Empowering Change: Unveiling the Synergy of Feminist Perspectives and AI Tools in addressing Domestic Violence

Received: 18 October 2024; Accepted: 02 December 2024 Dra. María López Belloso. Faculty of Human and Social Sciences, University of Deusto, 48007 Bilbao, Spain; Universidad de Deusto, Spain [email protected] ORCID 0000-0003-4857-2967 Dra. Ainhoa Izaguirre Choperena Faculty of Human and Social Sciences, University of Deusto, 48007 Bilbao, Spain; Universidad de Deusto, Spain [email protected] ORCID 0000-0002-5781-9768 Dr. Borja Sanz Urquijo, Faculty of Engineering, University of Deusto, 48007 Bilbao, Spain; Universidad de Deusto, Spain [email protected] ORCID 0000-0003-2039-7773 Empowering Change: Unveiling the Synergy of Feminist Perspectives and AI Tools in addressing Domestic Violence Abstract: Gender-based violence remains a widespread issue in our societies. Women who are victims-survivors often encounter significant barriers when seeking support services, and frontline responders frequently lack the necessary skills and capacities to provide an adequate response. In this context, artificial intelligence (AI), particularly through the use and development of chatbots, has emerged as a potential solution to enhance and simplify access to these services for women. This is where the European project IMPROVE (Improving Access to Services for Victims of Domestic Violence by Accelerating Change in Frontline Responder Organisations) comes into play. Using a qualitative methodology, this study captures the voices of victim-survivors, exploring their views on the use of AI tools in the context of domestic violence, while also comparing these perspectives with the general societal perception of chatbots as reflected in media coverage. Keywords: Gender-based Violence (GBV), Artificial Intelligence (IA), Chatbots, AinoAid, IMPROVE European project, Feminist perspective Introduction In recent years, there has been a growing recognition of the role that Artificial Intelligence (AI) and digital technologies can play in addressing gender-based violence (GBV), particularly in the domain of domestic violence (Rodriguez et al., 2021; Kouzani, 2023). The intersection between AI tools and feminist perspectives offers a unique opportunity to both enhance the detection and prevention of domestic violence, as well as to critically assess the potential risks these technologies pose. Since the advent of AI in the mid-20th century, its exponential development and growing role in contemporary society have prompted thorough analyses of its attributes, ethical implications, and potential impact on human rights. One of the most critical aspects of AI, particularly relevant to fundamental rights and vulnerable groups, is the presence of biases. These biases can either perpetuate and exacerbate existing inequalities or introduce new forms of bias based on the discipline’s construction and data processing methods. Feminist critiques of AI have underscored the need to address biases in algorithmic systems and to develop tools that are sensitive to the lived experiences of marginalized groups, particularly women (Eckstein & Danbury, 2020; PenzeyMoog & Slakoff, 2021). This paper explores the potential for AI-based tools, particularly chatbots, to support women experiencing domestic violence while integrating critical feminist insights to minimize the risks of exacerbating existing power dynamics. The significance of this study lies in its focus on the practical and ethical applications of AI in the context of domestic violence. As the IMPROVE project (Improving Access to Services for Victims of Domestic Violence by Accelerating Change in Frontline Responder Organisations) demonstrates, there is a pressing need to create digital tools that offer tangible support to victims while avoiding the reinforcement of harmful societal norms (Novitzky et al., 2023). AI technologies, including chatbots and predictive algorithms, have been increasingly integrated into systems aimed at detecting and preventing domestic violence, yet their application has often lacked a critical gender lens (Ledesma, 2022). This gap underscores the necessity of aligning technological innovations with feminist frameworks to ensure that they do not inadvertently contribute to the reinforcement of patriarchal structures (Al-Alosi, 2020). Empowering Change: Unveiling the Synergy of Feminist Perspectives and AI Tools in addressing Domestic Violence While there are several studies that explore the role of AI in addressing violence against women, the literature often approaches the issue from either a security or public health perspective, with minimal focus on feminist epistemologies (Nowitzki et al., 2023). This paper aims to address this gap by critically examining the use of AI tools, specifically chatbots, within a feminist framework. We focus on how these tools can be refined to better meet the needs of women, particularly in terms of prevention, detection, and support. Drawing on the literature, this study emphasizes the importance of addressing biases in AI systems and ensuring that they are designed to empower rather than disempower women. The use of conversational agents or chatbots to support gender-based violence victims and front-line workers Rodriguez et al. (2021) explicitly identify AI as a key tool in addressing GBV, classifying its contributions into four categories: offline detection, education, security, and online detection. Notably, more than half of the studies reviewed by Rodriguez et al. (50.7%) focus on the online detection of violent content, reflecting the importance of AI in monitoring harmful behaviors, such as misogyny, sexism, grooming, and peer violence. This approach aligns with the concept of "symbolic violence," as analyzed by Ledesma (2022), who argues that AI, through its inherent biases, can reinforce existing structural inequalities. Symbolic violence refers to a structural situation in which unequal power relationships are maintained through explicit or implicit cultural norms, which, in turn, consolidate the status quo (Ledesma, 2022). Ledesma further asserts that AI systems contribute to the perpetuation of this violence by embedding biases that disproportionately affect marginalized groups, particularly women and girls. Belen Saglam et al.’s (2021) work provides a valuable foundation for designing chatbots to support domestic violence victims, emphasizing empathy, security, privacy, and the provision of practical information and emotional support. These tools can significantly reduce barriers to accessing support services and encourage victims to report abuse, provided they are carefully designed to meet victims’ specific needs and ensure their safety. Designing such tools requires prioritizing safety from the outset. Careful planning of conversational flows is essential to ensure that interactions are sensitive and appropriate. Advising individuals experiencing abuse on how to seek safety is a critical responsibility, requiring detailed attention and collaboration from multidisciplinary actors, including the victims themselves. Another significant challenge is the “humanization” of these tools. Alan Turing’s famous test evaluated whether a machine could “pass” as human, but this test is more complex than it appears. Passing as human involves considerations of race, gender, and social class. Chatbots are judged not only on their ability to use natural language patterns but also on factors such as friendliness and empathy, which are influenced by racial and gendered expectations. Users project racial and gender identities onto these conversational agents through visual, textual, and auditory cues. Therefore, as Hussain and Spencer (2024) argue, limiting the use of chatbots without human follow-up is problematic, as user assumptions about the chatbot’s identity affect their communication and responses. Thus, the question of AI and chatbots is not just about replicating human intelligence but also about understanding the racial and gender dynamics inherent in these technologies. Empowering Change: Unveiling the Synergy of Feminist Perspectives and AI Tools in addressing Domestic Violence As Sasha Constanza-Chock (2018) points out, design principles that adopt a universalist approach often exclude certain groups, particularly those facing intersectional disadvantages under systems like capitalism, white supremacy, heteropatriarchy, and settler colonialism. When technologists address inequality in design (which is rare in most professional design processes), they typically do so through a single-axis framework (Constanza-Chock, 2018:7). As a result, contemporary design processes are structured in ways that fail to account for or remediate the unequal distribution of benefits and burdens. This aligns with Kimberlé Crenshaw’s argument that feminist or anti-racist theory and policy that do not adopt an intersectional understanding of gender and race cannot adequately address the experiences of Black women and other people facing multiple forms of oppression (Crenshaw, 1989). This principle should also apply to the “design demands” for AI systems, including technical standards, training data, benchmarks, and bias audits. Given these limitations, it is necessary to incorporate feminist perspectives into the design and application of AI tools. Feminist AI (FAI) shows how “feminism” and “AI” have multiple meanings. As Sophie Toupin rightly notes, despite the singular term, FAI encompasses diverse contemporary manifestations of feminism, including intersectional, Black, decolonial, and liberal feminist approaches. Haraway’s concept of situated knowledge was groundbreaking in highlighting that knowledge production is embedded in social relations and that perspectives from marginalized positions offer the most objective accounts of the social world. Drawing on these ideas, computer scientist Alison Adam began to critique AI from a feminist perspective, exposing its conservative foundations and raising questions about how AI is used, what knowledge it represents, and what kinds of knowledge are utilized in these systems. The following section reviews several AI-driven chatbots that have been designed to address GBV, highlighting their strengths, limitations, and the feminist critiques that shape their implementation. Empowering Change: Unveiling the Synergy of Feminist Perspectives and AI Tools in addressing Domestic Violence Table 1: Taxonomy of GBV chatbots MAIN FUNCTION ADVANTAGES LIMITATIONSCHATBOT Sophia1Assist women victims of domestic violence Uses AI to provide support and also to securely store any evidence of abuse submitted by victims. Designed for domestic violence victims and may struggle to provide tailored legal advice across different jurisdictions due to its international scope. Hello Cass 2Provide information and support on GBV via SMS. Simplicity and accessibility with minimal technological requirements. Reliance on SMS technology; limited in areas with poor network coverage. MySis3 Emotional support and practical guidance for legal procedures. Detailed support on emergency services, legal guidance, and protection. Automated responses limit personalization; focused only on Thailand. Law-U Legal guidance for survivors of sexual violence in Thailand. Accuracy based on Thai Supreme Court cases; LINE platform integration. Lack of human interaction; reliance ONLINE limits accessibility. Violetta4 Preventive and psychoeducational support for Spanish-speaking women. Personalized responses supported by psychologists detects high-risk words. Reliance on technology; lack of human intervention in critical cases. Empowering Change: Unveiling the Synergy of Feminist Perspectives and AI Tools in addressing Domestic Violence Themis5Free legal advice with a gender perspective. Handles multiple requests; psychological support; generates policy data. Exclusive to Facebook; inaccessible without internet or social media. Sara6 Assistance in six Caribbean countries for women and adolescents. Adapts responses to cultural and linguistic contexts. Limited by technology access and understanding of complex interactions. rAInbow Inform victims about abuse signs through stories and quizzes in South Africa. Real-time data training; considers LGBTQ+ needs. Geographically limited focus; does not address all victim diversity. Agile Improves Access to Information and Services Immediate, anonymous, and confidential access; aimed at adolescents, young women, and sexual and gender minorities; Reduces traditional barriers to care; Utilizes a user-centered approach; Chatbot availability (24/7) Effectiveness depends on access to technology; Excludes individuals with limited resources; May not capture the nuances and complexities of each situation; Lack of direct human interaction; Privacy and security concerns over collected data NajatBot Guidance for Women and Girls Victims of Violence Accessible for free through Messenger on Facebook; Developed in the Moroccan dialect; Includes a persistent menu and quick responses Use of predefined sentences; Initial development phase; Lack of direct human interaction; Costly and labor-intensive maintenance Empowering Change: Unveiling the Synergy of Feminist Perspectives and AI Tools in addressing Domestic Violence AinoAid7 Assistance and Support to Victims of Gender Violence; Offers Information on Available Services Immediate and anonymous access to useful information and support services Advanced technology in AI, Machine Learning, and Natural Language Processing Need to interact with human professionals for more detailed and personalized follow-up Privacy and security concerns over collected data Costly and labor-intensive maintenance 1Sophia a Chatbot Tool to Empower Victims of Domestic Violence, Jun. 2020, [online] Available: https://www.kona-club.com/sophia. 2https://hellocass.com.au/ 3https://changefusion.org/initiatives/11kdhvc0ebab7mgr9d85rviwj9axan 4https://holasoyvioletta.com/ 5https://www.facebook.com/pg/AbogadasVioletas/posts/?ref=page_internal 6https://infosegura.org/en/news/sara-new-artificial-intelligence-tool-tackle-gender-violence-central-america 7https://ainoaid.fi/ Empowering Change: Unveiling the Synergy of Feminist Perspectives and AI Tools in addressing Domestic Violence The European IMPROVE project and the development of the chatbot AinoAid™: Innovating Support for Victims of Domestic Violence IMPROVE, funded under Horizon Europe, is grounded in evidence showing that many survivors of GBV are unaware of their rights and available services, leading to low reporting rates and limited help-seeking behavior. Studies, such as Simmons et al. (2011), highlight that most women in abusive relationships do not engage with formal support systems like shelters or hotlines. Additionally, marginalized groups, such as those in remote areas or stigmatized communities, are less studied due to challenges in identifying and accessing them. Frontline responder organizations also struggle to detect diverse victims and coordinate between agencies, despite the availability of training and guidelines. Vulnerable populations, including refugees, the elderly, minorities, and people with disabilities, are often under-detected and underserved, lacking equal access to services and justice. The project aims to empower these victims by raising awareness of their rights and available resources. In this sense, the AinoAid™ chatbot aims to enhance access to services for victims of domestic violence by addressing both the needs of survivors and the challenges faced by frontline responders. IMPROVE's overarching goals include increasing reporting of domestic violence cases, improving service accessibility for underserved victims, accelerating policy implementation, and fostering inter-agency cooperation through targeted training. The project focuses on marginalized groups, such as refugees, the elderly, and individuals with disabilities, ensuring that all victims have equal access to justice and services. AinoAid™ specifically uses conversational AI to help survivors navigate available services by offering assessments, guidance, and emotional support. The chatbot addresses barriers to reporting, such as anonymity concerns and fears of judgment. Drawing on prior technological innovations, such as Rainbow and Hello Cass, AinoAid™ directs victims to nearby service providers and community justice initiatives. The chatbot incorporates a validated risk assessment questionnaire and offers tools for victims to document experiences securely. The design of AinoAid™ is centered on user experiences, with conversations based on real-world interactions and professional expertise in gender-based violence (GBV). It is continuously updated through user interaction and collaboration with local organizations, ensuring its relevance and accuracy. The chatbot is designed to provide empathetic, reliable support while maintaining user privacy and data security. However, the project also faces challenges. Ensuring data accuracy, avoiding biases in input data, and protecting user privacy are critical concerns. Additionally, steps are taken to prevent potential misuse of the chatbot by perpetrators pretending to be victims. AinoAid™ aims to mitigate these risks through robust data security measures, careful supervision of AI training, and a phased rollout with extensive testing to ensure effective functionality. The chatbot operates anonymously, without requiring user registration, enhancing its accessibility and safety. Empowering Change: Unveiling the Synergy of Feminist Perspectives and AI Tools in addressing Domestic Violence Methodology This paper adopts a qualitative methodology. In addition to a review of the existing literature on the development of chatbots for the prevention of GBV from a critical perspective, this work seeks to contrast the official discourse on the potential of these technological tools for the prevention and support of past victims of GBV with the victims' real perception of their usefulness. Narrative interviews Within the framework of the IMPROVE project, narrative interviews to survivors of GBV were conducted in five different countries: Austria, Finland, France, Germany, and Spain. However, for the purpose of this study, only the Spanish interviews have been taken into consideration. The research followed a multi-stage process. Initially, a detailed mapping of associations and organisations supporting GBV survivors was conducted. Purposeful sampling was used to select participants, with priority given to organisations with whom the researchers had established relationships, particularly those working with vulnerable groups. The participant selection and interview organisation process was led by the psychologists or the social workers of the services to acknowledge their knowledge of the cases and build the participant’s confidence. Secondly, an interview script was elaborated based on a previous literature review that helped to identify the relevant dimensions and aspects that the interview should address. In total, 30 interviews involving women from various backgrounds, including elderly and migrant/refugee participants were developed in the Basque Country, Cantabria, Castile and Leon, and Madrid. Of these, 15 were conducted one-on-one (interviewer and interviewee), 2 involved the presence of a shelter social worker (one guided by two researchers, one of whom had extensive knowledge of the interviewee’s culture of origin), and 4 were held in a group setting with 2 to 4 participants and 2 researchers. This information is collected in Appendix 1. In the final phase, in-depth interviews adhered to WHO (2001) guidelines to ensure participant anonymity and safety. These safeguards included providing safe spaces, using intermediaries when necessary, and ensuring strict data protection. The research followed ethical standards on Gender, Ethical, Legal, and Societal aspects (GELSA) as required by the European Commission and approved by the University of Deusto’s Ethical Committee. News search Secondly, in order to analyze the general social perception of chatbots and specifically of AinoAid, a news search was carried out in different media. At first, we considered conducting a general search across various written media to analyze the general perception of chatbots and their use in the field of GBV. However, initial experiments showed a lack of specific news related to this issue, so authors decided to limit the analysis to press articles focused on the chatbot developed in IMPROVE: AinoAid. A total of 40 articles in Spanish and 2 in Finnish covering the launch of AinoAid were identified. Due to language limitations, only the articles published in Spanish were included in the analysis. These 40 articles (see Table 2) Empowering Change: Unveiling the Synergy of Feminist Perspectives and AI Tools in addressing Domestic Violence Empowering Change: Unveiling the Synergy of Feminist Perspectives and AI Tools in addressing Domestic Violence j) Could you please write a story about the preparation of the vacation for the wife and husband? The story depicts a couple preparing for a long-awaited vacation, presenting them as a partnership referred to as "they." However, it becomes clear that traditional gender roles are at play. The woman is responsible for packing and checking the suitcases, while the man takes on a more passive role, focusing on planning the places they will visit. This dynamic suggests that while domestic tasks remain in the woman's domain, decisions related to outdoor activities are handled by the man. Essentially, the man holds the decision-making power, symbolizing authority within the relationship. The woman, in contrast, follows the plans made by the man, taking on a more passive role. “Sophia tackled the packing, carefully folding clothes and adding sunscreen, while Alex researched local attractions and made a list of must-see spots.” As with other stories, positive language permeates the narrative, with expressions like "relaxation," "excitement," and "joy," creating an upbeat and pleasant atmosphere. Commonalities and Differences in Stories When attempting to write stories independent of gender, it was found that ChatGPT consistently adopts a heterosexual perspective. In narratives involving family or couples, if no gender was specified, the model defaulted to imagining the family as consisting of a man and a woman. Similarly, the children were created as a boy and a girl, suggesting that ChatGPT's concept of an "ideal" family is a unit of four—composed of a man, a woman, a boy, and a girl. Another recurring theme is the division of labor within the household. The kitchen is typically portrayed as the woman's domain, while responsibilities outside the home fall to the man. The woman is shown performing domestic tasks, caring for her family, and sacrificing her career when necessary. Conversely, the man is depicted as someone who handles work outside the home, maintains a job, and assists with household tasks when he has time. Men are generally characterized as strong, authoritative, and emotionally stoic, while women are portrayed as more vulnerable—often crying, waiting for apologies, and being emotionally delicate. In the stories, the man's needs are prioritized, while the woman is depicted as the caretaker who ensures that all family members' needs are met. Another observation is the use of third-person plural "they" to convey happiness and shared experiences within the family. However, when describing roles, the narrative shifts to third-person singular, emphasizing the gender-based division of labor. This distinction highlights the differing roles assigned to men and women. A further point of interest is the portrayal of the man as the head of the household. In one story, the man is described as the person who decides where to sit in front of the television. In another, following a fight, the woman packs her belongings and leaves the home. This reinforces the notion that men are responsible for work outside the home, while women are the ones expected to leave the house in the event of conflict. Empowering Change: Unveiling the Synergy of Feminist Perspectives and AI Tools in addressing Domestic Violence • Gender Violence Victims as “Subjects” of the News A consistent theme across the analyzed articles is the notable absence of women's voices, particularly the voices of the primary users of tools like AinoAid: victims of gender violence. None of the articles provide individual testimonies or opinions from women about the IMPROVE project or the chatbot itself. This omission suggests a lack of engagement with the very people these technologies are designed to help. The feminist standpoint theory emphasizes the need to center the perspectives and experiences of marginalized groups—here, women victims of violence—to generate situated knowledge (Cabrera et al., 2020). Unfortunately, the articles fail to meet this standard, omitting both individual and collective perspectives from victims and advocates. Instead, the articles focus heavily on the contributions of law enforcement, especially the Policía Local de Valencia, which is highlighted across all articles. This emphasis reflects a pattern in which public authorities, particularly law enforcement, are the primary “subjects” of the news. Their role is highlighted both in terms of specific individuals (invidual police officers involved in the project), and collectively as a force integral to the project’s implementation. Additionally, municipal agents supporting the local police’s participation are also mentioned in all articles, although their day-to-day involvement in the project remains unclear. In contrast, the articles provide minimal coverage of other stakeholders, such as the technology companies responsible for developing the chatbot. This further skew the narrative towards a technocratic approach that prioritizes institutional and technological actors over the experiences of women, the intended beneficiaries. For example, only 1 article discusses the contributions of WeEncourage, the company responsible for developing AinoAid. However, none of the articles include testimony from any direct victim who might be a potential user of the chatbot, nor from any victim support organizations, some of which are participants in the project. This is despite the fact that some of the news reports include the opinion of other officers highlighting the importance of ‘listening to women’ and maximizing the reach of care: ‘The first thing I thought of when I heard about the chatbot was all the women we can't reach and who don't dare to tell a police officer in a police station what is happening to them. The chatbot can guide them to take those first steps’, (El País, 2024) • Gender violence victims as “objects” in the news. All the articles analyzed address the problem of gender-based violence as a security problem, focusing their arguments for the need for the chatbot on the magnitude of this violence and the low rates of reporting. Only one article (El País, 2024) discusses more than one tool (more than one chatbot), analyzing the different functionalities of each. All articles mention the issue of underreporting of this type of violence, but they fail to cite direct sources or explore the reasons why women may not report. Twenty out of the 40 news articles analyzed mention the number of “80% of victims of gender-based violence” not reporting the aggressions, but they do not provide the source for this percentage. Empowering Change: Unveiling the Synergy of Feminist Perspectives and AI Tools in addressing Domestic Violence Additionally, one out of 40 articles notes that AinoAid is not exclusively designed to address the needs of victims but also provides support for potential aggressors. This places victims and perpetrators on equal footing, downplaying the severity of gender-based violence. “Aino will be prepared to dialogue even with aggressors, those people who engage in violent behavior, who do not know how to avoid it, who want to reflect on their behavior and ask for help’. (El Periódico, 2024) News Analysis from a Feminist Perspective As previously mentioned, the Fourth Wave of feminism centers the use of new technologies in its agenda, both to assess the potential contributions of cyber activism and to analyze the new forms of violence that these technologies can generate. Only 1 out of 40 analyzed articles addresses the potential of these technologies to tackle a social problem that disproportionately affects women. These articles frame gender violence as a societal issue but fail to explore why victims may hesitate to report their abuse or how they experience interactions with tools like AinoAid. Furthermore, only one of the articles mention feminism’s contribution to the fight against gender violence or its role in the development of new technologies that integrate victims’ needs, but limiting the acknowledgement to the funding provided: ‘The project receives international funding and is part of the Feminist AI Research Network (FAIR), a global network of scientists, economists and activists whose purpose is to make AI and related technologies inclusive and transformative. There are other initiatives related to, for example, digital gender-based violence - such as the Chilean chatbot SOF+IA - and harassment on public transport with the SafeHER app, designed in the Philippines.’ (El País, 2024) The articles overwhelmingly prioritize the needs and perspectives of law enforcement agents, reflecting a technocratic and institutional focus. This imbalance indicates that the media coverage largely aligns with the press releases provided by project stakeholders, as evidenced by the similarity of content across different outlets. In some cases, this content is replicated almost verbatim, further reducing the space for critical or alternative perspectives, particularly those from feminist organizations or victim advocacy groups. The articles analyzed correspond to the early stages of the project, which may partially explain the limited discussion of the tool's impact on victims. Nevertheless, from a feminist epistemological standpoint, the omission of women’s voices, both individually and collectively, represents a significant gap in the coverage. The over-reliance on law enforcement perspectives and the minimal engagement with victims’ experiences illustrate a missed opportunity to align the discourse with the feminist commitment to privileging marginalized voices and integrating their needs into technological solutions. Empowering Change: Unveiling the Synergy of Feminist Perspectives and AI Tools in addressing Domestic Violence Discussion The analysis of the public perception of chatbots in the context of GBV, specifically regarding the AinoAid chatbot developed in the IMPROVE project, provides crucial insights into the broader social understanding of AI-based tools for victim support. Several key themes emerge from this analysis, including the potential benefits and limitations of using AI in this sensitive area, the role of feminist epistemologies in evaluating these tools, and the social impact of such technologies in the framework of the Fourth Wave of feminism. Firstly, it is clear that while there is general support for the use of AI and chatbots as an initial resource for victims of GBV, the anonymity of the service is paramount for its acceptance. Many interviewees expressed a willingness to engage with the chatbot only if it provided complete anonymity, reflecting the deep concerns around privacy and safety, particularly for individuals in abusive situations. This finding is consistent with feminist perspectives that emphasize the need for technologies to be designed with the lived experiences and vulnerabilities of women in mind (Cabrera et al., 2020). The widespread lack of experience with chatbots, however, leads many to prefer human interaction after the initial engagement with AI. This dual preference for anonymity and human empathy highlights a tension between technological solutions and the personal, emotional needs of victims, underscoring the importance of personalization in AI responses. The perceived advantages of the chatbot, such as 24/7 availability, instantaneous responses, and the absence of moral judgments, suggest that chatbots can play a crucial role in providing immediate, non-judgmental support. However, the concern that chatbots might be “cold” and lack the empathy necessary for sensitive issues like GBV was a significant barrier for many respondents. This reinforces the feminist critique that while AI can offer practical assistance, it must be paired with the possibility of human follow-up to provide the necessary emotional and psychological support. Moreover, the analysis reveals a gap in public understanding and representation of the role of AI in combating GBV, particularly in the media. The feminist epistemological analysis of the 40 Spanish-language news articles revealed a lack of inclusion of women’s voices, either as direct users of the chatbot or through advocacy organizations representing victims. This absence speaks to a broader issue of gendered exclusion in technological narratives, where the perspectives of those most affected by violence are often sidelined in favor of institutional or law enforcement viewpoints. The media’s focus on security forces and public agents, rather than on the needs and experiences of victims, reflects the androcentric tendencies identified in feminist sociological critiques (Smith, 1979, 1987). This imbalance in media representation further supports the need for a feminist epistemology that centers the lived experiences of women and critiques the power dynamics inherent in knowledge production (Hartsock, 1983; Del Moral Espín, 2012). In this context, the Fourth Wave of feminism, which places digital activism and the use of new technologies at its core, offers both opportunities and challenges. On the one hand, the Fourth Wave’s emphasis on intersectionality, empowerment, and the fight against sexual violence aligns with the goals of tools like AinoAid. On the other hand, as noted by Looft (2017), the risk of trivialization through commercialization and celebrity involvement remains a concern. This tension between technological Empowering Change: Unveiling the Synergy of Feminist Perspectives and AI Tools in addressing Domestic Violence innovation and meaningful social change mirrors the challenges faced by feminist movements in balancing widespread visibility with substantive impact. The limitations of the chatbot, particularly in addressing the needs of non-native speakers or those under extreme control from their abusers, point to important areas for improvement. The interviewees’ concerns about trust in the technology, anonymity, and the chatbot’s ability to handle complex and diverse experiences of violence suggest that further refinement is needed to ensure these tools are truly accessible and effective for all users. Moreover, the reluctance to use platforms like WhatsApp due to safety concerns indicates that developers must prioritize security features that protect the privacy and confidentiality of users in precarious situations. Finally, the potential of the chatbot to raise awareness, provide self-identification tools, and offer educational resources about lesser-known forms of violence, such as psychological and vicarious violence, presents a significant opportunity for preventative and educational interventions. This aligns with feminist calls for greater recognition of the complexities of GBV and the need for tools that not only respond to crises but also contribute to broader societal understanding and prevention of violence. Therefore, while the AinoAid chatbot shows promise as an initial point of contact for victims of GBV, its effectiveness will depend on addressing the identified concerns around anonymity, personalization, and emotional engagement. The feminist epistemological framework highlights the importance of situating these tools within the lived experiences of women and ensuring that their development and deployment are informed by the needs and voices of victims. Moving forward, further research and refinement of these technologies are essential to ensure they fulfill their potential as tools for empowerment, support, and ultimately, the prevention of gender-based violence Conclusions and Future Steps The IMPROVE project, through the development of the AinoAid™ chatbot, has demonstrated the potential of AI to provide significant support to women victims of gender-based violence (GBV). By creating a new entry point to access support services, AinoAid™ helps break down historical barriers and empowers victims while raising societal awareness about the prevalence and impact of GBV. However, while the implementation of AI in this sensitive area is promising, it also comes with significant challenges that must be addressed to ensure its effectiveness and ethical use. Key among these challenges are the issues of data accuracy, ethical handling, and maintaining user trust. AI systems, such as AinoAid™, rely heavily on algorithms that must be continuously refined to prevent biases and provide fair and accurate assessments. These algorithms must align with a victim-centered approach, ensuring that they are culturally sensitive, confidential, and safe for users. The General Data Protection Regulation (GDPR) and the AI Act (COM (2021) 206 final) highlight the importance of transparency, traceability, and accountability in algorithmic decision-making. Ensuring that these standards are met will be crucial to fostering trust in AI systems and protecting the rights and safety of victims. Empowering Change: Unveiling the Synergy of Feminist Perspectives and AI Tools in addressing Domestic Violence Looking ahead, interdisciplinary collaboration will be critical for the continued success of AI tools like AinoAid™. This collaboration must involve not only technologists but also human rights experts, health professionals, feminists, and the victims themselves. Such an inclusive approach will ensure that AI systems are developed and implemented in ways that are relevant to the cultural and social contexts in which they operate. Moreover, partnerships with local organizations will ensure that AI tools are properly contextualized and responsive to the specific needs of different communities. Future initiatives should focus on strengthening these collaborations in the design and development of chatbots and AI tools. Future actions should also focus on improving public awareness of the capabilities and limitations of AI systems in addressing GBV. Ongoing professional training for those who interact with AI tools, as well as the development of user-friendly interfaces, will be essential to maximize their utility and accessibility. As AI systems continue to evolve, there must be a concerted effort to ensure that they adapt ethically to new situations and challenges, always prioritizing the needs and safety of victims. Additionally, to increase the accessibility and effectiveness of chatbots, it is necessary to develop multilingual and culturally adapted capabilities. These tools should be translated into different languages and their responses adapted to the specific cultural contexts of the users. This way, linguistic and cultural barriers can be eliminated, ensuring that the technologies are useful and relevant to a diverse audience. For the long-term success of chatbots and AI tools, it is essential to conduct regular testing, gather feedback from users and other stakeholders, and adjust the systems based on the results obtained. The evaluation should include metrics of effectiveness, user satisfaction, and analysis of any biases or limitations detected in the functioning of the tools. In sum, while AI holds immense potential to transform the support systems available to GBV victims, its development and deployment must be approached with caution and a commitment to ethical practices. Ensuring transparency, fairness, and collaboration across disciplines will be critical to overcoming the challenges associated with AI in this field. By doing so, we can create robust, inclusive, and effective technological solutions that provide real support to women as they seek to rebuild their lives in safety and dignity. Empowering Change: Unveiling the Synergy of Feminist Perspectives and AI Tools in addressing Domestic Violence References Adam, A. (1995) Artificial intelligence and women’s knowledge. Women’s Studies International Forum, 18(4), 407–415. Al-Alosi, H. (2020). Fighting fire with fire: Exploring the potential of technology to help victims combat intimate partner violence. Aggression and Violent Behavior, 52, 101376. Alvarez, P. (2021). A systematic review of computer science solutions for addressing violence against women and children. IEEE Access, 9, 114622-114639. Belen Saglam, R., Nurse, J.R.C., Hodges, D. (2021). Privacy Concerns in Chatbot Interactions: When to Trust and When to Worry. In: Stephanidis, C., Antona, M., Ntoa, S. (eds) HCI International 2021 - Posters. HCII 2021. Communications in Computer and Information Science, vol 1420. Springer, Cham. https://doi.org/10.1007/978-3-030-78642-7_53 Chamberlain, P. (2017). The feminist fourth wave: Affective temporality. Springer. Cochrane, K. (2013). All the rebel women: The rise of the fourth wave of feminism (Vol. 8). London: Guardian Books. Costanza-Chock, S. (2018). Design justice, AI, and escape from the matrix of domination. Journal of Design and Science, 3(5), 1-14. Crenshaw, K. (1991). Mapping the margins: Intersectionality, identity politics, and violence against women of color. Stanford Law Review, 43(6), 1241–1299. Del Moral Espín, L. (2012). En transición. La epistemología y filosofía feminista de la ciencia ante los retos de un contexto de crisis multidimensional. e-cuadernos CES [Online], 18. doi: 10.4000/eces.1521 Eckstein, J. J., & Danbury, C. (2020). What is violence now?: A grounded theory approach to conceptualizing technology-mediated abuse (TMA) as spatial and participatory. The Electronic Journal of Communication, 29(3-4). Haraway, D. (1988). Situated knowledges: the science question in feminism and the privilege of partial perspective. Feminist Studies 14(3): 575–599. Hartsock, N. (1983). Money, sex, and power: Toward a feminist historical materialism. New York: Longman. Hussain, H., Spencer, N.K. (2024). Chatbots and the Complexities of Delivering AI-enabled Support to Survivors of Gender-Based Violence. Social Innovations Journal, (23) Intemann, K. (2010). 25 years of feminist empiricism and standpoint theory: Where are we now? Hypathia, 25(4), 778-796. Kaba, M., Smith, A., Adelman, L., and Gay, R. (2016). “Where Twitter and Feminism Meet.” The Nation. Accessed April 17, 2020. https://www.thenation.com/article/where-twitter-and-feminism-meet/ Kouzani, A. Z. (2023). Technological innovations for tackling domestic violence. IEEE Access. Ledesma, J.O. (2022). Algoritmos y género: inteligencia artificial al servicio de la violencia simbólica. Revista Llapanchikpaq: Justicia 4.5, 209-236. Looft, R. (2017)."#girlgaze: Photography, Fourth Wave Feminism, and Social Media Advocacy." Continuum: Journal of Media & Cultural Studies, 31.6, 892-903. Novitzky, P., Janssen, J., & Kokkeler, B. (2023). A systematic review of ethical challenges and opportunities of addressing domestic violence with AI-technologies and online tools. Heliyon. Parry, D.C., Johnson, C.W. & Wagler, F. (2018). Fourth wave feminism: Theoretical underpinnings and future directions for leisure research. In D.C. Parry (Ed.), Empowering Change: Unveiling the Synergy of Feminist Perspectives and AI Tools in addressing Domestic Violence Feminisms in Leisure Studies: Advancing a Fourth Wave (pp. 15-26). London, England: Routledge. PenzeyMoog, E., & Slakoff, D. C. (2021). As technology evolves, so does domestic violence: Modern-day tech abuse and possible solutions. In The Emerald international handbook of technology-facilitated violence and abuse (pp. 643-662). Emerald Publishing Limited. Rodríguez, D. A., Díaz-Ramírez, A., Miranda-Vega, J. E., Trujillo, L., & Mejía-Saglam, Rahime Belen, Jason RC Nurse, and Lisa Sugiura. "Designing chatbots to support victims and survivors of domestic abuse." arXiv preprint arXiv:2402.17393 (2024). Silvestre Cabrera, M., López Belloso, M., and Royo Prieto, R. (2020). The application of Feminist Standpoint Theory in social research. Investigaciones Feministas, 11(2): 307–318. https://doi.org/10.5209/infe.66034 Simmons, C. A., Farrar, M., Frazer, K., & Thompson, M. J. (2011). From the voices of women: Facilitating survivor access to IPV services. Violence against women, 17(10), 1226-1243. Shiva, N., and Kharazmi, Z., Nosrat (2019). The Fourth Wave of Feminism and the Lack of Social Realism in Cyberspace. Cyberspace Studies 3.2, 129-46. Sinders, C. (2020). Feminist data set. Clinic for Open Source Arts. https://carolinesinders. com/wp-con tent/uploads/2020/05/Feminist-Data-Set-Final-Draft-2020-0517. pdf. Smith, D. (1979). A sociology of Women. In Julia Sherman y Evelyn Beck: The prism of sex (pp. 135-187). Madison: University of Wisconsin Press. Smith, D. (1987). The Everyday world as Problematic: a Feminist Sociology. Boston: Northeastern University Press. Socatiyanurak et al. (2021). LAW-U: Legal Guidance Through Artificial Intelligence Chatbot for Sexual Violence Victims and Survivors», IEEE Access, vol. 9, pp. 131440-131461, 2021, doi: 10.1109/ACCESS.2021.3113172. Toupin, S. (2024). Shaping feminist artificial intelligence. New Media & Society, 26(1), 580-595 Toupin, S., & Couture, S. (2020). Feminist chatbots as part of the feminist toolbox. Feminist Media Studies, 20(5), 737-740. WHO. (2001). Putting women first: Ethical and safety recommendations for research on domestic violence against women. Geneva: WHO (www.who.int/gender/documents/violence/who_fch_gwh_01.1/en Acknowledgements The authors of the paper would like to express their gratitude to all the women who agreed to participate in this research process by sharing their personal experiences of GBV. Also, to the professionals in the organizations who contributed to the participant identification and interview planification process. The authors would also like to thank professionals in the ASKABIDE association, for their help in contacting some of the women and also for conducting some of the interviews. They also appreciate the contributions received during the “I International Congress: Women, Technology and Power” held at the University of Deusto (Spain). This work has been supported by the IMPROVE project (Grant Agreement No. 101074010), and the project Inteligencia Artificial: Oportunidades y Retos para la lucha contra las violencias machistas, funded by BBK Kuna through the Kuna-co Proiektuak 2023 call. Empowering Change: Unveiling the Synergy of Feminist Perspectives and AI Tools in addressing Domestic Violence Appendix 1. Sample of Women Interviewed in Spain.