A warning is only useful if it arrives. When a dam is released upstream at night, or a fire moves through a market at dawn, the last few kilometres are the hardest part of the journey, from an operations room to a phone, in a language the person holding it actually reads.

Ghana's National Disaster Management Organization (NADMO) now has a new way of covering that distance.

On 26 May 2026, at its headquarters in Accra, UNESCO handed over an AI-powered multi-hazard chatbot developed under the project "Strengthening resilience to water-related disasters under climate change for a sustainable society in Ghana", funded by the Government of Japan. The handover transferred the tool from project deliverable to national asset, owned and run by the agency that will depend on it.

Present at the ceremony were the Ministry of the Interior, the Embassy of Japan, the Ghana Meteorological Agency and the Ghana Hydrological Authority, alongside the Ghana Navy, the 48 Engineer Regiment, the Ghana National Fire Service and the National Ambulance Service, the services that will be reading the same warnings from the field.

One tool, seven hazards

The chatbot answers questions on floods, fires, storms, droughts, earthquakes, epidemics and localized environmental hazards. It offers safety guidance and weather advisories on request, delivers automated responses during an emergency, and supports several international and local languages, decisive in a country where a warning issued only in English reaches a fraction of the people who need it.

It is designed to hold open a channel that others lose exactly when demand peaks. During a flood, radio bulletins compete for attention and phone lines congest; a chatbot answers each person individually, at any hour, and keeps answering.

Training the regions to run it

A tool held at headquarters helps no one in a district. On 4 June 2026, officers from ten regional NADMO offices, Greater Accra, Ashanti, Eastern, Central, Western, Volta, Northern, Bono East, Upper East and Upper West, were trained online in both layers of the system: the interface the public sees, and the back end where content is configured and automated workflows are set.

The session reached beyond the IT department, bringing together officers from emergency operations, hydrometeorology, big data, communications, livelihoods and planning, so that the chatbot is run as an operational system rather than software owned by one unit.

Six regions, Oti, North East, Savannah, Bono, Ahafo and Western North, have no emergency operations centres yet, and will receive in-person training and the equipment to go with it. A public awareness campaign across radio, television, social media and community outreach will follow, since a chatbot only works once people know it exists. Training will then move down to district level.

The virtual format also exposed what the network still lacks: connectivity in remote regions was intermittent throughout the session, a reminder that digital early warning rests on infrastructure that has to be built first.

The chatbot advances Ghana's commitments under the Sendai Framework for Disaster Risk Reduction and the global Early Warnings for All initiative.