Wildfire Response Bolstered by AI Camera Systems

AI bots are being used across the country to detect wildfires, often beating human 911 callers in spotting the blazes, according to The Wall Street Journal.

Climate change and development in vulnerable areas have increased the frequency and intensity of wildfires across the West and now in many states in the East, including recent fires in North Carolina, South Carolina, and New Jersey.

The Aha Moment

In 2003, scientists from UC San Diego set up the area's first monitoring camera after the devastating firestorm to monitor fires, mudslides, and other natural disasters. Over the years, they added more cameras to vulnerable areas so they could track what was happening.

As AI advanced, the researchers realized they could train the bots to analyze camera-video footage for fire signals like thicker haze and rising smoke, creating a preventative tool rather than just a monitoring one. In 2023, they tested a handful of these cameras and soon deployed them to all as the bots proved they could accurately detect fires, particularly at night when human detection is weakest.

How It Works

The first time this system was used was on December 4, 2024, when, at 2 a.m., AI-equipped cameras spotted a fire in an Orange County canyon, unbeknownst to the residents who were sleeping nearby.

The system alerted a human-staffed command center at UC San Diego, which verified and notified the Orange County Fire Authority. Helicopters and fire engines were able to get to the location quickly and keep the blaze to under a quarter acre.

Had the fire spread, the more than 1,000 homes just beyond the ridge would have likely been burned to the ground. Officials said it was the county’s first time AI detected a wildfire that no human observer had called in.

Using multiple cameras in a fire-prone area has also proven effective in pinpointing the exact location of a fire, as the different bots can triangulate the location for more precise coordinates.

The Current Stats

Over 1,200 AI bots are in use, mostly in California, with more scheduled for the rest of the country. These bots have detected 1,200 confirmed fires, beating out human 911 callers a third of the time.

Colorado is utilizing this system with 42 AI cameras installed by Xcel Energy between 2023 and 2024, and another 93 to be installed this year.

21 AI camera sites were contracted by Xcel Energy in Colorado in 2023, another 21 in 2024, and plans for 93 more this year.

56 buckets of water were dropped over nine hours by helicopters responding to a fire detected by AI cameras, preventing a potential disaster.

Concerns

Some residents in monitored areas have expressed concerns about intrusiveness and invasion of privacy. The cameras, however, are designed to blur identifying information like faces and license plates.

Another concern is the prevention of false alarms. Normally, when an AI bot detects smoke or heavy haze, a notification appears on command-center screens via text or email. To eliminate the possibility of false alerts, a staff member analyzes the video and determines the next steps. AI does not direct the action.

Firefighter unions support AI cameras but worry that other AI applications, such as pilotless helicopter drones, could endanger human-piloted aircraft working out the fires, emphasizing the need for careful implementation and human oversight.

The System Has Proven Itself Time and Again

One of the first flares detected by an AI bot/camera was near the town of Sonoma in 2023. As a result, it was quickly put out.

In Colorado, AI cameras spotted smoke after a lightning strike on a mountain above Denver. Multiple cameras triangulated exact coordinates. Two helicopters were immediately dispatched, dropping 56 buckets of water over nine hours. This kept the fire at bay until two dozen firefighters could contain it. The weather that day had been ideal for a rapidly spreading fire - 90°F and strong winds. The early alert prevented a potential disaster.

Multiply these types of events by 1200 (the number of fires AI bots have detected), and you get a clear idea of the success the system has had.

"AI is not going to put the fire out," says Phillip SeLegue, Cal Fire's staff chief of intelligence, "but the successes are the fires you don't read about in the newspaper."

Sources: MSN, Wall Street Journal, UC San Diego, CNN