3. Guidelines for the Use of Generative AI 1) Verification of Information Accuracy Since outputs generated by generative AI may contain inaccurate information, their accuracy must always be verified through reliable means. ① Particular caution is required when using generative AI for tasks in which factual accuracy is critical. ② Facts should be cross-checked using multiple sources, such as internet searches and expert consultation. ③ Users should verify whether the data or materials are up to date. ④ Authoritative sources and official documents should be prioritized. ⑤ Where possible, preference should be given to AI outputs that reflect recent information (e.g., AI systems based on web search). ⑥ Clear and structured prompts should be used, and the AI should be asked to provide sources. ⑦ Caution is required when relying solely on summarization features without reviewing the original materials directly. By design, generative AI models predict the next word probabilistically based on the data on which they have been trained. In other words, AI does not determine whether something is true or false when generating a response; rather, it produces sentences that are most plausible within a given context. As a result, outputs generated by generative AI do not guarantee factual Generative AI Guide for Civil Society