News reporting on feminicides and gender-related violence is critical for 1) framing this violence as a structural, systemic social problem, 2) as a means to keep specific cases in the public agenda to press for justice, and 3) as a source of data for activists monitoring this violence. However, when coverage lacks a gendered perspective, it can unintentionally perpetrate harm—for example, by re-victimizing survivors and families, obscuring structural causes, or contributing to state negligence and public inaction.
Data Against Feminicide is collaborating with Data in Society Collective (DISCO Lab) at Brown University on a project that explores how collaborative news annotation and artificial intelligence (AI) tools might support more responsible, feminist reporting of cases of gender-related violence. Working closely with feminist activists, journalists and scholars across cultural contexts, our project has three objectives:
- To co-develop taxonomies of both harmful and constructive journalistic practices that are sensitive to different languages and regions.
- To experiment with collective and collaborative data annotation, in contrast with conventional, extractive processes for creating datasets for AI training.
- To ideate and co-design an AI tool to promote responsible feminicide news coverage.
This project is a continuation and a deepening of the feminist and participatory methods for the co-design of machine learning tools explored through Data Against Feminicide. Currently, we are working with 45 participants from 25 countries. Through a participatory research process, we collectively read and annotate news articles to build a repository of harmful and constructive journalistic practices in feminicide reporting across contexts. Using this dataset, we will also ideate and co-design potential AI tools to support the production of feminist and contextualized reporting, from a writing assistant that flags problematic framings to an interactive tool that empowers news readers and activist groups to audit and challenge news coverage.