the settings required to comply with this policy, and determining which functions should not be used. Decisions on whether to approve the use of a particular generative AI service should be made by an appropriate internal decision-making body, such as the steering or executive committee, and the organization’s AI officer should manage the approved list. This list may include information such as the name of the AI service, the provider, version, pricing plan, usage policy, and date of approval. Where the introduction of generative AI is likely to partially replace or significantly alter tasks previously performed by staff members, it is necessary to engage in prior consultation with those affected. Civil society organizations that place a high value on labor rights and human rights should approach the adoption of generative AI with these considerations in mind. Rather than unilaterally replacing the labor of staff members who previously carried out specific tasks, organizations should discuss what kinds of changes generative AI may bring, how human roles should be redesigned accordingly, and how the resulting burdens and benefits should be distributed. Even where the use of AI is expected to improve efficiency, there may be unforeseen issues or tasks that AI cannot replace. Routine or repetitive tasks may be streamlined through AI, while new roles can be created in response, or work can be reorganized around functions that only humans can perform, such as relationshipbuilding and other forms of interpersonal engagement. Generative AI Guide for Civil Society