Alberta's AltaLink and ATCO Electric are deploying AI-driven camera systems to spot wildfires near power lines. These tools, provided by ALERTWest, aim to reduce the risk of infrastructure-led blzaes in remote regions.

How a 30-minute response time prevented a multi-kilometre blaze

The practical utility of this technology was demonstrated during a pilot project run by ATCO Electric between July 2024 and February 2026 near Slave Lake, Swan Hills, and Whitecourt, Alberta. According to the report, one AI camera detected a wildfire start last June,allowing crews to reach the ignition site within 30 minutes. The fire was extinguished within an hour, a critical window that prevented a disaster.

ATCO Electric noted that a post-fire analysis indicated the blaze would have spread for several kilometres and likely become uncontrollable had it burned for just one more hour. This success has led the company to integrate the cameras into a broader, layered approach to wildfire management, moving beyond the initial testing phase to active operational use.

Scanning the Bow Valley from Kananaskis to Canmore

AltaLink has been integrating AI-powered detection since last June, specifically targeting high-risk zones in the Bow Valley. the utility placed cameras north of Kananaskis Village and west of Canmore, Alberta, identifying these as the most vulnerable points within its existing network. While AltaLink already operates 17 cameras in the Bow Valley, those previously required manual monitoring by staff to identify hazards.

The transition to AI allows for automated scanning of feeds, which triggers alerts for human review only when a potential fire is spotted. This shift is particularly vital because, as the report says, roughly 10 per cent of all wildfires in Alberta are linked to utility infrastructure, making the power grid both a potential cause and a primary victim of forest fires.

5,600 fires and the 360-degree scan

The technology provided by California-based ALERTWest relies on cameras that rotate 360 degrees every one to two minutes to maintain a continuous view of the landscape. These images are processed by an AI model trained on known wildfire visuals.. To prevent false alarms, ALERTWest staff review every detection before notifying utilities or fire agencies.

Scott Schifando, vice-president of operations at ALERTWest, noted that these systems have detected approximately 5,600 fires across the western United States over the last two years. This capability is essential for remote power lines where human presence is minimal and 911 calls may not arrive until a fire has already festered for a significant period.. Schifando emphasizes that the system provides live feedback to firefighters, improving real-time decision-making during an active blaze.

The 2022 test and the challenge of "smoke-like" clouds

Despite the technological gains, AI is not yet a replacement for human intuition. Mathieu Bourbonnais, an assistant professor at the University of British Columbia Okanagan, points out that algorithms still struggle with false positives, such as clouds or heavy truck exhaust that mimic the appearance of smoke. This remains a primary focus of his research into improving AI wildfire detection.

Furthermore, a 2022 test comparing AI cameras with human lookout observers found that humans were more successful at identifying wildfires. Because Alberta maintains one of North America's most extensive networks of staffed lookout towers,the utility companies view AI as a supplement. For instance, AltaLink noted that while a manned tower in the Bow Valley has only a partial view of transmission lines, the AI cameras provide the 24/7, infrastructure-specific monitoring that human towers cannot.