Humanitarian communication is a critical component of effective crisis response, enabling the coordination of relief efforts, timely dissemination of information, and engagement with affected populations. Although digital tools such as mobile platforms, social media, and data systems have improved communication reach and speed, challenges remain, including information overload, fragmented and siloed data, language and cultural barriers, limited situational awareness, misinformation, and unequal access to technology. Artificial Intelligence (AI) presents significant potential to address these challenges by enabling real-time data analysis, adaptive messaging, multilingual engagement, personalised risk and recovery communication, and participatory feedback mechanisms. This study employs a qualitative desk review of peer-reviewed literature, organisational reports, policy documents, and case studies to examine the role and challenges of AI in enhancing humanitarian communication. Grounded in the Constitutive Communication (CCO) framework and Participatory Communication Theory, the findings indicate that AI-mediated communication not only improves operational efficiency but also constitutes new organisational practices and decision-making structures while promoting inclusive, community-centered engagement. To address the challenges identified, recommendations include ensuring data quality and representation through inclusive data collection and bias audits, implementing robust ethical oversight and data governance policies, and maintaining human judgment through training, cultural sensitivity, and human-in-the-loop decision-making. These strategies aim to support the responsible integration of AI, enhancing effectiveness, accountability, and equity in humanitarian communication and strengthening crisis response outcomes.