Microsoft’s HASTE turns satellite images into disaster-response maps within hours

Microsoft says its open-source HASTE system is helping emergency responders identify damaged areas, blocked routes and likely aid priorities within hours rather than the days or weeks a manual assessment can take. The practical point is simple: faster maps can help people decide where to send scarce rescue and relief resources first.

Microsoft AI Watch analysis

What happened

HASTE, part of Microsoft’s AI for Good programme, analyses before-and-after satellite imagery to identify damaged buildings and other clues about a disaster’s footprint. Microsoft says Catholic Relief Services used it after the June 2026 earthquakes in Venezuela, producing a damage map by the next morning and adapting the system to assess roads, bridges and possible aid routes.

The company says the tool has supported responses to 23 major disasters in 16 countries since 2023. In Venezuela, the resulting analysis was shared through a humanitarian network used by governments, charities and emergency-response teams. Microsoft also says responders used HASTE after wildfires in Spokane, Washington, to estimate how many households might need support while the fires were still burning.

Why it matters

Disaster information often arrives as a messy bundle of field reports, messages, news coverage and incomplete observations. Satellite imagery gives responders another layer of evidence, while machine learning can scan much larger areas than people can inspect manually. That can mean fewer teams sent down the wrong road and earlier estimates of shelter, food and equipment needs.

The system is not a magic damage oracle. Microsoft’s own account stresses that accuracy matters as much as speed, and disaster scenes vary considerably. HASTE needs people to train it for a particular situation, and its maps still need to be interpreted alongside local knowledge and other sources.

Our read

This is one of the more convincing uses of AI because the payoff is concrete: not a shinier dashboard, but a faster answer to “where should we go?” The open-source model matters too. Smaller organisations can adapt the tool instead of waiting for a bespoke system or a large technology contractor to arrive after the useful window has narrowed.

The sensible test is whether these maps improve decisions on the ground, not whether the model sounds impressive in a product story. Speed without reliable confidence information would merely produce wrong answers at impressive velocity, which is not the sort of innovation anyone needs in a collapsed neighbourhood.

What to watch

• Accuracy in new disasters:
  whether local teams find that HASTE remains dependable when terrain, building styles and damage patterns change.
• Open-source uptake:
  whether charities and public agencies adapt the tool for smaller emergencies without Microsoft’s direct involvement.
• Route planning:
  how often damage maps help responders avoid blocked roads and reach harder-hit communities.
• Human oversight:
  whether local knowledge remains central when satellite evidence is incomplete or ambiguous.

Discussion spark: Should emergency agencies make open-source AI damage mapping a standard part of disaster response, or is the risk of fast but misleading assessments too high?

not affiliated with or endorsed by Microsoft