of using these tools, and therefore must exercise caution. As with personal data, outputs generated by generative AI—particularly music, images, and videos—may incorporate or closely resemble copyrighted works that the AI was trained on, giving rise to claims of copyright infringement by the original rights holders. In such cases, and irrespective of the liability of the AI provider, the user who generated and used the output may also bear responsibility for copyright infringement. This may apply even where the user was unaware of the similarity to a copyrighted work or had no intention to infringe. Accordingly, to prevent harm to the organization’s credibility and to avoid legal disputes, users of generative AI should take care not to inadvertently infringe on the copyrights of others. In particular, prior review is essential when AI-generated outputs are used publicly, such as for organizational communications or campaign materials. To this end, it is important to check whether there are existing works that are identical or similar to the output generated by generative AI. This can be done by conducting internet searches or consulting relevant copyright databases. Textual outputs can be verified by searching specific passages through search engines, and images can likewise be checked using reverse image search tools. Another way to reduce the risk of copyright infringement is to use generative AI outputs only as a starting point and then substantially revise, adapt, or edit them through human effort. While outputs generated solely by generative AI are generally not protected by copyright, the addition of meaningful human creative input may Generative AI Guide for Civil Society