked publics (boyd 2014). These perspectives are complementary rather than competing, and the present article draws on all three. Communication privacy management supplies the vocabulary used at the interpretive stage—privacy ownership, privacy rules, boundary coordination, and boundary turbulence—and is applied to the findings in Section 4.2 and Section 5.6. The analysis itself is descriptive and organized by the five domains set out in Section 3.4 rather than derived from the theory; communication privacy management is used to name and connect patterns that the domain-by-domain comparison brought into view, not to generate the categories in advance. Users themselves have long disagreed about whether social network sites constitute public or private space (Burkell et al. 2014).

Users respond with practical strategies: private profiles, restricted messaging, selective acceptance of followers, self-censorship, multiple accounts, and differentiated platform use. Instagram users’ imagined audiences vary according to affordances and disclosure goals (Yao et al. 2024). Such findings support a relational understanding of privacy in which the crucial question is often not whether information is secret, but whether it reaches the intended audience in the intended context.

In this article, “privacy negotiation” is therefore used descriptively rather than as a claim to introduce a wholly new concept. It refers to the continuing adjustment of boundaries around visibility and accessibility. The study’s contribution lies in examining how this boundary work differs across generational locations and how it intersects with both interpersonal observation and perceived platform surveillance.

2.4. Generations as Technological and Historical Locations

Conceptually, this article follows Mannheim’s (1952) understanding of generations as shared historical locations rather than fixed age brackets. Operationally, this article defines Generation X as those born between 1965 and 1979, Generation Y as those born between 1980 and 1999, and Generation Z as those born from 2000 onward, drawing on classifications used in popular and organizational literature (Lancaster and Stillman 2002; Twenge et al. 2010). Such boundaries are conventions rather than empirically fixed thresholds, and they are contested: the Pew Research Center, for example, dates Millennials to 1981–1996 and places those born from 1997 onward in Generation Z (Dimock 2019). Because cohort boundaries—especially those of Generation Y—vary across classifications, the principal sample was deliberately concentrated in narrow birth ranges (see Section 3.2) to minimize classification overlap.

Generational labels can be analytically useful only if they are not treated as deterministic. In Mannheim’s (1952) classic formulation, a generation shares a location in historical time, but members of the same generational location can form distinct “generation units” that respond differently to shared experiences (Pilcher 1994). The same cohort contains substantial variation in education, gender, occupation, digital competence, and platform use. Accordingly, the present article treats generation primarily as a technological and historical location: a shorthand for different life-course encounters with digital media rather than a fixed predictor of privacy attitudes.

This caution is reinforced by empirical evidence. Generational differences in privacy concerns and self-disclosure have been observed, with social media use frequency modulating these relationships (Qian and Zhang 2026). Research on digital inequality points in the same direction: internet skills vary substantially within the so-called net generation, particularly by education and socio-economic background (Hargittai 2010), and privacy-related literacy is unevenly distributed across users and shapes protective behaviour (Park 2013). Age differences are also non-linear: younger Facebook users disclosed more information but also adopted more privacy-protective measures than older users (Kezer et al. 2016). Comparative work on Generations X, Y, and Z in Indonesia has likewise linked generational differences in social media use to distinct perceptions of risks and benefits (Nuzulita and Subriadi 2020). The qualitative approach used here complements such evidence by examining how participants narrate the meaning of privacy and surveillance within everyday platform use.

2.5. The Turkish Context: Mahremiyet and Political Visibility

The Turkish term mahremiyet carries connotations that the English “privacy” only partly captures. In Göle’s (1996) analysis, the mahrem denotes the domain of the intimate and the forbidden—the interior space of the family and, especially, of women’s bodies, shielded from the gaze of outsiders—and it has been central to debates on modernity, gender, and visibility in Türkiye. In this sense, mahremiyet refers less to control over information than to a moral boundary around the intimate—bodies, homes, and family relations—whose protection has historically been tied to the regulation of women’s visibility and to family honour (namus) (Kandiyoti 1987). This understanding differs from the informational conception of privacy that dominates platform-privacy research. Studies of social media in other societies where privacy is similarly bound up with modesty and family reputation show that users extend such norms to online visibility, for example by managing what images of women and family life circulate beyond trusted circles (Abokhodair and Vieweg 2016). A comparable moral register may therefore shape how users in Türkiye judge exposure on social media.

Political visibility forms a second contextual layer. In July 2020, an amendment to Türkiye’s internet law obliged social media platforms accessed by more than one million users daily to appoint a legal representative and to host user data in Türkiye (Tunç 2021), and on 27 May 2022—during the fieldwork for this study—a bill criminalizing the dissemination of disinformation was submitted to parliament (Esen 2022). The bill was enacted in October 2022 (Cantekin 2022). The interviews were therefore conducted at a time when the political consequences of online visibility were a matter of public debate.

This political dimension has a longer history. Law No. 5651 on the regulation of internet publications, in force since 2007, has served as the legal basis for extensive access blocking; Wikipedia, for instance, was blocked in 2017 and remained inaccessible while the case moved through the courts, including individual applications to the Constitutional Court (Arvas 2019). Tufekci’s (2017) account of networked protest in Türkiye shows how social media became simultaneously a space of political expression and of exposure to state and social sanction. Turkish research on social media privacy has also documented variation along the lines examined here: users’ privacy expectations and concerns about personal data rose with educational level (Alkan 2022), and Generation X Facebook users shared according to privacy limits they set for themselves (Çevik 2019)—an early indication of the perceived-control discourse identified in this study.

3. Materials and Methods

3.1. Research Design

This article reports a descriptive qualitative interview study derived from the author’s doctoral dissertation (Atabey 2023). The dissertation described its design as phenomenological (olgubilim) to emphasize participants’ experiences and perceptions of privacy and surveillance in everyday social media use; this designation did not mean that interpretative phenomenological analysis (IPA) was undertaken. IPA generally emphasizes idiographic interpretation of individual cases before cross-case comparison (Smith et al. 2009). By contrast, this study employed a heterogeneous, stratified sample to compare experiences across generations, gender, and educational levels through descriptive systematic analysis (betimsel analiz). The article therefore characterizes the analytic approach as descriptive and comparative rather than retroactively presenting it as IPA; the analytic procedure is detailed in Section 3.4.

3.2. Participants and Sampling

Participants were selected through stratified purposive sampling, a non-probability technique that allows the characteristics of predefined subgroups to be described and compared (Büyüköztürk et al. 2012). In qualitative designs of this kind, the depth of the data sought and the number of participants are inversely related, and selection is guided by the purpose of the study rather than by representativeness (Türnüklü 2000; Yıldırım and Şimşek 2021). The sample was accordingly limited to 21 participants in order to obtain more meaningful and detailed information about participants’ privacy perceptions. Because this rationale concerns the sufficiency of information rather than the exhaustion of themes, it is developed further in Section 3.4 in terms of information power.

The principal X, Y, and Z groups each included six participants: three women and three men, with primary/lower-secondary, high-school, and university education represented by one woman and one man at each educational level. To limit overlap at cohort boundaries, participants within each principal cohort were concentrated within relatively narrow age ranges. The wide overall age range (19–55 years) followed from the research questions: comparing experiences of privacy among people who encountered digital media at different life stages required three distinct generational locations. Participants’ ages at the time of the 2022 interviews are reported in Table 1. The sample was designed for structured qualitative comparison rather than statistical representativeness, so observations about the cohorts should not be generalized to entire generations.

Table 1. Participant structure.

Group	n	Age Range in 2022	Gender	Education
Generation X	6	50–55	3 women, 3 men	2 primary/lower-secondary; 2 high school; 2 university
Generation Y	6	35–42	3 women, 3 men	2 primary/lower-secondary; 2 high school; 2 university
Generation Z	6	19–22	3 women, 3 men	2 primary/lower-secondary; 2 high school; 2 university
Additional LGBTİ+ participants (Generation Y)	3	24–39	Recorded as LGBTİ+	3 university

Note. Ages are reported as participants’ ages at the time of the 2022 interviews. Participant codes combine generation (X, Y, Z), gender (K = kadın, woman; E = erkek, man), and education (İ = primary/lower-secondary; L = lise, high school; Ü = university). LGBT-1, LGBT-2, and LGBT-3 denote the three LGBTİ+ participants; to reduce the risk of identification, their codes do not include age, and their occupations, marital status, and city of residence are not reported. Their age range reflects a deliberate spread across the broad Generation Y definition rather than the narrower age ranges used for the principal sample.

Three additional university-educated LGBTİ+ participants were recruited to broaden the perspectives represented in the study, including experiences concerning identity, visibility, and privacy. Their inclusion was also intended to give space to LGBTİ+ perspectives that could be overlooked by the principal sample’s binary gender stratification. Although recruitment was attempted across generations, only Generation Y participants agreed to take part; their ages spanned the broader Generation Y definition (1980–1999), rather than the narrower range of the principal sample. All three answered the same core questions as the principal sample. Their accounts provided concrete illustrations of audience segmentation through separate accounts (LGBT-1), routine inspection of others’ profiles (LGBT-2), and perceived algorithmic classification (LGBT-2), while also differing from one another across some issues. These information-rich interviews complemented the analysis but were not treated as a separate comparison group or used to attribute differences to sexual identity. All counts in Table 2 and Table 3 refer only to the 18 participants in the principal sample; none of those counts depends on the three additional interviews.

Table 2. Cohort-level counts.

Indicator	Generation X (n = 6)	Generation Y (n = 6)	Generation Z (n = 6)
Knew their own follower count	1	4	6
Reported using platform friend suggestions	1	4	6
Located privacy in the home, family, or partner	6	2	5
Explicitly acknowledged some privacy violation	2	2	2
Believed they were monitored through social media	3	4	3

Note. Values are numbers of participants. One further Generation Z participant (ZKL) was ambivalent about being monitored, and one (ZEL) about privacy violation. Percentages are not reported because each generation–gender–education combination is represented by a single participant.

Table 3. Cross-cutting patterns by education and gender.

Pattern	Participants	Count
Primary-educated women who did not think they were being watched	XKİ, YKİ, ZKİ	3 (one per cohort)
University-educated women who rejected friend suggestions	XKÜ, YKÜ, ZKÜ	3 (one per cohort)
University-educated participants among those who explicitly acknowledged a violation	XEÜ, YEÜ, ZKÜ, ZEÜ	4 of the 6 acknowledging participants

Note. Counts refer to the 18 participants in the principal sample (six per generation); the three additional LGBTİ+ cases are excluded. These patterns are exploratory, not population estimates.

3.3. Data Collection

Data were collected between 1 April and 1 June 2022 through semi-structured interviews; nine interviews were conducted online and twelve face-to-face. Geographical distance was the principal determinant of mode: participants in the interviewer’s city were interviewed face-to-face, whereas those in other cities participated via Zoom. As documented in the dissertation, the online interviews were video-recorded and face-to-face interviews audio-recorded, with participants’ consent; recordings were transcribed and considered together with field notes. The same 30 core open-ended questions were asked across modes, with follow-up probes used as appropriate. Participants received information about the study and consent was documented through signed consent and audio-recorded confirmation in face-to-face interviews and recorded verbal consent online. The interviews lasted approximately 30–60 min. The interview questions (preceded by six demographic questions) were developed for the study and are provided in English in Appendix A. Mode was not treated as an analytic category, and responses were not systematically compared by mode. Because mode was linked to participants’ locations, any possible differences related to interview mode cannot be separated from those related to geographical context; this limitation, including possible effects on discussing sensitive matters, is acknowledged in Limitations and Future Research Section.

3.4. Data Analysis and Reporting

The dissertation employed descriptive systematic analysis (betimsel analiz): responses to the shared interview questions were summarized and compared across generation, gender, and educational level, using direct quotations to illustrate similarities, differences, and exceptions. For this article, related findings from the dissertation were reorganized into five cross-cutting Results sections without a new round of coding. All reported counts were checked against the original thesis excerpts; the quotations and counts are drawn from the dissertation (Atabey 2023). Quotations were translated into English from the verbatim Turkish excerpts, with generative AI assistance (Section 3.6), and subsequently checked by the author against the originals; no formal back-translation was conducted. The Turkish originals are supplied alongside the English translations in Table S1 (Supplementary Materials). Translation prioritized semantic fidelity, and culturally specific terms (e.g., mahrem, namus) retain their Turkish form with an English gloss. The original participant codes are retained except for the additional LGBTİ+ cases, whose age-based codes were replaced to reduce identifiability.

The analysis proceeded through four connected stages. First, interviews were recorded and transcribed, and the transcripts were considered alongside interview notes. Second, responses to the 30 core questions were organized under five overarching headings established from the research aims and interview guide before analysis: social media use habits (Questions 1–10 in Appendix A), social media sharing (Questions 11–13), individual privacy (Questions 14–23), surveillance practices (Questions 24–26), and perceptions of others’ privacy (Questions 27–30). Third, answers within each question and heading were summarized and compared across generations, gender, and educational levels; convergent and divergent examples were presented through participant quotations. Fourth, the dissertation related selected similarities and contrasts to the literature, while the present article consolidated findings across the original thematic areas. For example, Section 4.3 brings together accounts about terms of use and third-party access (Question 15), privacy violation (Question 23), and perceived monitoring (Question 26). This pre-specified thematic organization resembles an element of framework-oriented qualitative analysis (Ritchie and Spencer 1994), but the dissertation did not report a separate coding matrix, formal framework-analysis protocol, or independent inductive theme-generation exercise.

Because the stratified design fixed the sample size in advance, the study does not claim thematic saturation in the sense of continuing data collection until no new themes emerged. Sample adequacy is instead argued in terms of information power (Malterud et al. 2016): the study’s aim was narrow, the sample was specified to the comparison of interest, and all interviews followed the same question set. Put plainly, the sample size was fixed by the design (three generations × two genders × three educational levels, plus three complementary cases) rather than by an iterative judgement during fieldwork; information power is invoked not to justify stopping data collection but to explain why a small, tightly specified sample suited the study’s aim. The evidence of adequacy lies in the structured comparison itself: every cell of the design was represented, and each pattern reported in Section 4 can be traced to named participants. At the same time, the one-participant-per-cell structure means that the education and gender patterns reported here should be treated as exploratory.

3.5. Ethics and Trustworthiness

Participation was voluntary, identities were anonymized through participant codes, and the study received ethics approval from the Social and Human Sciences Scientific Research and Publication Ethics Committee of Ege University (26 January 2022; meeting/decision no. 01/15; protocol no. 1282). As a form of peer debriefing, the interview transcripts were read and checked by two academics in communication studies. Because the three LGBTİ+ participants form a small and potentially identifiable group, their age-linked participant codes were replaced and no occupational or marital-status information is reported for them. The author conducted all interviews and the analysis. As a communication researcher with a professional background in journalism and news editing, and as a member of the same national and cultural context as the participants, the author shared much of the vocabulary through which participants spoke about privacy, honour, and political caution. This familiarity facilitated rapport and interpretation but may also have made some culturally shared assumptions less visible. The author’s position as a male Generation Y researcher may also have shaped the interviews, particularly when women participants discussed bodily privacy and when older or younger participants evaluated other generations. To limit interpretive bias, the analysis retains extensive verbatim quotation and reports disconfirming cases alongside dominant patterns. Because the present article is based on previously collected qualitative interviews, claims about platform monitoring or algorithmic classification are reported as participants’ perceptions unless the research design independently established them.

3.6. Use of Generative AI

Generative AI tools were used to assist with the English translation of selected participant quotations. All AI-assisted translations were checked by the author against the original Turkish excerpts to ensure semantic accuracy. Generative AI was not used for data collection, coding, or analysis.

4. Results

The findings show differences across generations, but also substantial variation within them. Five patterns are especially important for the article’s argument: different degrees of embeddedness in social media; plural definitions of privacy and audience-based boundary management; a widespread discourse of perceived control; the coexistence of watching and being watched; and a moral reading of privacy that focuses on other people’s visibility.

4.1. Different Degrees of Social Media Embeddedness

Here, “embeddedness” refers descriptively to the extent to which social media use is woven into participants’ everyday communication, routines, and networks. Among participants in this study, the Generation X group generally described a more cautious relationship with social media. Their first encounter with social media was predominantly through Facebook, and their networks tended to mirror existing offline relationships. Only one Generation X participant knew their follower count, compared with four in the Generation Y group and all six in the Generation Z group. Attitudes towards platform-generated friend suggestions followed a similar gradient. Five of the six Generation X participants evaluated the feature negatively; XKİ found it “absurd: why would I add someone I don’t know just because it is suggested?” Four Generation Y participants reported using it, often instrumentally: YEÜ likened it to advertising logic—“it offers you your friends’ friends through an algorithm”—and used it mainly to expand a professional network. All six Generation Z participants reported adding contacts this way, ZEÜ noting that he could “even find a childhood friend there straight away.”

Dominant purposes also differed: Generation X participants emphasized everyday communication and news, Generation Y participants professional interests and hobbies, and Generation Z participants online communication itself. Yet the pattern was not a simple age gradient. The sharing frequency of primary-educated Generation Y and Z participants resembled that of Generation X, and in both Generations X and Z only the university-educated participants described social media use as a necessity of the times rather than a choice—“It is an obligation, of course. Things happen outside you, and you shouldn’t remain outside them either” (XKÜ). Friend suggestions also revealed a cross-cohort pattern among the three university-educated women in the study. All three—XKÜ, YKÜ, and ZKÜ—rejected the feature, but for reasons that went beyond unfamiliarity. XKÜ said she was “afraid of these” because they gave her “a feeling that we are being controlled” and she had “no idea how those connections are made”; YKÜ felt that adding a friend’s friend would mean “doing something the other person doesn’t want”; and ZKÜ observed: “If they appear to me, it means I appear to them, and that disturbs me.” The same feature was thus read through the lenses of opaque platform control, the other person’s consent, and reciprocal visibility.

4.2. Privacy as Home, Relation, Information, and Audience

Participants did not share a single definition of privacy, and the variation did not map neatly onto cohort. All six Generation X participants defined privacy through the home and the family: “My home, my room” (XKİ); “Privacy is first of all one’s family, one’s private life” (XEL). XEÜ extended this to his guests, whose privacy he felt obliged to respect, and to being photographed outdoors and posted by others. In Generation Y, only two participants (YKL and YEİ) located privacy in the home; the others described it in relational and behavioural terms—“a space I share with very special people or only with myself” (YKÜ), or “the limits you permit and your values” (YEÜ). In Generation Z, five of the six participants again named the home, the family, or a partner, closely resembling Generation X. The exception was ZKÜ, who offered the most informational definition in the sample: “Privacy is what stays with me. The person closest to you is your mother, and there are things even your mother shouldn’t know; that is my privacy.” Within this sample, it was the middle cohort, not the youngest, that most often defined privacy in relational terms—a pattern that runs against the assumption of linear generational change.

Participants also disagreed about whether social media is a public or private space. For XEÜ, “there is a very thin line, as if you enter the private with half a step and the public with half a step.” Others made accessibility the criterion: “However much I share, the other person can see that much” (YKÜ); “For me it is private, because the people whose requests I accept follow me; it is a space I define” (ZKÜ). Among the additional LGBTİ+ participants, LGBT-2 formulated the same criterion most explicitly: “Any post with the public setting open is public space. Private posts are private space. I think it depends on which audience can access that post.”

Audience segmentation was not confined to one subgroup. ZKÜ used Instagram’s close-friends list as a semi-private layer: “There is a close-friends section; that goes a little into the intimate… I might post a photo in pajamas there, but I don’t post an unkempt picture of myself anywhere else.” YEÜ restricted students and a craftsman he worked with, who could remain “friends” without seeing his posts, because he did not want professional hierarchies to erode: “When they see your photo with your child, you don’t want them to cross that line.” The most elaborate architecture came from LGBT-1, who maintained three Instagram accounts: “I have three different Instagram accounts: one completely public account that I use for work, one locked account with people around me, and one private account with only four or five people. I can share much more comfortably there. I have, frankly, categorized my life a little more.” In each case, privacy was reconstructed not through withdrawal but through differentiated audiences. In communication privacy management terms, these are practices of boundary coordination: participants decided which audiences would become co-owners of which information and built the boundaries to match (Petronio 2002).

4.3. Perceived Control, Accommodation, and Protective Action

When asked whether social media violated their privacy, most participants answered through a discourse of personal control: “As long as I don’t share, it doesn’t” (XKİ); “I don’t think so, because I don’t allow it” (XKÜ); “Not as long as I don’t permit it” (XEİ); “No, because I am the one who sets the limits; if I want, I show nothing at all” (YKÜ). ZKL framed the question in terms of bodily display, feeling unviolated “because I don’t post anything revealing.” This discourse spanned all three cohorts and all educational levels, and it located the privacy boundary in individual posting decisions rather than in platform data practices. Analytically, it expresses a strong sense of privacy ownership that is attached to what users author, while the behavioural data that platforms extract and monetize remain outside the boundary that participants believed themselves to control (Zuboff 2019).

A smaller group acknowledged violation, and did so mostly in informational terms. Six participants clearly acknowledged some violation (XEL, XEÜ, YEL, YEÜ, ZKÜ, and ZEÜ); four of them were university-educated (XEÜ, YEÜ, ZKÜ, and ZEÜ), and ZEL was ambivalent. ZKÜ described what she understood to be a chain of account linkages: “When we join Instagram we accept lots of things. Every Android phone has a Google account… Instagram connects through the Google account and they access all my information—my age, my school, my location. This is really disturbing, but there is nothing I can do.” This account describes how the participant understood the platform’s access to her information; the study did not verify the data flows she described, and such accounts are reported here as perceptions rather than as evidence of particular data-collection practices. In communication privacy management terms, such moments are experienced as boundary turbulence: information appears to flow along routes that the participant neither authorized nor can see (Petronio 2002). Others combined acknowledgment with the control discourse—“If we are here, there is of course a certain degree of violation, but what matters is how much of it you allow” (YEÜ)—while XEL, a high-school graduate, defined violation interpersonally, as others copying and re-sharing his posts. Education thus appeared to shape less whether participants cared about privacy than which privacy—bodily, domestic, interpersonal, or informational—they had in view when judging violation.

Acknowledgment rarely led to withdrawal. YEİ, who suspected that his phone was listening because “when we say tire, a tire advertisement appears,” added: “we still don’t stop using social media.” YKÜ described habituation: “I used to feel very anxious, but now I have got used to it.” Among the LGBTİ+ participants, LGBT-2 stated: “Of course I think [social media violates my privacy], but I suppose we have become used to it. In order to be there, we overlook many things.” Protective action nonetheless remained visible: closed profiles, approval-based messaging, refusing app permissions (ZKL) and cookies (LGBT-2), and withholding political views from profile information (XKL, XEİ). XEÜ, who described himself as a programmer, captured the ambivalence of such actions: “I always mark consent requests as rejected; but as a programmer, let me tell you, reject as much as you want, it doesn’t work—it just satisfies you psychologically.” Terms of use and privacy policies were read more often in Generation X than in Generation Z, where those who reported reading them described skimming; several Generation X participants recalled reading them only vaguely. None of the three primary-educated men reported reading them at all. Concern about third-party access was often translated into concrete fears such as account theft.

4.4. Watching and Being Watched

Observation of others was unevenly distributed. In the Generation X and Y groups, looking at strangers’ profiles was generally described as unusual or pointless—“I never look; after all, it doesn’t concern me” (XKL)—or limited to checking who had sent a request. In Generation Z, the practice differed by gender and education: the women and the primary-educated man said that they did not look at strangers’ profiles, whereas the high-school and university-educated men did. ZEL looked “if there is a reason… for example, what someone from a past relationship is doing now,” and ZEÜ out of curiosity or by clicking through tags. ZKL recalled that she and her friends had opened fake accounts in high school to look at circles they were curious about. For the three LGBTİ+ participants, profile inspection was more routine; LGBT-2 explained that when meeting someone new, they might search that person on social media and Google “to get to know people,” while drawing an ethical boundary around using such searches in professional decisions.

Perceived surveillance also varied within cohorts. Three Generation X participants (XKL, XEİ, and XEÜ) believed they were monitored through social media, while the remaining three did not: two (XKİ and XEL) said they had never noticed anything of the kind, and one (XKÜ) attributed the low likelihood to spending little time on social media. Among those who did believe they were monitored, the grounds differed: XEİ was unconcerned because he did not share very private content, XKL suspected that she was being watched or listened to, and XEÜ regarded it as a given in digital life. In the principal Generation Y group, four of the six believed they were monitored, and all four cited personalized advertising as evidence; YEL attributed it to cookies, and YEİ to the phone “listening.” Generation Z responses were mixed—three participants (ZKÜ, ZEL, and ZEÜ) believed they were monitored and ZKL was ambivalent: for ZKÜ, “in the digital age we can do everything with a phone, and the price of that is being tracked,” whereas ZEİ saw no privacy problem “as long as we are not doing anything illegal.” One observation spanned all three cohorts, although it rests on three individuals: the three primary-educated women (XKİ, YKİ, and ZKİ) all said that they did not think they were being watched—“If they are watching, I am not aware of it” (XKİ). These distributions are reported as qualitative sample patterns rather than estimates of generational prevalence.

LGBT-2 offered the most explicit account of perceived algorithmic classification. Referring to Instagram’s search/recommendation environment, the participant said: “This means: we coded you as gay. These are the things you like, so this is what we show you.” The participant continued by objecting to being reduced to a narrow inferred category. The significance of this account is not that the study technically verifies Instagram’s inference process; rather, it shows how personalization can be experienced as surveillance that moves from observation to classification.

Taken together, the interviews show a recursive surveillance environment. Users may search for others, anticipate being searched, interpret personalized content as evidence of data collection, and modify disclosure practices in response. Privacy management therefore occurs within a field in which watching and being watched are intertwined. From the perspective of communication privacy management, this recursive field complicates privacy ownership: participants claimed ownership of what they posted, yet friend suggestions, personalized advertising, and profile inspection by others redistributed access to their information without explicit boundary coordination. Interpersonal inspection resembles social surveillance, in which users routinely watch one another and adjust their self-presentation in anticipation of being watched (Marwick 2012), while the reading of advertisements and recommendations as evidence of monitoring registers, at the level of everyday experience, the data-extraction logic that surveillance-capitalism scholarship describes (Zuboff 2019); the two forms are experienced not as separate domains but as overlapping sources of exposure.

4.5. Privacy as a Moral Norm About Others’ Visibility

Asked whether social media had changed society’s understanding of privacy, almost all participants agreed that it had, but most located the change in other people’s bodily and domestic exposure rather than in data collection. YEİ explained the change through bodily visibility and described swimwear and bikini posts as “dangerous posts”; ZEİ observed that people now post photographs “in their nightgowns at home, and with that photograph privacy disappears”. ZKL explained the change through imitation—because people see swimwear and bikini photographs on social media, “many people want to imitate this.” YEÜ made the moral register explicit, arguing that in societies like Türkiye privacy is equated in people’s minds with honour (namus) rather than with private life alone: “privacy, in societies like ours, is equated in people’s heads with something corresponding to honour, beyond private life”. Generation X and Z participants voiced the sharpest disapproval of other people’s swimwear photographs, while Generation Y participants more often framed such posts as a matter of personal choice—“they concern their own preferences”; “I think it is very personal”—although two of them, YKL and YEÜ, held that such posts required a sense of proportion, and the remaining four reported no discomfort with them. The boundary between judging others’ visibility and regulating one’s own was not drawn consistently. YEİ, who called swimwear posts “dangerous,” also said that he would share a swimwear photograph of himself—“I would share my swimwear photograph from the sea; it wouldn’t be much of a problem”—and, asked directly about other people’s swimwear posts, replied only that they were “about them, not my concern”. His objection was thus directed at posts that appeared to exceed or erase privacy rather than at bodily visibility as such, a distinction YEÜ drew explicitly when he said that he did not approve of photographs that “greatly exceed or entirely eliminate the perception of privacy” and that “there has to be a balance”. This is consistent with the pattern identified in Section 4.3: participants located the privacy boundary in their own posting decisions rather than in what others chose to display.

A smaller number of participants linked the change to recording and data practices instead. XKL pointed to the ubiquity of cameras and to the fact that “everything on mobile phones is recorded,” and said that these fell within the domain of privacy. XKİ noted that even shopping is registered—“you shop and they know even that; they are probably recording it”—and said that this ought to be a cause for concern. XKÜ recalled that after her family looked for a new refrigerator, “the refrigerator now appears in front of me everywhere; there is no privacy”. ZEÜ argued that the cookies users are obliged to accept on most websites “disrupt and remove privacy,” because they collect personal data. Notably, the same participant described himself as an open-profile user who shares most information about himself and wants people to see it, which shows that a participant could treat data collection as an erosion of privacy while accepting the visibility of his own content. The coexistence of these two registers matters for interpretation: for many participants, the most salient privacy problem on social media was what others chose to show, not what platforms collected or inferred.

Political opinion formed a further, carefully guarded boundary. Participants who otherwise shared personal details withheld their political views from their profiles; YEİ and YEL stated that they did not make political posts, and XEÜ did not follow parties or party leaders because “that view is private to me”. YEL linked this reticence directly to the political climate: “Because I think there is not sufficient freedom of expression in the current situation, I do not much prefer to share my political comments on social media”. LGBT-2 was not concerned about companies acquiring personal data but worried that, “given the way the world is going,” such data might “fall into the hands of an oppressive, totalitarian regime,” adding that preventing this would require complete isolation from social media, “so it does not seem very realistic”. In these accounts, privacy protected not only bodies and homes but also political identity, and the relevant observer was not only the platform or the peer group but also the state.

5. Discussion

Taken together, the interviews answer the three questions set out in Section 1. With respect to how privacy is conceptualized, no single definition was shared, and the variation did not follow a cohort gradient: Generation X participants defined privacy almost entirely through the home and family, five of the six Generation Z participants returned to the same register, and it was the middle cohort that most often described privacy relationally or behaviourally. Across the sample, privacy was more often located in a boundary that participants believed they drew themselves—“however much I share, the other person can see that much”—than in secrecy.

With respect to privacy practices and perceptions of surveillance, the clearest generational gradient concerned platform embeddedness, from follower awareness to the use of friend suggestions (Table 2). Perceptions of surveillance did not follow the same gradient. Instead, the most consistent pattern ran along education and gender rather than cohort (Table 3). The dominant answer to the question of violation—that privacy is not violated as long as the participant does not allow it—spanned all three cohorts and all educational levels.

With respect to boundary management, participants did not withdraw from visibility but redistributed it. A close-friends list, the restriction of professional contacts, and the maintenance of three separate Instagram accounts each allocated different kinds of visibility to different audiences, and these arrangements were accompanied by selective disclosure, closed profiles, approval-based messaging, and refusal of permissions. Boundary management therefore appears less as a choice between public and private than as an ongoing allocation of accessibility—private from whom, in which account, and under which conditions.

5.1. Generational Differences Without a Linear Privacy-Decline Thesis

The findings do not support a simple narrative in which privacy concern declines steadily from Generation X to Generation Z. Within this sample, generational differences were clearest in platform embeddedness—first platforms, follower awareness, and the use of friend suggestions—whereas education and gender cut across cohort lines in how privacy was defined and how violation and surveillance were judged (Table 3). Generation Z participants’ definitions of privacy were, in this sample, closer to those of Generation X than to those of Generation Y. Given the one-participant-per-gender–education cell design, these cross-cutting patterns should be treated as analytically suggestive rather than stable group differences. This reading accords with evidence that skills and privacy literacy vary substantially within generations along lines of education and social position (Hargittai 2010); later survey research likewise reports that generational differences can vary with patterns of use (Qian and Zhang 2026). A further caution concerns age and life course: because cohort and life stage coincide in this cross-sectional design, the similarities in the privacy definitions of Generation X and Generation Z may reflect life-course positions, including their relationships with home and family, rather than cohort alone. Differences in reported uses of social media—existing relationships in Generation X and work and self-presentation in Generation Y—also echo findings from Indonesia (Nuzulita and Subriadi 2020).

Theoretically, this supports reading generation in Mannheim’s (1952) sense of a shared historical location that is internally differentiated, rather than as a uniform psychological type. The additional Generation Y LGBTİ+ participants, who displayed intensive multi-platform use alongside the highest awareness of surveillance in the sample, further illustrate how other social positions shape technological experience within a cohort. Such variation warns against translating cohort labels into fixed personality traits or universal privacy orientations.

5.2. Perceived Control, Resignation, and Everyday Accommodation

The dominant answer to the question of violation—“not as long as I don’t allow it”—was the clearest cross-generational pattern in the data. It locates privacy in what users decide to post and thereby recasts platform data collection, which users can neither see nor directly control, as a matter of personal discipline. The privacy-paradox concept offers one vocabulary for the coexistence of concern and continued participation, but recent replication evidence suggests that the concern–disclosure relationship is not stable enough to be treated as a universal paradox (Masur and Ranzini 2026). The interviews point instead to a combination of perceived control, habituation, and selective protection. XEÜ’s remark that rejecting consent requests “just satisfies you psychologically” shows that even protective action can be experienced as largely symbolic.

These accounts resonate with work on privacy fatigue (Dong 2024), privacy cynicism (Hoffmann et al. 2016), and digital resignation (Draper and Turow 2019). The present interviews cannot test mediation, but they show how resignation is narrated in everyday language—“now I have got used to it,” “we still don’t stop using social media,” “in order to be there, we overlook many things.” At the same time, closed profiles, restricted audiences, permission refusal, and selective disclosure demonstrate that fatigue does not necessarily eliminate protective action. The perceived-control discourse may itself serve a protective function: it allows users to maintain a sense of agency in an environment in which the most consequential data flows lie beyond individual posting decisions.

This interpretation also aligns with work that emphasizes users’ risk-mitigation strategies (Gruzd and Hernández-García 2024). The apparent contradiction between “I am concerned” and “I still use social media” can partly dissolve when the analyst examines what users do between those two positions.

5.3. From Context Collapse to Audience Segmentation

Audience segmentation was one of the clearest practices in the qualitative material, and it was not confined to one subgroup. ZKÜ’s close-friends list, YEÜ’s restriction of students and workers, and LGBT-1’s three Instagram accounts each assign different kinds of visibility to different audiences. Privacy was achieved not through invisibility but through differentiated accessibility.

This finding resonates with research on context collapse and imagined audiences (Litt 2012; Marwick and boyd 2011) and with communication privacy management theory’s account of privacy rules and boundary coordination (Petronio 2002). The present study shows how users construct their own privacy architecture from platform features and, where these are insufficient, from workarounds. YEÜ’s concern with preserving professional hierarchies also shows that segmentation protects not only personal information but social roles. Audience segmentation is thus a response not only to context collapse but also to the reciprocal visibility that social surveillance entails, in which users monitor one another and anticipate being monitored in turn (Marwick 2012). The same logic appeared in the interviews in the form of a symmetry test: as ZKÜ put it, “if they appear to me, it means I appear to them” (Section 4.1). The relevant question becomes not only “What is private?” but “Private from whom, in which account, and under which conditions?”—a formulation consistent with Nissenbaum’s (2010) view of privacy as the appropriateness of information flows in context.

5.4. Perceived Algorithmic Surveillance and Classification

The account of being “coded” by Instagram extends the analysis beyond interpersonal visibility. The participant experienced recommendation as evidence that behavioural traces had been translated into an identity category. This is analytically distinct from the belief that a platform simply collects data: it involves the perception that the platform interprets data and acts on the resulting classification.

A less articulated version of this experience appeared among older participants. XKÜ read friend suggestions as a sign that “we are being controlled” while admitting that she had no idea how such connections were made. Perceived classification thus did not require technical knowledge; the opacity of recommendation systems could itself generate a sense of being watched. The belief that phones “listen” worked in the same way, although it was voiced in two registers. YEİ inferred it from a strikingly well-timed advertisement—“when we say tire, a tire advertisement appears”—and asked whether he was being listened to, whereas XEL, who had said when asked directly that he had not concluded he was being watched through social media, treated the listening of mobile phones as something he had already heard was happening—“as far as we hear, even mobile phones are being listened to now”—and said the situation troubled him. Whether or not audio is actually captured, opaque personalization and the conviction that devices register what is said around them are experienced as eavesdropping, and it is this experience, rather than technical knowledge, that shapes how participants understand being watched.

Algorithm-awareness research provides a useful bridge: awareness of personalized algorithms is associated with privacy concerns and self-disclosure (Shin et al. 2022), and algorithm awareness has also been linked to communication privacy management and privacy fatigue (Craig 2025). The present qualitative account adds the subjective experience of classification: the participant objected not only to being known but to being reduced to a category that then structured what the platform displayed.

Because the study did not audit platform systems, this finding must remain explicitly framed as perceived algorithmic surveillance. That limitation is also theoretically useful: perceptions of surveillance can influence privacy behaviour even when users cannot verify the technical mechanism behind a recommendation.

5.5. Moral, Political, and Informational Registers of Privacy

The moral reading of privacy change identified in Section 4.5 helps explain why the perceived-control discourse was so widespread. When privacy is understood primarily as modesty about bodies, homes, and families—and, in YEÜ’s words, equated with honour—its violation appears to depend on what one chooses to display, and a person who “doesn’t post anything revealing” can reasonably feel protected. Informational and inferential privacy, by contrast, concern flows that users neither see nor author. The coexistence of these registers in the interviews suggests that survey items asking whether users “care about privacy” may conflate distinct normative frameworks. ZEİ’s view that there is no problem “as long as we are not doing anything illegal” reproduces the “nothing to hide” argument, which has been criticized precisely for reducing privacy to the concealment of wrongdoing (Solove 2007). In the Turkish context, the moral register resonates with Göle’s (1996) account of the mahrem as a gendered domain of the intimate that must be shielded from outside gaze, which may help explain why bodily and domestic exposure dominated participants’ judgements. The guarded treatment of political opinion points to a further register, in which privacy shields political identity from institutional and social sanction—a dimension that platform-centred models of privacy tend to overlook and that future research in politically constrained settings should address directly.

5.6. Theoretical and Practical Implications

The study’s main theoretical contribution is twofold. First, it connects privacy boundary management to a recursive surveillance environment in which users protect some boundaries while accepting other forms of exposure, observe others while anticipating observation themselves, and interpret personalization as evidence of platform classification. Second, it suggests that the meaning of privacy through which such boundary work is conducted—domestic, bodily, relational, or informational—appeared, in this sample, to vary not only by generation but also across education and gender. Generational comparison therefore adds a temporal and technological-socialization dimension to privacy research only when it is read together with the other social positions that cut across cohorts. Disclosure and protection are not necessarily opposites, and the observer/observed distinction is unstable in everyday social media use.

The five domains were analysed descriptively, but communication privacy management theory (Petronio 2002) offers a vocabulary that organizes the cross-cutting patterns more precisely. The perceived-control discourse expresses a strong sense of privacy ownership: participants treated information as theirs to release and located the relevant privacy rules in their own posting decisions. Audience segmentation—close-friends lists, restricted contacts, and multiple accounts—is a form of boundary coordination through which users allocate co-ownership of information to different audiences. Platform inference, friend suggestions, and personalized advertising, by contrast, appeared in the interviews as sources of boundary turbulence: moments at which information seemed to flow along routes that participants had not authorized and could not see. The recurring gap between strong perceived ownership and limited control over these routes is, in CPM terms, the core tension that participants managed through habituation, selective protection, and, in some cases, resignation.

For platform design, the findings tentatively suggest that granular audience controls may serve users better than a binary public/private model. The multiple-account and close-friends strategies observed here indicate that some users create their own workarounds when platform-level audience separation seems insufficient or cumbersome. Given the small and exploratory sample, these implications should be read as directions for further research rather than as design recommendations, and they would need to be tested with larger and more diverse samples.

For digital literacy, the findings suggest a need to distinguish explicit disclosure from behavioural inference. Users may understand that posts are visible while remaining less certain about how clicks, likes, follows, searches, and other traces can shape personalization (Li et al. 2025). Privacy literacy should therefore include audience management, platform data practices, algorithmic inference, and interpersonal privacy ethics. If the pattern observed here—informational framings of violation concentrated among university-educated participants—holds in larger samples, literacy initiatives should not assume that the problem is a lack of concern; rather, they should make visible the difference between controlling what one posts and controlling what platforms infer.

6. Conclusions

This study examined how 21 participants from Generations X, Y, and Z in Türkiye understood privacy and surveillance in social media environments. It found generational patterns in platform embeddedness, but no simple progression from privacy-conscious older users to privacy-indifferent younger users. Within this sample, education and gender cut across cohort lines in how privacy was defined and how violation was judged, and the most widespread cross-generational pattern was a discourse of perceived personal control over privacy.

Privacy persisted across the sample, but its boundaries were constructed differently. Some participants emphasized the home and family, while others defined privacy through selective accessibility or information that should remain unseen by particular audiences. Strategies such as restricted profiles, approval-based messaging, selective disclosure, and multiple accounts demonstrate that privacy can be reconstructed inside highly visible platforms (boyd 2014).

The study also highlights the dual position of social media users as observers and observed subjects. Participants could inspect other people’s digital traces while simultaneously interpreting advertising, recommendations, and algorithmic categorization as forms of surveillance. Privacy under platform conditions is therefore not simply retained or lost; it is continually managed within relationships of visibility, observation, and datafication. From a media and communication perspective, the findings suggest that privacy on social media is produced through the interaction of audience segmentation, boundary management, interpersonal observation, and perceived platform surveillance rather than through withdrawal from visibility alone.

Limitations and Future Research

The study is based on a purposive qualitative sample of 21 participants and does not support statistical generalization to entire generations. It was conducted in Türkiye, so the findings are embedded in a particular sociocultural setting. The interviews were conducted in 2022 and should be read as a time-bounded account of how participants understood privacy and surveillance at that moment. Platform affordances, dominant services, recommendation systems, and public awareness of data practices have continued to change. The article therefore uses 2024–2026 research to place the 2022 accounts in dialogue with subsequent scholarship, not to imply that later studies validate unchanged technical conditions. The value of the data lies in the depth of participants’ accounts and in the opportunity to compare those accounts with later developments in privacy research.

Generational categories also simplify substantial within-cohort heterogeneity (Stark and Poppler 2018). The three LGBTİ+ participants were all members of Generation Y, so their accounts cannot support cross-generational conclusions about sexual identity; LGBTİ+ participants from other generations could not be recruited. Because cohort and life stage coincide in this cross-sectional design, differences attributed to generation may partly reflect age and life-course position. Quotations were translated without formal back-translation, although their Turkish originals are provided in Table S1; this is a particular limitation for culturally loaded terms such as mahrem and namus, whose moral connotations have no exact English equivalent; these terms are therefore retained in Turkish with a gloss. Nine of the 21 interviews were conducted online, which may have affected rapport and participants’ willingness to discuss sensitive topics such as political views. Because face-to-face interviews were conducted with participants in the interviewer’s city and online interviews via Zoom with participants elsewhere, interview mode and geographical context were linked. Any possible effects of mode on rapport or disclosure therefore cannot be distinguished from those of geographical location; no systematic comparison by interview mode was undertaken. Because each gender–education combination was represented by a single participant in each generation, the education and gender patterns reported here are indicative and should be tested in larger samples. The author’s position as a male researcher may also have shaped women participants’ accounts of bodily privacy, which bear directly on the moral register identified in Section 4.5. Some interview questions also used the study’s own framing; for example, Question 29 in Appendix A referred explicitly to the “surveillance society,” which may have encouraged participants to adopt that frame. Finally, statements about monitoring, listening, or algorithmic categorization represent participant perceptions unless independently verifiable; the study did not technically inspect platform data systems.

Future research could compare these patterns longitudinally, distinguish age and cohort effects, examine platform-specific privacy strategies, and investigate how algorithm awareness changes disclosure and audience management. Mixed-method designs could also test relationships among perceived surveillance, privacy fatigue, audience segmentation, privacy literacy, and continued social media use.