As public health emergencies unfold, communities often speak first through digital platforms. During COVID-19 in Tanzania, thousands of social media posts carried signals of public concern, awareness gaps, misinformation and trust dynamics that were not always visible through routine surveillance alone. Drawing on Twitter data from Kinondoni District in Dar es Salaam, this study shows how AI-enabled infodemiology and social media analytics can complement formal public health systems by turning unstructured online conversations into timely intelligence for risk communication, community engagement and pandemic preparedness. The findings demonstrate that, with the right safeguards, artificial intelligence can help health authorities and development partners listen better, respond faster and strengthen digital health resilience in Tanzania and beyond.