GenAI Safety for Civil Society and Organizations Supporting Marginalized Populations: Framework, Evaluation Methodology, and Transparency Guidance

Title:GenAI Safety for Civil Society and Organizations Supporting Marginalized Populations: Framework, Evaluation Methodology, and Transparency Guidance (2025) Description:This bundle contains a set of integrated contributions developed by Ashley Khor (AI in Common) to support emerging practices in GenAI safety evaluation for civil society contexts, with a particular focus on organizations serving marginalized and vulnerable populations. Contents include: GenAI Safety Evaluation Framework (conceptual model) Safety Evaluation Recommendations (practical guidance) Safety Rubric for participatory and internal testing Transparency Hub structure for safety-related user communications This version is now publicly shared to support active collaborations and academic citation. This work complements a separately published Literature Review of GenAI and AI Safety Frameworks [https://doi.org/10.5281/zenodo.15576021], which provides background context for the evaluation and transparency approaches presented here. If you are an organization or researcher interested in using or adapting this work, or in contributing to its further development through feminist-aligned and participatory methodologies, please contact the author via AI in Common.