AI trip-planning chatbots are leading some hikers into real backcountry danger
Hikers who leaned on Gemini, Claude and other chatbots for trail advice have ended up lost, injured, or stranded — and rangers say the pattern is becoming harder to ignore.
What happened: Search-and-rescue teams across North America are reporting a pattern: hikers getting into trouble after leaning on AI chatbots instead of official trail resources. On Mount Shasta, three underprepared 20-year-olds used Google's Gemini to plan a summit, pushed on past dark, and ended up stranded with an injured hiker. Rangers there also met a camper who pitched her tent next to a 'No Camping' sign because a chatbot told her it was fine, and a man following an AI-generated map that misplaced major landmarks. In British Columbia, a 16-year-old who'd consulted Anthropic's Claude had to be pulled off a ledge on Crown Mountain by helicopter. At Grand Teton National Park, the share of visitors citing AI for trip planning jumped from zero to 26 percent in a year.
Why it matters: These aren't just embarrassing anecdotes — they show how differently people treat a chatbot's answer versus a map or a ranger's warning. Chatbots respond with confident, detailed instructions regardless of whether the information is accurate, current, or even real. Clemson researcher Chris Flathmann calls the result 'cognitive surrender': once someone treats an AI as an authority, they stop questioning it, especially on topics where they lack the expertise to catch mistakes. In the backcountry, where conditions change constantly, that unearned confidence can turn a planning shortcut into a life-threatening situation.
How it works, plainly: Chatbots don't have real-time knowledge of trail closures, snow conditions, or which side of a mountain a landmark actually sits on. They generate plausible-sounding text, sometimes pulled from scraped sources like old Reddit threads or outdated YouTube videos never vetted for accuracy, and sometimes simply invented. Unlike a search engine, which offers links to check, a chatbot delivers one confident, conversational answer — making it harder for an inexperienced user to tell a credible route from a hallucinated one, or a real rule from a made-up one about where camping is allowed.
What the data — and history — actually show: It's not yet clear AI is driving more rescues overall. Wyoming's search-and-rescue calls are actually down in 2026 compared with the prior two years, and Mount Shasta's rescue count is tracking average. Experts note this echoes earlier tech waves: Instagram sent hikers to dangerous, poorly marked viewpoints a decade ago, and smartphone maps led people astray before that. Their advice hasn't changed with the technology — cross-check any AI trip plan against official land-agency sites, trail-association resources, or guidebooks, and talk to an actual ranger or guide before heading into terrain that can kill you.