AI Chatbot Advice Leads to Multiple Mountain Rescue Calls Across Continents

A 16-year-old climber named Bryce Vincent Gowryluk was trapped on a near-vertical ledge of the Widowmaker route near Vancouver after following route advice from Anthropic’s Claude chatbot. North Shore Rescue deployed a helicopter to extract him, noting that Claude had recommended a scramble over loose rock that led him onto a multi-pitch climb requiring gear and partners. This rescue is the latest in a series of AI-related incidents, including two hikers in British Columbia, a pair in Poland’s Tatra Mountains, and three novices on Mount Shasta, all of whom were misdirected by chatbots such as ChatGPT or Google’s Gemini.

Rescue officials and researchers attribute the problem to chatbots’ unwarranted confidence, which users often accept without verifying, especially when they lack local expertise. Psychologists and AI designers stress that answer quality depends on question quality, and that AI lacks mechanisms to signal uncertainty. Gowryluk says he will avoid AI for future hikes, though he may still use it for other tasks.