<![CDATA[The development of digital technology has brought significant changes to the dynamics of public communication, with social media becoming a primary platform for government criticism. The controversy surrounding the release of Indonesia’s 2024 poverty data by the Central Bureau of Statistics (BPS) sparked a negative response, challenging the legitimacy of public institutions. In this context, accurate, real-time monitoring of public sentiment is crucial for supporting crisis communication responses. However, comparative studies on the reliability of free AI-based sentiment analysis tools in the public sector are still limited. This study evaluates the effectiveness of ChatGPT, DeepSeek, and Parabola.io in building an early warning system for crisis communication. A quantitative-descriptive approach was used to analyze 351 comments from the Instagram account @bps_statistics related to the release of poverty data. The results showed that DeepSeek (82.93%) and Parabola.io (82.62%) had higher accuracy than ChatGPT (40.24%). These findings underscore the importance of AI reliability in diagnosing crises early in the issue cycle. Theoretically, this study develops a crisis communication framework by integrating sentiment analysis and the issue cycle. Practically, this research emphasizes the need to consider accuracy, bias, cultural context, and human oversight in the use of AI.]]>
Copyrights © 2026
