For an organization’s AI policy to function effectively in practice, it must be understood and implemented by its members. From the policy-development stage onward, it is important for members to engage in collective discussion, and regular training is essential, particularly in light of staff turnover and the onboarding of new members. To facilitate understanding of the policy and meaningful discussion about the need for updates, it is also helpful to include education on recent developments and trends in AI. To properly grasp fundamental issues in generative AI outputs—such as hallucinations and bias—some level of training on the technical characteristics of AI may also be necessary. Collecting and sharing case studies of problems that have arisen within or outside the organization (for example, instances of biased outputs) can further help members better appreciate and recognize these risks in practice. We hope that this guide will serve as a useful reference for internal training within civil society organizations. If an organization decides to adopt generative AI, it is undesirable for significant gaps in AI-related skills to emerge among staff members. For this reason, training on how to use generative AI may be necessary. There is no need to treat generative AI as something exceptional; rather, such training can be provided as part of the organization’s regular instruction on the use of tools required for day-to-day work. In some cases, an organization may choose to place policybased restrictions on the use of specific generative AI tools by certain members for a defined period of time. Properly assessing 98 99