Canada Develops AI Tool to Predict Wildfire Hazards

Researchers at Mila and Natural Resources Canada created an open-source model to speed up fire mapping.

Updated on Oct. 23, 2026 in Artificial Intelligence

Isometric editorial illustration showing a geometric forest canopy with a heat hazard pattern, representing wildfire prediction technology. — Natural Resources Canada and Mila have launched an open-source AI model to improve the mapping and prediction of wildfire spread. AI Illustration. Upload story photo >

Live Poll

Do you trust artificial intelligence to accurately predict natural disaster hazards in your country?

Mila and Natural Resources Canada have launched an open-source AI tool designed to predict wildfire behavior, including burn likelihood and spread intensity. This new model aims to replace the existing BurnP3+ simulator to improve forest management.

Why it matters

As wildfire frequency and intensity rise, the federal government is prioritizing the development of faster, free, and open-source models to assist in hazard mitigation. This technology is essential for managing the increasing threat fires pose to communities across the country.

The AI model runs 800 to 3,400 times faster than the BurnP3+ simulator and can map fire hazards within hours instead of months. It was specifically trained using data derived from physical fire simulation maps recorded throughout 2026.

The players

Mila

Located in Montréal, this research institute specializes in artificial intelligence and machine learning development.

Natural Resources Canada

This federal department is responsible for the sustainable development and management of the country's natural resources.

Public Safety Canada

This federal agency coordinates the government's response to emergencies and disasters across the country.

The details

The tool provides real-time mapping of fire hazards, offering insights into burn likelihood and rate of spread. It was developed to help agencies better manage the significant increase in land loss and resident evacuations observed during the 2026 fire season.

Over the course of 2026, 5,400 wildfires burned 4.5 million hectares across Canada.
In October 2026, Public Safety Canada published an update on the wildfire season.

The Tech Race

The transition to this AI-driven model replaces the outdated BurnP3+ simulator to modernize how the country handles forest fire threats. This move represents a strategic push by the federal government to implement open-source solutions to outpace the escalating risk of climate-driven fire events.

The tool is designed to improve hazard forecasting, which may help authorities issue more accurate warnings to residents in high-risk areas. By mapping hazards in hours rather than months, it provides emergency services with faster data for planning evacuations and forest management.

The takeaway

The implementation of high-speed AI tools is essential for maintaining safety as fire patterns become more volatile. These technologies serve as critical upgrades to legacy systems that can no longer keep pace with the scale of environmental disasters.