Chilling or resisting? Exploring the influence of technology-facilitated (gender-based) violence on female feminists in Colombia and Costa Rica

2.1 Forms of technology-facilitated (gender-based) violence: misogynistic hate speech, doxing, and beyond

The rapid advancement of ICTs and the affordances offered by social media, including visibility, scalability, and anonymity, have supported female feminists to engage in activism, mobilize support, amplify their voices, and advance their agendas in public discourse. ^{22 , 28 , 29} Nevertheless, as Suárez et al. (2022) argue, ³⁰ the use of ICTs for both personal expression and feminist activism has also intensified intersecting forms of violence, which serve to curtail women’s participation in political and public sphere. Misogynistic narratives, which are deeply ingrained in the sociocultural mindset, ¹⁰ fuel gendered TFV as a vehicle to uphold hegemonic masculinity, branding any attempt from women to gain public visibility and active participation as a transgression of traditional gender norms, thereby making them targets for violent reprisal. ^{22 , 31} As Esposito ³² (p. 463), mentioned “all forms of gender-based violence have been working as fairly effective gate-keeping practices, fostering the silencing and exclusion from the public and political arena (…)”. Since the onset of the COVID-19 pandemic, cases of GBV have increased significantly, ¹⁶ and considering the following social distancing measures, there has been a notable shift towards online interactions. ³³ These circumstances have increased the “toxicity of online communities” and the likelihood of experiencing TF(GB)V (p. 7). ³⁴ Hinson et al. ³⁵ have defined TFGBV as the action perpetrated by one or more individual(s) that harm others based on their gender or sexual identity or by enforcing harmful gender norms using mobile technology or the internet. Additionally, Vaughan et al. ³⁶ (p. 5), identify TFGBV as “[…] any act that is committed, assisted, aggravated, or amplified by the use of [ICTs] or other digital tools, that results in or is likely to result in physical, sexual, psychological, social, political, or economic harm or other infringements of rights and freedoms”. It includes bullying, stalking, defamation, sexual harassment, exploitation, and hate speech in digital spaces. ⁴ So far, the study of TFGBV is characterized by conceptual ambiguity and inconsistent terminology. ³³ Scholars employ a range of term, including gendered TFV, gendered cyberhate, cyber violence against women and girls (VAWG), cyber misogyny, ⁸ and technology-facilitated sexual violence and abuse (TFSVA). ³³ While some terms are applied consistently, others, like gendered cyberhate, are used to draw attention to specific forms of violence such as hate speech. Building upon these definitions and trying to navigate the conceptual ambiguity, we understand the term TF(GB)V as the intersection of technology-facilitated violence, defined as harm enabled or facilitated by technological tools, ³⁷ and gender-based violence, defined as acts of violence directed at individuals based on their gender identity (Aghtaie and Gangoli 2015) (see Figure 1). ³⁸ The term technology-facilitated points to the blurred line between digital and physical violence, meaning that while TFV always has a digital component, its manifestation often spills into the physical realm. ^{20 , 39} Still, as Chen et al. highlight, TF(GB)V “isn’t just the same old sexual violence wrapped in a shiny technological coat, it presents unique challenges to prevention and mitigation”, ⁴⁰ (p. 2772).

Figure 1:

Overview of context factors influencing technology-facilitated (gender-based) violence, the various forms it takes, the female actors affected, and their responses. Source: Own depiction.

Drawing from Vlahakis’ ⁹ report on TFGBV against women’s rights activists in Zimbabwe, Nepal, and Kenya, factors such as gender identity, socioeconomic status, sexual orientation, ethnic belonging, and age underpin women’s experiences with TF(GB)V. Generally, a significantly higher proportion of BIPoC and LGBTQAI+ individuals report instances of TF(GB)V due to (intersecting) identity attributes. ²⁷ A study conducted by Posetti et al. ³⁴ with 901 female journalists from across the world reveals that TFV is often closely linked to the journalists’ gender and frequently contains sexist remarks. Not only are (publicly known) women disproportionately affected by TFV, but the topics with which they are frequently engaged – such as domestic violence, feminicides, sexual harassment, queer rights, and reproductive rights (especially abortion) – are often perceived as controversial in many (patriarchal) societies and serve as triggers for attacks. These forms of violence are not only perpetrated by external actors, like right-wing conservative groups, but can also arise from within families and communities. ¹⁸ An additional factor identified as being significantly associated with the increase of TF(GB)V is the surge in disinformation, which seems intricately intertwined with TF(GB)V. It has been observed that disinformation tactics are frequently repurposed to perpetrate misogynistic attacks and that reporting on such disinformation can itself become a trigger for coordinated pile-on attacks. ³⁴ Digital platforms, through their specific affordances, such as anonymity, interactivity, and the ability to amplify content rapidly, often function as amplifiers of violence by perpetuating and reinforcing longstanding systems of domination, exclusion, and discrimination. ^{14 , 29 , 41} Affordance theory provides a useful analytical tool to understand how specific platform features mediate users’ experiences. When viewed through a feminist lens, as in Salamoun et al., ⁴² affordances like anonymity or virality are not neutral but deeply entangled with power, culture, and gendered violence. For instance, several affordances enable perpetrators to harass, threaten, and manipulate victims with minimal accountability: The anonymity offered by platforms enables perpetrators to hide their identity, making it harder to trace or prosecute them, while the ability to share content quickly and widely can increase, amongst others, the reach of harmful messages. ³³ Furthermore, Im et al. ⁴ and Treem and Leonardi ⁴³ identify “persistence” as a key affordance of social media platforms; one that, in the context of TF(GB)V, can contribute to an environment where TF(GB)V endures and potentially escalates over time, intensifying the emotional and psychological impact on those targeted. In the context of TF(GB)V, the lack of transparency, inadequate reporting mechanisms, and the frequent emphasis on virality over safety reflect design decisions that neglect feminist HCI principles of reflexivity, advocacy, and contextual awareness. ^{26 , 44}

As Fal-Dutra Santos and Pourmalek ⁴⁵ (p. 85) state, “terms like ‘cyber violence’ and ‘technology-facilitated violence’ frame technology as the cause of violence and obscure the structural causes”. Furthermore, they highlight that “GBV does not occur in isolation but rather is a part and result of a web of structural inequalities and institutional discrimination” and thus TF(GB)V can be seen as a “continuum” of direct physical violence. Consequently, it is imperative to bear in mind that technology itself does not inherently cause GBV, but it can be used as a tool to facilitate, amplify, or perpetuate it. Thus, GBV perpetrated online and through technology is largely a reflection of underlying, deeply entrenched patriarchal structures. Thus, it is difficult to differentiate between the impacts and consequences of the violence/abuse that are initiated in digital spaces from offline realities and vice versa. ⁴⁶ Examining these issues in isolation without a concomitant understanding of the wider problems they give rise to is misleading.

2.2 Responding to technology-facilitated (gender-based) violence

How individuals navigate and respond to TF(GB)V is influenced by a combination of social, economic, political, and contextual factors and is thus inherently personal. Therefore, it is important to recognize emotional, physical, and situated experiences. ⁴⁷ As Higson et al. ⁴⁸ elucidate, in situations where human beings encounter threats, a predominantly instinctual reaction is typically observed. In violent situations, this often leads them to act based on their intuition, on fear, or a drive to fight back and “destroy the aggressor” (p. 27). In relation to TF(GB)V, the literature commonly differentiates between two main responses to TF(GB)V: (1) self-censorship and self-regulation (chilling effect), where targets limit their online activity to avoid further harm; and (2) active resistance mechanisms. Jane labels these as flight or fight, noting that individual responses “can involve both flight and fight modes” simultaneously, ⁸ (p. 58). Liang ⁴⁹ (p. 1473) highlights that “personal factors” (e.g., privacy concerns), “platform factors” (e.g., unclear platform policies), and “social factors” (e.g., fear of political retribution) can manifest in (1) self-censorship, (2) self-restraint, (3) switching platform, and (4) camouflaging. While (1) and (2) can lead to varying levels of chilling, (3) and (4) may cause anti-chilling. This section briefly introduces these strategies and reviews existing empirical research.

The first response frequently highlighted by scholars is self-censorship, meaning that individuals affected by TF(GB)V reduce their engagement to minimize the actual and potential risks linked to their online participation. ^{50 – 54} The study by the Economist Intelligence Unit ¹² shows that one in three women think twice before they post something on the internet. Violent online threats, such as insults, intimidation, shaming, and discrediting, are powerful forces often triggering deep psychological states of fear, anxiety, and severe emotional distress that then, in turn, amplify social conformity. ^{12 , 27} This phenomenon is frequently termed the chilling effect, ^{55 , 56} with self-censorship and self-regulation being typical manifestations. ⁴⁹ The term chilling effect was first introduced in the US to denote the discouragement or suppression of free speech due to the fear of legal repercussions or government action. ⁵⁷ Now, it broadly describes how laws, policies, or social conditions can discourage or force individuals from exercising their (constitutional) rights and activities due to fear of legal, social, economic, or professional repercussions. It is commonly influenced by the intensity, proximity, and perceived likelihood of swift punishment inherent to threats. ²⁴ Within patriarchal societies raising voice against men, who are often the perpetrators of TF(GB)V, is often challenging and women are often blamed and delegitimized. ⁵⁸

Nadim et al. (2021), in an analysis of online harassment in Norway, argue that personal threats directed at the target’s being instead of its ideas or thoughts are more likely to trigger self-censorship. Along the same lines, Penney, ⁵⁶ looking at the chilling effect caused by online regulation and surveillance, argues that more personal regulatory actions (e.g., individualized legal threats) lead to a strong chilling effect by evoking fear of penalty and prosecution, whereas impersonal regulations provoke a weaker form of the chilling effect. In the context of state-led prosecution in autocratic regimes in Southeast Asia, Ong ⁵⁹ argues that “social media self-censorship occurs when the expected pay-off for expressing one’s political opinion online is negative” (p. 143). Schauer ⁵⁷ similarly states that the occurrence and extent of the chilling effect is a question of benefit-risk assessment by saying that individual risk aversion is a decisive factor. In short, this means that individuals affected by TF(GB)V subjectively seem to weigh the benefits of their online engagement, such as wide outreach and easy networking, against actual and potential risks linked to TFV. These limited and heterogeneous findings support the notion that the chilling effect and other response mechanisms are influenced by social, economic, political, and contextual factors and are thus inherently personal and context-dependent. ⁴⁹

Employing a gender-sensitive lens, the large-scale study with 901 journalists by Posetti et al. ³⁴ shows that TF(GB)V against women journalists is meant “to humiliate […] and chill their active participation (along with that of their sources, colleagues, and audiences) in public debate” (p. 6). The study reveals that 30 % of respondents admitted to self-censoring on social media, while 20 % withdrew from all online interactions without reporting or publicizing experiences (ibid). The findings demonstrate that TF(GB)V is intended not only to silence individuals directly affected by it, since individuals who may not have personally experienced TF(GB)V can still be deeply affected by its chilling effect, leading them to self-censor or avoid public discourse. They observe how others in similar positions face severe repercussions, such as physical and sexual violence or death threats, which are intentionally used as warnings to silence dissent. As a result, TF(GB)V not only harms direct targets but also discourages broader participation, reinforcing systemic suppression of marginalized voices. ^{7 , 22 , 60}

While extensive research has investigated the impact of TF(GB)V on female journalists, scholars, and politicians in the Global North, comparatively fewer studies have focused on this phenomenon among (female) activists and in the Global South, particularly Latin America, ^{4 , 18 , 61} despite these groups and regions being particularly affected. ⁶² One exception is the survey conducted by Soto and Sanchez’, ⁶³ who found that 71.2 % of surveyed Chilean feminists and activists (n = 163) faced TF(GB)V, mostly misogynistic verbal attacks leading to self-censorship. Benítez ⁶⁴ and Barrios and Miller ⁶⁵ examine tendencies towards self-censorship among journalists, focusing on Colombia and Central America, concluding that this practice is more prevalent when covering contentious issues such as forced displacement and illegal mining. They also note that journalists in rural regions are more prone to self-censorship due to the extensive territorial dominance maintained by illegal armed groups, which amplifies the visibility of their work and renders them more vulnerable to physical threats. In 2023, the Constitutional Court of Colombia published a report shedding light on the issue of TF(GB)V against female journalists, highlighting the lack of reporting mechanisms and legal mechanisms for action. ⁶⁶ This is more generally an issue since solely some forms of TF(GB)V, such as cyberporn, can be treated as criminal or civil offenses; many other types are neither legally criminal nor unlawful. ⁶⁷ One reason for limited research is, according to Bedi, ⁶⁸ the difficulty of measuring the chilling effect “for reasons of data collection, data analysis, causal events, and confounding factors” (p. 270). Especially the human behaviors underlying the chilling effect – i.e., what triggers self-censorship in response to TFV – are rarely researched.

Besides chilling, distancing (e.g., ignoring, blocking, muting, and deleting content and users), improving one’s own digital hygiene, and limiting internet use (e.g., avoiding interactions with strangers, disengaging from contentious debates, and removing or greatly restricting access to personal information, privatizing accounts, etc.) are often identified coping strategies women seem to employ when being targeted with TF(GB)V ^{8 , 12 , 58} . Liang ⁴⁹ elucidates that social media content creators (from China) occasionally switch platforms and evade sensitive terms, e.g., circumventing the platform’s algorithm and mitigating the risk of encountering TFV. These coping mechanisms reflect the highly individualized ways in which people respond to TF(GB)V: A reality also shaped by the frequent trivialization of such violence and the lack of awareness or competence among official institutions, including law enforcement. ²⁰ When individuals persist in their online engagement despite being targeted, their continued presence can be read as a form of resistance and feminist mobilization. As Jane’s ⁸ analysis of TF(GB)V indicates, a “women’s response to gendered cyberhate cannot always be neatly categorized as one thing or another” (p. 54). She observed that numerous women use both strategies such as retreating by modifying their writing style and content (flight), while simultaneously maintaining active participation in overt feminist activism (fight). Nah ⁶⁹ found that some activists persist in their resistance and solidarity, feeling responsible for addressing their hostile realities, regardless of the risks involved. Within this trend, female feminists have demonstrated greater readiness to engage in discussions regarding emotional well-being and advocate for both personal and collective care within human rights defense. Similarly, Jane’s ⁸ concept of fight responses encompasses individual and collective dimensions, ranging from personal behavioral adaptations in using ICTs to coordinated campaigns to combat TF(GB)V. Hashtag feminism has paved the way for amplifying narratives countering violence from intersectional viewpoints, ^{70 , 71} while restoring agency to targets of TF(GB)V. Beyond shedding light on persisting forms of oppression, hashtag feminism promotes proactive engagement to address present and future social challenges. ^{72 , 73} Hence, collective self-defense against TF(GB)V has consolidated as a political practice articulating women’s political struggles beyond Latin America. ⁶ This has sparked feminist social mobilization and collaboration, extending to activities such as awareness-raising, advocating for policy reforms related to TF(GB)V, and providing support to women who have experienced TF(GB)V. Jane has further outlined that feminists have increasingly resorted to practices like digilantism, which involves extrajudicial actions like exposing perpetrators to ensure they are held accountable. Some individuals also seem to directly engage with attackers (e.g., confronting perpetrators privately or publicly, responding humorously to attacks).

In summary, the literature illustrates a spectrum of responses to TF(GB)V, ranging from self-censorship and total disengagement to active resistance and a combination of both. All forms of TF(GB)V violate basic human and democratic rights, such as the right to express one’s opinion, free speech, and live free from violence. ⁵⁷ Thus, individuals affected by TF(GB)V are deprived of their basic democratic and human rights, and their consequent withdrawal from social media platforms leads to a democratic deficit with voices being silenced.

3.1 Case selection

The decision to concentrate on Latin American countries, i.e., Costa Rica and Colombia, was influenced by several factors, including personal connections, active feminist movements, high rates of individuals identifying as BIPoC, and high numbers of TF(GB)V. Female feminist activists in all three countries use social media to communicate content related to their activism and to raise awareness, such as through Facebook pages dedicated to targets of GBV. Numbers of social media users exceed 60 % of the population in each country, with Colombia at 70.3 % and CR at 73.8 %. ⁷⁴ Costa Rica and Colombia exhibit elevated levels of violence, including high homicide and feminicide rates, with Colombia having 0.7 femicides per 100.000 inhabitants and Costa Rica 0.9. It is evident that Latin America exhibits a prevalence of GBV that is among the highest globally – partly due to “deep historical and structural roots of patriarchal, discriminatory and violent cultural patterns” that “have proven among the most difficult to dismantle” (p. 2). ⁷⁵ Despite existing action plans and legislation addressing GBV (e.g., Ley de Penalización de la Violencia Contra las Mujeres in Costa Rica (INAMU 2013), ⁷⁶ GBV in digital spaces remains largely unaddressed in these countries. In February 2024, for instance, the Congress of the Republic of Colombia intensified the discussion on measures that could be adopted to prevent, protect, and penalize TF(GB)V (Congreso de la República de Colombia 2024). ⁷⁷ Despite these efforts, or conceivably by reason of these efforts being so limited in scope, the rates of TF(GBV) remain high and those affected are neither safeguarded from the experience of violence in the first instance, nor assisted in managing it. ¹²

3.2 Data collection

To gain profound insights into the experiences of female feminists with TF(GB)V and their coping mechanisms, 23 in-depth interviews were conducted. Prior to conducting the interviews, risk assessments were carried out and the first author’s personal devices were encrypted to ensure data security. The participants (I0–I22) were female feminist activists who all self-identified as cis-women. The interviews were conducted in Spanish between June and August 2023. The sample comprised 16 participants from Colombia and six from Costa Rica. The interviewees represent organizations and social movements specializing in various subject matters.^[5] Their agendas revolve around the defense of Indigenous rights and territories, the monitoring and reporting of femicides, global awareness rising regarding GBV, the advocacy for sexual and reproductive rights, social mobilization for gender issues, and the psychological support of individuals who have experienced TF(GB)V. The interviewees largely reflect the diversity of feminist causes championed in Latin America, although they may not fully capture the entire spectrum of grassroots feminist groups in their contexts. Their multifaceted advocacy efforts and activism unfold in urban and rural areas, engaging individuals from diverse backgrounds, including Afro-descendant, Indigenous, peasant, migrant, and mestiza women. While some groups have a long history spanning decades, others have emerged more recently in response to the ongoing challenges faced by women in their countries.

For the interviews, a semi-structured questionnaire was utilized, incorporating inquiries regarding participants’ engagement in activism, the impact of ICTs on their activism efforts, challenges encountered in their digital activism endeavors, and strategies employed to mitigate instances of TF(GB)V. We want to acknowledge that given the evolving nature of digital platforms, sociopolitical conditions, and activist strategies, both the forms of TF(GB)V and responses to it are subject to change over time. The insights presented here reflect the realities and perceptions captured during the period of data collection and may shift in future contexts. Access to interview participants was established through a combination of local NGOs, academic networks, personal contacts, and referrals, employing purposive and snowball sampling techniques. ^{78 , 79} All participants granted verbal informed consent and did not receive compensation for their involvement, ensuring voluntary participation and minimizing the risk of creating undue incentives. Ethical clearance was obtained from our university’s Institutional Review Board before the study, and ethical considerations were paramount throughout the research process. The study adhered to the principles of do-no-harm and the ethical guidelines outlined in the Belmont Report, given the sensitive nature of the topic. ⁸⁰ The interviews were recorded via a secure recording device and transcribed offline and locally using Open AI Whisper (a speech-to-text model in Python), with corrections applied manually during the process. Upon completion of the transcription process, anonymized transcripts, free of any personal identifiers, were securely stored in encrypted form and are accessible only to the research team, ensuring the confidentiality and privacy of participants.

3.3 Data analysis

The study was based on the methodology^[6] of Grounded Theory (GT), building upon its theoretical underpinnings and core principles. Particularly GT’s emphasis on openness to emergent themes, iterative coding, and context-sensitive analysis informed the data analysis of this study. The analytical process (i.e., the method) used GT techniques, such as the different stages of coding (open, axial, and selective), as a flexible guide rather than a rigid framework, following Charmaz’ ⁸¹ approach of constructivist GT. This allowed for both a structured process of data analysis and the flexibility to work iteratively and cyclically, meaning that the stages could inform each other. While this study’s primary aim was not to develop a new formal theory, the tools offered by GT were used to sensitize the analysis, allowing for emergent patterns and context-specific interpretations grounded in the interviewee’s experiences. Thus, it enables a greater sensitivity to contextual factors (of which the analysis might go beyond strict GT procedures) and enhanced reflexivity of the researcher’s positionality, thereby prioritizing analytical richness over procedural purity. ⁸¹ Our coding process via the data-secure software MAXQDA began with detailed coding, followed by creating sub-categories, including “types of TF(GB)V”, “potential perpetrators”, and “chilling effect”, through code grouping (see excerpt in Appendix, Table 2). Guided by concepts from prior research, such as forms of TF(GB)V (e.g., doxing, cyberstalking), chilling effects (e.g., self-censorship), and resistance strategies, initial coding categories were shaped. Through inductive engagement with the data, additional grounded categories emerged, such as distrust in official institutions, perceived impunity, generalized societal violence, infrastructural deficits, and the vulnerability of female feminists.

The initial categorization was collaboratively discussed by the first two authors, and the agreed-upon approach served as the foundation for the subsequent stages of analysis. Through an iterative process, additional codes were identified and organized into overarching thematic categories. In total, we have identified 12 main categories, including 2.381 codes. Throughout the coding process, we engaged in a constant comparison between our data and relevant literature. Through this iterative analytical process, data excerpts were continuously organized and reorganized to uncover patterns, relationships, and emergent themes grounded in the participants’ narratives. ⁸¹ Recognizing that our positionality influences both the interpretation and coding of data, we find it important to be transparent about our identities as researchers. Both researchers identify as cis-gendered women, one from Colombia and one from Germany. The two researchers critically reflected on their positionalities throughout the coding process. This involved acknowledging how their personal identities, cultural backgrounds, and lived experiences might shape their interpretations of the data, thereby fostering a more thoughtful and reflexive approach to the research. While the study’s findings are not intended to be representative, they aim to offer profound, contextually specific insights into how Costa Rican and Colombian female feminists encounter TF(GB)V.

4.1 Technology-facilitated (gender-based) violence

Before addressing TF(GB)V, all interviewees emphasized that both gender-based violence and violence against activists are alarmingly prevalent in their respective countries. This broad context of “normalized” violence forms the backdrop against which their digital experiences unfold. When using social media, all interviewees have reported having experienced at least one incident of TF(GB)V, with instances including (misogynistic) hate speech, bot attacks, doxing, hacking, and death threats (e.g., I1, I4, I13, I22). The indication that most female feminists have received threatening calls, texts, WhatsApp messages, and pamphlets suggests that their private information was accessible to different actors, including illegal armed ones, who are believed to utilize ICTs to target and surveil them. I9 shared: “We know that different armed actors may also be, let’s say, using technologies for tracking and so on”. Similarly, I8 and I0 reported feeling persecuted and under surveillance due to their efforts to publicly denounce alleged perpetrators of GBV at their work. This sense of pervasive surveillance is closely tied to how TF(GB)V is experienced differently from physical violence by some interviewees. I15 reflected on this distinction, stating that one of the most unsettling aspects of TFV is the anonymity of the perpetrators:

When it’s anonymous, when you don’t know who it is – it’s terrifying. when you don’t know who’s behind it, it’s overwhelming because you don’t know what’s happening. Psychologically and consciously, your mind creates an image – it can be the worst-case scenario, like thinking someone is already following you. You develop paranoia, the feeling of being watched. It becomes all-encompassing. In that sense, it can be even worse – a deep psychological impact.

The interviewees illustrate that these incidents have frequently targeted their physical appearance and employed derogatory language to discredit their public image and activism. Furthermore, numerous interviewees shared that they have received threatening messages intruding upon the most intimate aspects of their lives, including death threats, threats of sexual violence, and violence against their families (I9, I12–13, I19–21).^[7] For example, I4 shared: “I have received messages containing humiliating language about the way I look, comments about my body, my family, and my partner. […] These constant attacks are different from those directed at men. They reflect sexualized violence rooted in patriarchal society and machismo. That’s why I believe there are threats specifically targeting women’s bodies”. The interviewees explained that they are targeted by TF(GB)V because of their activism and causes that are often perceived as controversial in their home countries, where conservatives and patriarchal traditions prevail. I22 mentioned, “we do post things in support of women. We work in communities on women’s empowerment, right? But when it comes to visibility, we don’t like to expose ourselves, you know? Because of that same fear of digital violence – where just one comment can get you targeted, and it’s like … let’s say, it’s like being marked, stained with the mark of the devil”.

According to most interviewees, the fact of being a woman personalizes the threats they face, often manifesting as identity-based attacks. I9 shared that she had received numerous messages containing threats of sexual violence, including “threats to be raped”. Furthermore, I9 noted that traditional gender roles, which portray women as caregivers and mothers, make threats against their children particularly common and effective. Furthermore, in relation to intimate aspects of their personal lives, I9 further emphasized that attacks on female feminists target their

intimacy and bodies. And another aspect that we could say is a constant, differentiated attack from men is the issue of sexual violence, also by patriarchal society and macho violence. So, I believe there is a specific threat regarding the attack on women’s bodies. The messages are of this type of threat of being raped, of not being able to travel certain territories, certain spaces at certain times, about the control and life of women’s bodies that add to, let’s say, this patriarchal structure of violence against women, they add up or become visible in the digital scenario.

In addition to misogynistic discourses deeming women inadequate for assuming leadership roles and vocal positions in public discourse, entrenched racism has solidified into enduring sociocultural beliefs, contributing to heightened violence against women and BIPoCs in social media platforms:

[…] in the comments, one can perceive that they are more offensive […] when it comes to people with diverse bodies, when it comes to people with diverse sexualities, when it comes to black or Afro-descendant individuals. We see a greater attack when […] the diversity is accentuated (I20).

In essence, deeply ingrained racism and sexism, combined with longstanding socioeconomic disparities and institutional neglect, have exacerbated the marginalization of, amongst others, ethnic communities. In the Colombian context, regions rich in natural resources hold significant geostrategic importance for the territorial dominance of illegal armed groups, such as the National Liberation Army (ELN), and economic pursuits of extractive industries. ⁸² This creates fertile ground for female feminists to confront disproportionate levels of structural violence as they strive to transform the material realities of their territories from a gender-sensitive standpoint. While doing so, the provision of official support, for example by the government, appears to be limited, thereby engendering feelings of distrust in official institutions. I13 disclosed that the level of violence within her territory is at an alarming level and that there are a significant number of armed groups operating in the area. However, there is an absence of any form of early warning from the government, and the government is not taking any action. Furthermore, I9 shared that while the government’s protective measures are ineffective, they are particularly deficient in safeguarding women.

Now, it must be said that when it comes to protection, what we encounter in most cases with women is…let’s say, an inefficient government when it comes to responding to the protection needs of women. So, the entire government infrastructure is not adequately prepared to protect all women defenders, considering all the conditions that exist in the territories. […] It’s a very high level of distrust and a very high level of insecurity that also manifests in the digital sphere.

These testimonies exemplify the broader pattern of institutional distrust expressed by many interviewees. Furthermore, interviewee 11 noted that even some NGOs had proposed implementing panic buttons as a protective measure. While such tools were acknowledged as potentially useful in theory, she questioned their practical value due to widespread skepticism toward law enforcement. As she put it, activating a panic button seemed pointless: “What for?” – underscoring her belief that police intervention was either unlikely to occur or ineffective when it did.

4.2 Examples of chilling

Most interviewees described how such experiences of TF(GB)V impacted their everyday lives and that as a result they took measures to mitigate the associated harm. During the interviewees it became clear, that the lines between TFV and physical violence are frequently blurred and that individuals who experience TF(GBV) also often experience physical violence and vice versa. In instances where the danger of physical harm was present, interviewees reported having to relocate. For instance, I12 reported having to move multiple times after receiving online threats, driven by concerns pertaining to her own safety and that of her family. In other instances, interviewees decided to leave digital spaces or reduce their online activity to avoid experiencing further digital violence and to prevent this violence spilling into the physical world. For example, in two instances where interviewees decided to leave digital spaces, potential physical risks appear to have deterred them from maintaining the same level of online engagement as before. Despite the harassment occurring on social media, the threats extended to their inner circles, compelling them to take additional security measures and adopt self-censorship tactics to avoid being in the spotlight: “Even our female [colleague] would tell us not to meet on [location] or would tell us not to work on gender issues anymore” (I8).

In general, surveillance was identified as a potential factor contributing to some of the interviewees’ re-evaluation of their online activity. Another aspect that has led I13 to abstain from social media for several weeks was the emotional distress caused by a deepfake depicting her as being armed. Despite her consistent and peaceful opposition to extractive economies in the region, she received numerous messages labelling her as a “murderer” after the publication of the deepfake. Other reasons for disengaging from social media for several months were a series of racist and misogynist insults on Facebook (I19, I21). I21 shared:

About a month and a half ago, we had a case of harassment from a woman that started as a form of false exposure against the media, but mainly against me, in a very violent manner, and I think that changed my life forever with Twitter […]. I absolutely do not want to be public about anything […] I feel like I cannot do it as this woman took away that right from me.

This example illustrates that, in certain instances, activists may receive violent messages for a comparatively brief period before deciding to abstain from specific social media platforms. Conversely, in other cases, interviewees have reportedly received threats over an extended duration, ranging from months to years, before temporarily suspending their online engagement. This observation shows that the frequency or duration of threats made is not an objectively determining factor in the outcomes observed. The interviews do not indicate a clear trend but instead highlight that these decisions are highly individual, influenced by various external and internal factors, including the availability of support and whether close family members are also targeted.

The deterrent effects of TF(GB)V similarly affected female feminists who have not personally experienced threats to their physical integrity or that of their relatives. In this sense, some have opted to engage cautiously in online spaces, often adopting anonymous profiles due to the fear of encountering technology-facilitated (gender-based) violence and stigmatization and showing reluctance to express certain viewpoints. This hesitancy is rooted not only in the fear of TFV and stigmatization (I1, I4, I10, I13, I22), but also, for instance, in the context of Costa Rica, where abortion remains criminalized, in concerns about potential legal repercussions for guiding pregnant individuals seeking to interrupt their pregnancies. Overall, this apprehension makes it difficult to find women willing to assume prominent spokesperson roles: “[…] often people do not want to be spokespersons, they do not want to speak publicly about a topic, which is completely understandable due to the risk involved”.

As demonstrated by these example, certain incidents of TF(GB)V have resulted in analog consequences, such as relocation and the rejection of high-level positions, while others have opted to disengage from online spaces, seeking professional support (e.g., after a doxing incident (I10)), and increased skepticism toward certain (especially unfamiliar) social media platforms and digital technologies. For instance, I9 expressed a general distrust toward newer technologies due to a “general sense of distrust, especially toward social networks”. Following their exposure to TF(GB)V, most interviewees demonstrated some level of self-censorship and reduced their social media activity, particularly on platforms where they faced increased harassment. I19 mentioned: “because of the violence, I just withdrew from social media for at least four months. I censored myself”. Furthermore, three interviewees reported completely discontinuing the use of a specific platform (I0, I8, I10). For instance, I10 described how the violence she experienced on Twitter profoundly impacted her, stating that it “changed [her] relationship with Twitter”. She explained: “I absolutely do not want to be public about anything anymore. In fact, I was on a trip at the time, and sometimes when you’re traveling, you post pictures of where you are, but I haven’t posted anything at all since then.”

4.3 Examples of resistance

In this study, none of the interviewees mentioned withdrawing completely from participation within their activist group, they described solely a temporary decrease in participation and thoroughly examined long-term prevention and coping strategies at individual and collective levels. Despite the imminent risks inherent in their activism and the resulting physiological repercussions, some interviewees have forged alliances and campaigned for greater social mobilization, as their struggles are intimately tied to securing liveable conditions for their communities (I12–13). “Life goes on, life continues with or without threats; it is something we must deal with. We must continue defending rights. We must keep reporting and keep shedding light on what is truly happening in our territory” (I12). According to the interviewees, in the absence of effective institutional protection mechanisms and prevalent impunity, certain feminist groups within Colombia and CR have united in solidarity to establish preventive measures and capacity-building mechanisms for digital security customized to the needs of high-risk activism in these poorly connected regions (I7–8, I15, I17). For instance, the NGO to which I17 belongs has provided digital security workshops to female feminists, considering their available economic and infrastructural resources for implementing basic protective measures. Acknowledging these feminists’ strong reliance on free messaging services like WhatsApp, the workshops focused on enhancing security settings on the messaging app to prevent the unintentional sharing of sensitive information. Furthermore, female feminists within ethnic communities have shown a firm commitment to empowering their communities in their pursuit of social justice. Despite facing multiple racist attacks, I19 has launched campaigns and workshops to raise awareness about racism as a colonial legacy, aiming to gradually dismantle stereotypes surrounding Indigenous women and prioritize narratives that dignify Indigenous identity over those that exoticize the realities of these territories.

In short, the chilling effects of TF(GB)V have not occurred without evident resilience and resistance. Regardless of the context surrounding their activism, eleven interviewees have established partnerships with grassroots feminist groups and institutions advocating for gender issues at local, national, and international levels. I6 shared: “we are very much connected with [other feminist groups] all over Latin America. We are in numerous group chats with feminists from the south and the north”. These alliances encompass various activities, spanning from sharing information from other groups and forging media alliances (I0, I3, I10, I20) to coordinating joint demonstrations (I3, I14), collectively monitoring and mapping feminicides across the continent (I1, I12, I15), and engaging in political advocacy (I16, I17, I20). For instance, I16 describes her involvement in developing a TFV barometer during electoral campaigns, aiming to advance legislation on TFV. Additionally, this organization provides political guidance to empower community female leaders, who, according to their research, are particularly vulnerable to encountering insults and threats that aim to undermine their intellectual capacities and physical appearance with racist and misogynistic undertones. For many interviewees, these alliances hold profound significance, offering emotional support and solidarity amidst the shared challenges women face in Latin America. Partnerships with local women’s groups play a pivotal role in establishing comprehensive mapping systems for feminicides and GBV, amplifying the voices and experiences of Afro-descendant, Indigenous, and peasant women. This is particularly crucial in regions where mainstream and centralized media lack records of feminicide cases (I15, I20). Moreover, these networks facilitate capacity-building through close collaboration and knowledge exchange, aiding in preventing further violence in the face of widespread impunity (I8–9, I17). For example, one interviewee mentioned the development of technological safeguards to combat GBV in Bogotá (Colombia). This includes a mapping application designed to pinpoint high-risk areas for women, and a video game aimed at promoting awareness of street harassment.

The prevalence of impunity significantly contributes to the reliance of female feminists on individual self-care and collective coping strategies over institutional mechanisms. I19 both emphasized the role of protecting other women defenders, their territories, lives, and dignity, while also raising the question: “But then – who protects us?” This deficiency stems from various yet interconnected factors, as revealed in the interviews. Firstly, the interviewees highlight that there is a lack of state recognition regarding the extent of TF(GB)V. Consequently, no specific legal framework addressing TF(GB)V seems to typify its diverse forms and results in adequate legal approaches. Secondly, reports of TF(GB)V often seem fail to prompt thorough investigations due to insufficient technical infrastructure and capacity within public prosecutor’s offices. Additionally, personnel lack adequate sensitivity and training to grasp the seriousness of potential risks, leading to the dismissal of reports and further victimization. Thirdly, there seems to be insufficient accountability among media platforms regarding measures to identify and counteract content that may incite violence.

While the geographical context does not seem to determine whether female feminists form feminist solidarity alliances, it does appear to influence their ability to develop large-scale technological safeguards for enhancing digital security. The interviews reveal that there are often significant disparities between rural and urban areas. These are exemplified by extensive infrastructure and connectivity gaps, digital illiteracy and language barriers, and inadequate technological equipment among populations residing outside major urban centers. In line with this observation, capacity-building efforts, including digital security education and workshops in rural areas, are overwhelmingly undertaken by urban-based feminist organizations to bridge this gap. I16 shared: “We had to do a huge amount of digital literacy for the women who only knew how to use messenger chats and answer phone calls. Nothing else. Then you start dealing with other things, other platforms like Instagram. Also to understand the dangers of that.” Despite this tendency, the interview findings also highlight the readiness of female feminists, irrespective of their geographical context, to adopt individual self-care strategies using the resources available to them. Twelve of the interviewees implemented digital security measures at an individual level. These encompassed limiting discussions of sensitive topics to face-to-face interactions, controlling access to WhatsApp groups, using nicknames, declining calls from unknown numbers, refraining from sharing personal information, ignoring, and deleting offensive messages and opting for more secure platforms. Moreover, I13 mentioned constantly reporting offensive comments and messages on social media platforms and to official institutions: “I report. I always take screenshots and say ‘this happened to me this way’”. In addition to personal reporting, I19 described seeking assistance from an NGO to address violent messages she had received. However, the reporting of such incidents remains rare, largely due to fears of escalating violence, skepticism, the uncertainty of whom to denounce since many perpetrators remain anonymous (I21), and prior negative experiences. As I2 explained: “We are experiencing outrageous acts where trolls and men deliberately publish photos of feminists’ children – photos that reveal which school they attend. Even when this content is removed, nothing changes”. Furthermore, I2 shared that the social media platforms have not taken any measures to protect and defend her, nor to provide her with tools that could be useful in dealing with her “aggressor”. These testimonies highlight a pervasive sense of impunity: although specific harmful content may be taken down, there are no meaningful consequences or protective measures following reports. Such experiences contribute to ongoing frustration and reluctance to report incidents, underscoring the profound distrust in official institutions previously discussed in Section 4.1.

5.1 Contextualizing technology-facilitated (gender-based) violence against female feminists from Latin America

Social media platforms are often celebrated for fostering participation and serving as vital spaces for feminist advocacy and mobilization. ⁸³ However, our findings in line with earlier work by Bernal Sarmiento et al. ⁴¹ and Brydolf-Horwitz, ¹⁴ also reveal that these platforms can simultaneously function as sites of exclusion and (gender-based) violence. The rise of TF(GB)V and the reproduction of misogynistic structures online challenges the often rather optimistic view that social media “empowers” women to engage more actively. Empirical studies reveal that women and other frequently marginalized individuals are disproportionately targeted by TF(GB)V. ^{13 , 27} The prevalence of (misogynistic) hate speech, doxing, and death threats across the interviews suggest that TF(GB)V operates not as an isolated incident but as part of a broader, systemic pattern of online gendered repression aimed at silencing dissent and feminist advocacy. Interestingly, the interviews revealed no clear differences in the types of TF(GB)V or responses. However, this was not the primary focus of the study, and the findings should not be considered highly conclusive due to the limited sample size; they intended to provide an initial impression of the phenomenon and to lay the groundwork for future research. Numerous interviews noted that topics such as reproductive rights – including access to abortion, the fight against GBV – including feminicides, and queer rights are often seen as controversial and frequently provoke both online and physical attacks. The persistent levels of violence, along with ongoing public debates and restrictive legal frameworks, such as the criminalization of abortion in Costa Rica, reflects the continued influence of patriarchal and conservative structures across the region. While the specific laws and degrees of restriction vary, numerous Latin American countries share broader struggles against gendered power dynamics embedded in institutional and cultural systems. Overall, the prevalence of widespread misogyny and high levels of physical violence often intensify incidents of TF(GB)V, reinforcing Jane’s ¹⁰ assertions. The interviewees emphasized that the experience of TF(GB)V, as well as the responses to it are shaped by numerous context factors, which can be specific to the country but may also be more nuanced, relating to regional, communal, or even individual circumstances. Important influences to consider are, amongst others: (1) socio-political conditions (e.g., distrust in official institutions and social exclusion (e.g., Ref. 84]), (2) technological access and infrastructure (e.g., limited internet connectivity or digital literacy (e.g., Ref. 85]), (3) cultural norms (e.g., the persistence of strong traditional gender norms (e.g., Ref. 86]), and (4) legal frameworks (e.g., insufficient or poorly enforced regulations addressing TF(GB)V). For instance, widespread lack of trust in state institutions often discourages individuals from reporting incidents of TF(GB)V (e.g., I2). Many interviewees described their countries as having long-standing histories of state violence, impunity for crimes against women, and systemic patterns of institutional neglect (e.g., I4, I9, I19). Participants expressed concerns about victim-blaming, police inaction, and the perception that TF(GB)V is not treated as a serious offense by legal authorities. This institutional distrust is not only rooted in personal experiences but also reflects broader social patterns of discrimination and exclusion. Within such patriarchal structures, the normalization of TF(GB)V appears to be both a symptom and a consequence of this systemic failure to protect. Moreover, numerous interviewees stated that reporting can potentially increase the risk of harm.

Another important finding is that female feminist activists seem to especially prone to TF(GB)V due to the intersection of their activism and their gender identity. While engagement in (online) activism in general increases the vulnerability of individuals by increasing their public visibility, the controversy of feminist topics that feminist activists work on elevates this vulnerability even further. Numerous interviews noted that topics such as reproductive rights – including access to abortion, the fight against GBV – including feminicides, and queer rights are often seen as controversial and frequently provoke both online and physical attacks. Furthermore, however, some interviewees noted being targeted not only because of their activism and the topics they are fighting for but also due to their gender, or more specifically, the intersection of both factors. Explicit gender-related hate was demonstrated to be highly personalized (e.g., I9), targeted the female identity or the bodies of the activists, but also close family members, abusing the females’ roles as caregivers. This aligns with other empirical studies, which show that individuals advocating for or reporting on issues perceived as controversial are frequently targeted. ^{34 , 64 , 65} Additionally, numerous studies have highlighted that often marginalized individuals and groups, such as women and BIPoC, are often more affected by digital and physical violence. ^{45 , 67 , 87} This highlights the importance of recognizing the intersectional threats associated with TF(GB)V, such as racism and homophobia. Due to these intersecting forms of violence, it can sometimes be difficult to definitively determine whether a person is targeted because of their identity or the cause they are advocating for or because of both.

Additionally, as sometimes hinted at in the literature, this study strongly underlines the notion of the line between technology-facilitated and physical violence being blurred. ⁸⁸ Several interviews recounted experiencing TFV in the form of threats of physical assault. In some cases, those threats led not only to reactions in the digital sphere (e.g. blocking, chilling, etc.) but also to far-reaching consequences in the physical sphere, including relocation. Emotional reaction to digital violence, such as fear and distrust, have also been reported to stretch beyond the digital sphere, affecting those targeted in their everyday lives. This interconnectivity between the digital and physical sphere requires further attention in future research. As Chen et al. ⁴⁰ indicate, digital and physical violence should not be treated as one and the same in different spheres, as they present different challenges and require different methods of prevention and mitigation. Still, both forms seem to strongly inform each other and gaining a better understanding of this bilateral influence can help reduce violence in the future.

5.2 Shrinking secure spaces: consequences for political and social participation and responses to TF(GB)V

Building on previous (empirical) research on TF(GB)V and its consequences (see Section 2), our findings provide insights into how Costa Rican and Colombian female feminists respond to and cope with TF(GB)V. So far, this topic has received little (academic) attention, reflecting a broader gap in the literature, which often overlooks (female) activists – particularly female feminists and those from Latin America. Research on TF(GB)V targeting female feminists is crucial for understanding whether, as Jane ⁸ puts it, they are fighting, by e.g., developing coping strategies to resist TF(GB)V, remaining active online, and continuing their activism, or fleeing by consciously distancing themselves from certain digital spaces to protect their well-being,^[8] or engaging in both approaches simultaneously. In cases of self-censorship and withdrawal, digital engagement will be reduced. Over time, this constrained participation not only (further) minimizes feminist voices in online discourse but also undermines broader democratic processes by narrowing the diversity of perspectives in online public debate. In contrast to many other actors, activists typically operate in a rather personal capacity and rely on grassroots, peer-led support networks, such as feminist collectives, for, amongst others, solidarity and protection. This reliance reflects a broader systemic issue: the inadequacy of institutional and platform-level responses to TF(GB)V, which forces activists to assume responsibility for their own safety in the absence of formalized support.

By the time of the interview, social media and posting behaviors were often influenced by the TF(GB)V they had experienced, with some individuals withdrawing more than others. The interviews reveal that many participants have adopted both fighting and flighting mechanisms over time, with some even adopting both simultaneously (see Figure 1).

This supports the notion that fight and flight mechanisms are not mutually exclusive binary opposites but can instead, as Jane ⁸ demonstrated, coexist, inform each other and evolve over time. This observation is in alignment with the tenets of feminist HCI, which advocate for pluralism and the challenging of binary response models. ²⁶ The coexistence of both chilling and resistance responses demonstrates that female feminists are not a homogenous group but that they navigate incidents of TF(GB)V with nuanced, context-dependent strategies, building upon individual cost-benefit evaluations. Feminist HCI also acknowledges the complexity of digital agency and underscores the importance of reflecting on local circumstances. ²⁶ Notably, none of the interviewees mentioned completely disconnecting from their group, showing that, despite restricting their online behavior, they remained connected to other feminist group members, offering support. This suggests that collective identity and solidarity play an important role in this regard, as support networks function as both emotional anchors and political instruments, thereby facilitating the sustenance of resistance in the face of adversity. ⁸⁹ Moreover, the interviewees highlighted that resistance if often a reaction not only to the direct threat, rather is part of an ongoing historically rooted struggle against patriarchy, state neglect, and colonial power structures.

Some influencing factors mentioned were the personalization of threats, fear of surveillance and that someone closely related gets attacked (e.g., I15), the rural-urban divide, controversial topics, lacking regulation, and subjective emotionality. Drawing on Liang’s ⁴⁹ influencing factors, our results (illustrated in Figure 1) indicate that personal factors were the most frequently cited and that they have substantial impact on the interviewees’ reactions following incidents of TF(GBV). In comparison, social and platform factors elicited comparatively less attention. Thus, whether an individual continues to be vocal online depends on a range of individual psychological, social, and economic factors, leading to a subjective analysis of perceived risk to oneself and one’s friends and family. Interviewees consistently mentioned that they personally evaluate the costs and benefits to determine whether to maintain, adjust, or withdraw from their online presence. The degree of visibility and the issues they champion have been shown to influence the risks of violence that female feminist activists face, making them particularly susceptible to violent attacks and death threats. Additionally, regardless of context and geographical setting, personal (often embodied) threats and those directed at family members particularly provoke feelings of paranoia, persecution, fear, and helplessness (I12), often leading to stronger chilling effects, including self-censorship. This aligns with Nadim et al. (2021) and Penney, ⁵⁶ who posited that when threats are perceived as more personal, they are more likely to evoke feelings of fear and elicit behaviors such as avoidance, influencing how public discourse is accessed and who gets to speak. At the same time, these incidents have compelled some of our interviewees to adopt additional security measures irrespective of their specific circumstances, recognizing the potential for digital threats to escalate into physical violence and the concurrency of self-silencing and resistance mechanisms. The decision by some interviewees to partially or fully withdraw from (certain) online spaces or to engage in self-censorship, consistent with findings from prior studies with female (journalists) has significant implications for democratic engagement and activist participation. ^{12 , 34} The withdrawal of individual or communities, especially among marginalized groups undermines efforts to e.g., challenge oppressive systems and advance gender equality. ²⁸ Furthermore, it can facilitate the rapid dissemination of misogynistic and anti-democratic narratives. In instances where interviewees reported to continue being present online, the connection of their activism to personal stories (e.g., because someone close has experienced GBV) seem to have motivated them to take greater risks.

The ease with which TF(GB)V can be replicated, go viral, and remain anonymous compared to offline contexts (I21) causes profound mental and physiological harm, while also complicating efforts to proper accountability and reparation. Echoing Jane’s ⁸ fight responses, numerous female feminists resorted to individual protection mechanisms to navigate the risks associated with their activism, as the lack of support and protection from e.g., digital platforms and government compelled them to take measures for their own safety. Thereby, this study supports empirical findings revealing that many female feminists lack trust in institutional action, leading them to prioritize “personal and collective protective measures” over relying on “state-led safeguards” (p. 23) ¹⁸ . Concerning the collective sphere, our findings show that many interviewees generated an understanding of collective security and fostered alliances and partnerships with other feminist groups to join forces aimed at addressing common challenges. In his context, the strategic structures that many activists move within offer an advantage over less organized targets of violence. Networks and structures originally established for activist reasons thereby serve the additional function of a support and security network. Vice versa, newly formed support alliances in response to TF(GB)V can additionally be used for activist reasons. In accordance with Pedraza Bucio ²² and Abascal Mena and Pedraza Bucio, ⁹⁰ especially feminist groups based in urban areas particularly championed initiatives for digital capacity-building among rural-based activist networks. These efforts contributed to bridging the digital literacy gap and provided context-sensitive good practices and tools to enhance their digital security. Moreover, some of them engaged in digilantism practices to publicly expose alleged aggressors of GBV, particularly within work settings, as discussed by Jane. ⁸ Notably, the reactions were marked by relentless intimidation and harassment.

5.3 Reflections for HCI

While the interviewees already use various coping and security strategies, we argue that they should not bear the greater responsibility for ensuring their digital security and privacy: Social media platforms also play a critical role in reducing TF(GB)V and adopting an intersectional approach that recognizes how individuals with different identities, including gender, race, sexuality, or disability, experience TF(GB)V in unique ways. Instead of meeting this responsibility, misogynistic discourses that trigger or normalize GBV are often solidified into platform’s structures and institutional cultures, thereby overlooking the disproportionate challenges faced by female feminists in Colombia and Costa Rica, as they consistently advocate for feminist causes. This persistence of TF(GB)V reveals a systemic failure, amongst others by social media platforms, to address gendered violence (I2). On social media numerous affordances do not operate in isolation – they gain meaning through how people use them within specific social contexts. ^{29 , 42 , 43 , 91} Whilst numerous platform affordances, such as anonymity, visibility, communication, sharing, and automation, are frequently perceived as rather positive in HCI, the present findings demonstrate how these same affordances can be co-opted to sustain or intensify gender-based power imbalances. For instance, anonymity might enable users to harass women without accountability, while visibility and virality can be manipulated to shame or silence those who speak out. These technological features, when actualized in settings where misogyny and gender inequality is normalized, can facilitate various forms of TF(GB)V, such as cyberstalking or image-based abuse. Therefore, understanding affordances as socially and culturally situated helps explain why and how certain technologies may disproportionately harm often marginalized users, including women and gender-diverse individuals in patriarchal contexts. ^{48 , 92 , 93} Salamoun et al., ⁴² drawing on both feminist and affordance theory, stress the importance of considering the cultural, historical, political, and gendered dimensions of users’ lived contexts. They argue that affordances need to be analyzed not solely in terms of their technical functionality, but also in relation to the social power structures they both reflect and reinforce. In line with this, numerous critical and feminist HCI scholars underscore the importance of centering the experiences and needs of those most frequently excluded, such as women, queer communities, and other marginalized individuals across diverse contexts, including Latin America. ^{19 , 26 , 44 , 94 , 95} Thereby, they call for solutions that are not only technically effective but also addressing the users’ material and social realities. In relation to this, Sultana et al. ⁹⁵ explain that meaningful and responsible design in deeply patriarchal contexts requires working within existing power structures, rather than seeking to immediately overturn them, by enabling situated, context-sensitive interventions that respect local realties while still opening space for transformation. They propose ways to design within the patriarchal system, even if they wish to subvert it.

In sum, effectively addressing TF(GB)V requires a holistic approach, moving beyond technical solutions and towards systemic accountability, including platform-level responsibility. The failure to adopt intersectional and gender-sensitive design strategies reflects a broader pattern of exclusion in technological development. ⁹⁶ Although dismantling such discourses requires profound social changes, institutional and political reforms are essential to gradually transform the culture of impunity and recognize TFV as a critical issue. Suzor et al. ⁹⁷ argue that based on the United Nations Guiding Principles on Business and Human Rights (UNGPs), the platforms have responsibilities to prevent and counter TF(GB)V actively (provider responsibility). They list several steps platforms should follow in this regard: increasing transparency of internal processes, identifying impacts on human rights, developing policies to respond to abuse, and mitigating harm by enacting these policies. Additionally, they advocate for “social pressure” and “binding legal obligations” to hold the platforms accountable (p. 87), ensuring that they don’t merely “defuse the tensions without any real impact” on their daily business (p. 92). So far, some forms of TFV, such as hate speech, have been addressed through the application of natural language processing (NLP) and machine learning techniques aimed at classifying and automatically detecting abusive language. While these approaches show potential, they often fall short by overlooking critical contextual factors, socio-cultural dynamics, and linguistic nuances that shape how violence is enacted and perceived. ²⁷ Therefore, platforms should implement robust, multilingual and multimodal content moderation – a topic that is still highly discussed: While many urge for stricter content moderation and control, others, including Meta CEO Mark Zuckerberg, ⁹⁸ argue for loser content moderation policies, with a view to preserving “free speech” without restrictions. ⁹⁸ Several studies have demonstrated that a lack of fact-checking and insufficient content moderation increase (gender-based) violence, potentially causing individuals to withdraw from online spaces. ^{99 , 100} Besides this, it is important to acknowledge that content moderation practices on social media platforms have the potential to disproportionately impact marginalized groups, since “treating content and users as equal entities falsely presumes that their underlying identities and experiences are equitable” (p. 2) ⁴ . For instance, these practices may involve the censorship of representations of women’s bodies, or the misclassification of reclaimed language used by feminist or queer communities as hate speech. This is because automated moderation tools often lack the context to understand that some terms are used in empowering ways and not as attacks: For example, the word “slut” is commonly recognized as a derogatory term and flagged as hate speech by automated systems. But in some contexts, this term has been reclaimed and thus used intentionally by individuals or communities to challenge and subvert its negative connotation, transforming it into a symbol of sexual agency or resistance. Furthermore, sometimes, individuals who have encountered harmful content re-post it to raise awareness and draw attention to the problem. Subsequently, these posts are frequently removed because they violate platform content policies designed to limit harmful material. ⁴ , ¹⁰¹

Given that certain threats are conveyed through images, sometimes containing text that is not easily processed by automated tools like bots, it is crucial to develop effective strategies for handling such incidents. This challenge is particularly significant because the content often lacks overt abusive language, making detection more difficult. ¹⁰² In relation to content moderation, the question of what is classified as “harmful” emerges, as well as the extent to which largely automated moderation systems can address this, especially given the fact that moderation policies are commonly shaped within patriarchal structures. Given the prevalence of misogynistic narratives in our society, it is highly probable that numerous of instances of TF(GBV) go unreported or unflagged. Current community guidelines, which form the basis for moderation, need to be developed with greater sensitivity to GBV. Furthermore, as these guidelines are typically applied globally, they tend to overlook crucial differences across cultural, linguistic, and regional contexts. It is therefore recommended that platforms localize their guidelines, employ a feminist and culturally sensitive lens, and increase transparency around moderation decisions. The rationale behind takedowns or inaction should be clearly communicated to foster trust and accountability. ¹⁰² Moreover, inequalities in content moderation are often fueled by the absence of financial incentives for social media companies in less “profitable markets”. ¹⁰³ Here, we agree with Suzor et al. ⁹⁷ that the supposed conflict between free speech and preventing abuse online is a false dilemma. It overlooks how power dynamics, systemic discrimination, and abuse limit the agency and participation of affected individuals, making it impossible to engage in online spaces fully.

Empirical evidence demonstrates that numerous existing approaches are so far ineffective, underscoring the pressing need for policy overhaul and the development of new legal, legislative, and normative solutions that adhere to international standards of free speech and privacy protection. ³⁴ Existing efforts (e.g., the Online Safety Act 2021 in Australia) demonstrate that such gender-sensitive legislation on TFV holds promise for restoring trust in institutions and holding social media platforms accountable for either fueling hate speech, disinformation, and other forms of TF(GB)V, or for failing to prevent and address them promptly. ¹⁰⁴ In this regard, big tech companies bear significant responsibility for creating safe online spaces that foster women’s participation online. ⁴

In sum, as demonstrated throughout this section, addressing TF(GB)V requires coordinated, multi-level responses that engage diverse actors and strategies in parallel (see examples in Table 1). Only through intersectional, participatory, and contextually grounded approaches digital spaces can be made safer and more inclusive.