For a generative AI policy to be coherent, it is essential to clearly articulate the principles on which it is based. This policy framework proposes six core principles. First, full responsibility for all outputs produced using generative AI, as well as for any decisions made on the basis of those outputs, rests with the organization. No matter how much autonomy an AI system may appear to have, responsibility cannot be assigned to a tool. While it may be possible to raise issues with AI developers in cases where problems arise due to technical flaws in a generative AI system, the primary responsibility lies with the organization that used the output. Accordingly, each organization must establish procedures to ensure that it fulfills its responsibilities as the accountable actor using generative AI throughout the entire process of use. For example, outputs generated by generative AI should always be reviewed under the organization’s responsibility, and internal procedures should be in place to determine how to respond if problems arise. This principle also means that, at every stage of using generative AI— even when AI-generated outputs are used largely as they are—final judgment and oversight must remain under human responsibility, specifically that of the organization’s activists or staff. Second, generative AI is merely an auxiliary tool and must not replace the judgment or expertise of activists. This principle is closely linked to the first. The primary assessment of outputs generated by generative AI must be carried out by the organization’s Generative AI Guide for Civil Society