Teen Rescued After AI Gives Dangerous Hiking Route

Published: 09 October 2026. The English Chronicle Desk. The English Chronicle Online

A 16-year-old boy in British Columbia, Canada, had to be rescued by a specialist mountain rescue team after following inaccurate hiking directions generated by the artificial intelligence chatbot Claude. The incident has renewed concerns about the growing reliance on AI tools and digital mapping applications for wilderness navigation, particularly in mountainous areas where incorrect route information can quickly become a serious threat to life.

Bryce Vincent Gowryluk, an enthusiastic indoor climber, had set out to hike Crown Mountain near Vancouver last Saturday. He reportedly asked Claude to plan a return journey from Grouse Mountain to the nearby summit, a destination known for its spectacular views across Greater Vancouver, the Fraser Valley and the Capilano watershed. Instead of following a suitable hiking route, however, the teenager found himself at the base of a steep rock face associated with a demanding climbing route known as Widowmaker Arete.

The route he encountered was far beyond the requirements of an ordinary mountain hike. Widowmaker Arete rises approximately 1,700 feet and involves exposed rock, steep sections and technical climbing. Route descriptions indicate that although parts involve relatively manageable slab climbing, other sections can require considerable experience and specialist equipment. Ropes, climbing protection and a carefully prepared ascent plan are essential for tackling the more difficult terrain safely.

Gowryluk had not set out to undertake such a climb. As he moved through the mountain landscape, the directions he received led him away from his intended route. After crossing steep scree slopes, gaining more than 2,000 feet in elevation and navigating through rocky terrain, he travelled down Crown Pass and continued through Crater Slabs towards a boulder-filled valley.

At some point during the journey, the route suggested by the chatbot took him towards the final headwall of Widowmaker Arete. As he attempted to make progress, the terrain became increasingly steep. He eventually realised that he had reached a cliff face he could not safely climb and was unable to continue along the path he had expected to follow.

Recognising the danger, the teenager contacted his mother before calling emergency services. Police were notified and coordinated with North Shore Rescue, the specialist volunteer organisation responsible for mountain search-and-rescue operations in the region. A helicopter was dispatched to locate him and assist with the evacuation.

The rescue proved challenging because Gowryluk was positioned too close to the rock face for the helicopter to approach safely with its rotors. Rather than attempting a direct extraction, rescuers deployed two members of the team by hoist. The rescuers reached the stranded teenager and helped lift him to safety before he was flown to the rescue team’s base near Cleveland Dam.

The operation ended without a reported serious injury, but the circumstances highlighted how quickly a seemingly ordinary hiking trip can turn into an emergency when navigation information does not accurately reflect the terrain. During a subsequent debrief, members of the rescue team said the teenager had ended up far from the route he intended to follow.

Although the incident is believed to be among the first reported cases in which a hiker became stranded after relying on AI-generated directions, it is part of a wider pattern of navigation-related problems in southwestern British Columbia. The region attracts hikers with its dramatic peaks, forested trails and panoramic viewpoints, but its mountainous landscape can be unforgiving for people who underestimate distances, elevation changes or technical difficulties.

In July, three people were rescued after using Google Maps to plan a journey along the Howe Sound Crest Trail. According to reports of that incident, the application provided a highly unrealistic estimate of the time required to complete the route. A journey that can take as long as 14 hours was presented as achievable in slightly more than five hours.

Such cases demonstrate that digital navigation tools can create a misleading sense of certainty. A route displayed on a screen may appear straightforward, but the real-world conditions can involve loose rocks, exposed ridges, steep ascents, difficult descents and sections where a person cannot safely turn back. Even a small navigational error can have serious consequences when a hiker enters terrain that requires specialist skills.

Paul Markey, a search manager with North Shore Rescue, said people should not depend on artificial intelligence as their primary source of mountain route planning. He stressed that preparation, experience and an understanding of the environment remain essential for anyone heading into the wilderness.

Markey noted that Gowryluk had carried several useful items, including a headlamp, suitable clothing, sturdy boots, a spare battery and additional food. Those provisions helped make him reasonably prepared for hiking and scrambling. However, the equipment he had brought was not sufficient for the technical rock climbing terrain where he eventually became stranded.

The distinction is important because hiking, scrambling and technical climbing require different levels of preparation. Hiking generally involves travelling on established paths, while scrambling can require using hands to negotiate steep or rocky sections. Technical climbing may demand ropes, protective equipment, specialised techniques and a detailed understanding of the route. Someone equipped for one activity may be dangerously unprepared for another.

Markey also questioned the ability of AI systems to provide reliable descriptions of specific mountain conditions. He explained that Claude had not physically visited the area where the teenager became trapped and could only provide information within the limits of the data and descriptions available to it. A chatbot may produce convincing directions without being able to verify whether a suggested route is passable, safe or appropriate for the user’s experience.

The incident raises broader questions about how people interpret information supplied by artificial intelligence. Modern chatbots can generate detailed itineraries, describe landscapes and answer follow-up questions in natural language. Their confidence and fluency can make recommendations appear authoritative, even when the underlying information is incomplete, outdated or inaccurate.

For wilderness navigation, the difference between a plausible answer and a verified route can be critical. AI-generated directions may fail to account for temporary trail closures, seasonal conditions, unstable slopes, snow, difficult terrain or the actual technical requirements of a route. Unless the information is checked against reliable mapping resources and current local guidance, users may have little reason to recognise that a suggested journey is unsafe until they encounter the danger themselves.

Digital mapping applications have also changed how people prepare for outdoor activities. Instead of studying topographic maps, learning to use a compass or consulting experienced hikers, some visitors increasingly depend on electronic devices to determine where they should go. These tools can be useful, but they can also encourage users to follow a route simply because it appears on a screen.

Markey warned that some applications identify routes in areas where the rescue organisation would prefer people not to travel. When a path is displayed digitally, users may assume it is an established trail suitable for general access, even when the terrain is difficult, poorly maintained or inappropriate for inexperienced hikers.

His warning points to a wider challenge for technology companies and outdoor enthusiasts alike. Mapping systems need to communicate uncertainty clearly, while users must understand that a digital line on a map does not guarantee the existence of a safe, maintained trail. Accurate distance estimates alone are also insufficient if an application fails to account for steep gradients, exposure or technical obstacles.

Mountain safety specialists recommend preparing for journeys with reliable topographic maps, checking official trail information and studying the route before departure. Hikers should understand the expected distance, elevation gain, weather conditions and technical difficulty, while carrying suitable equipment and a backup navigation method. Informing someone about the planned route and expected return time can also help rescuers respond more effectively if a person fails to return.

Electronic devices remain valuable when used appropriately, but batteries can run out, mobile coverage can disappear and digital directions can be wrong. Traditional map-and-compass skills provide an additional layer of safety when technology fails or produces misleading information.

For young adventurers in particular, the incident is a reminder that enthusiasm and confidence cannot replace preparation. Experience in an indoor climbing gym does not necessarily translate into the skills required to navigate exposed mountain terrain, where weather, loose rock and the consequences of a fall create very different challenges.

Gowryluk’s rescue ended with him being brought safely off the mountain, but it also provides a warning for anyone considering using a chatbot to plan an outdoor adventure. Artificial intelligence can help people research destinations and organise trips, yet it cannot replace sound judgement, verified local knowledge or practical navigation training.

As digital tools become more deeply embedded in everyday decisions, their limitations matter just as much as their convenience. In remote and dangerous environments, following a confident but incorrect answer can have consequences far beyond an ordinary online mistake. The message from mountain rescuers is clear: technology can support preparation, but it should never be treated as a substitute for experience, proper equipment and responsible decision-making in the field.