activists. Generative AI should not substitute for activists; rather, it should serve as a tool that strengthens their capacities. To achieve this, activists must possess the skills and competencies necessary to use AI in a responsible and effective manner, and organizations should support them in developing their expertise, experience, and capabilities. Accordingly, this policy framework proposes that, where necessary, organizations may place limitations on the use of generative AI in the course of work—for example, by restricting its use in document drafting in order to support the capacity-building of early-career activists. This is not only a matter of individual capacity. Within an organization, it is essential to engage in discussion and deliberation on specific issues and to maintain a shared understanding and collective position. Generative AI must not be allowed to replace these processes. In particular, when drafting organizational statements or positions, the use of generative AI may weaken or substitute for internal discussion and collective reflection. For this reason, some organizations may choose to adopt policies that prohibit the use of generative AI for such purposes. Third, outputs generated by generative AI must not include any form of bias or discrimination against marginalized or vulnerable groups, nor negatively affect fundamental rights. The data used to train generative AI systems often reflect existing social biases, inequalities, and stereotypes, and AI-generated outputs may reproduce these patterns. The use of biased or discriminatory outputs can cause secondary harm to marginalized and vulnerable groups, while simultaneously 56 57