Full Library

Generative AI (GenAI) is evolving rapidly, and its use in violence against women research remains new and largely experimental. While researchers are beginning to explore GenAI’s potential, uncertainty around its risks, ethical challenges, and real-world application persists. Recognising the need for clear guidance, the Sexual Violence Research Initiative (SVRI) partnered with The MERL Tech Initiative (MTI) to develop this resource. This guide aims to provide background on GenAI, strengthen researchers’ AI literacy, and offer practical advice to help the violence against women research community critically and carefully engage with GenAI while minimising risks. We start off with an overview of GenAI and potential areas where researchers might use it at different stages of the research process. We highlight some low-risk, practical applications of GenAI that can help streamline tasks while ensuring that researchers remain in control. (Appendices provide clear definitions of key AI terms and a simple explanation of how generative AI works for those less familiar with the technology.) Following these examples, we discuss some of the key risks and potential harms of using GenAI for research on violence against women, including reliability, bias, dehumanisation, data privacy, power and patriarchy, relevance, transparency and replicability, inclusion and participation, worker’s rights, inequities and environmental harms. To support informed decision-making, we offer a simple decision framework with critical questions to consider before using AI for any task: • Does AI provide a clear benefit over existing methods? • Does its use comply with ethical guidelines, survivor consent, and data protection laws? • Are humans in the loop to validate results and take responsibility? • Can potential risks be effectively managed? • Is there a process in place for making and documenting decisions about the use of GenAI? If the answer to any of these is “no,” the guidance suggests reconsidering or avoiding AI in that context. This framework can help researchers to weigh AI’s advantages against its limitations, ensuring thoughtful and responsible use. Recognising AI’s fast-changing nature, we encourage a cautious approach to GenAI. We suggest starting small—testing GenAI tools on non-sensitive, publicly available data—to better understand their capabilities and limitations. If you choose to integrate GenAI into your research, we encourage you to document and share your experiences, so we can learn collectively and refine best practices together. The guide reinforces a key message: human expertise is irreplaceable. AI can improve efficiency, but final analysis and interpretation should always lie with researchers. By experimenting in a controlled way and keeping ethical principles at the forefront, researchers can build confidence in AI while avoiding unintended harm. This guidance is intended to be a living document—SVRI will maintain and update it as the AI landscape evolves and new insights emerge. SVRI and MTI are grateful for the critical feedback received from colleagues during the production of this guide, which has helped strengthen its content. (See Appendix B for details on the methodology used.)