Travelers increasingly rely on artificial intelligence to plan outdoor trips, leading to many dangerous incidents requiring emergency rescue.

Chatbots can synthesize less-than-reliable information from social media to create maps and route advice that appear reputable but are not practical, posing serious risks when traveling in wild or rugged places.

According to Wired, three climbers had to call for help from a cliff on Mount Shasta in Northern California, USA. In late August, they made a series of consecutive mistakes. The group did not bring enough food and overestimated their physical strength. Instead of returning in the late afternoon when they found they hadn't reached their destination, they tried to climb to the top at 7:30 p.m. and then had to grope down the mountain in the dark. Later, the tourist group admitted their lack of preparation and poor planning was partly due to overreliance on Google's Gemini chatbot. As a result, the next morning, they got lost in the rugged mountain terrain and one person in the group suffered a knee injury.

They aren't the only ones making mistakes in the wilderness when seeking help from AI. According to Nick Meyers, manager at the Mount Shasta Wilderness Climbing and Avalanche Center, another visitor who pitched a tent near the "No Camping" sign insisted on staying because the AI ​​told her she was allowed. Meyers' team also encountered a man wandering at the trailhead with an AI-generated map that marked famous locations on the wrong side of a neighboring mountain.

According to People, in early October, the non-profit rescue organization North Shore Rescue had to mobilize a helicopter to rescue a 16-year-old boy stuck on a ledge on the Widowmaker Arete route behind Crown Mountain in British Columbia, Canada. This teenager uses Anthropic's chatbot Claude to plan a trip.

Meyers emphasized the need for monitoring to determine whether overuse of AI increases the number of accidents and rescues in remote areas. His office manages a large wilderness area in Northern California with an average number of rescues per year of 12 and 2026 is about the same as every year. On the contrary, the number of rescue calls in Wyoming has even decreased compared to last year, from the beginning of 2026 until now there have been only 312 calls while the number in 2025 is 466 and in 2024 is 377.

WiseVoter said that Grand Teton National Park in Wyoming, USA, recorded a significant increase in the number of incidents related to tourists using AI to obtain travel information in the past year, from 0 to 26%. Chip Jenkins, director of the national park, predicts the trend will continue.

A recent report from travel research firm Phocuswright shows that about 58% of American travelers are using AI to plan their itineraries and more than 60% of Millennial and Gen Z travelers have done so. However, Jungho Suh, a professor of management at the George Washington School of Business, said at least one study found that travel planning chatboxes were only accurate in 10% of cases. While technology has improved since the study he cited was conducted in 2024, chatbots still frequently make mistakes because they "rely on fixed data sources that can be outdated and incomplete."

Using AI chatbots to ask for directions or advice always comes with its own risks. Chatboxes can make up false information but it sounds convincing and real. According to Chris Flathmann, a professor of computing at Clemson University in South Carolina, using them can easily lead to lazy thinking and complete abandonment to machines.

"When working with AI systems, people can get so deep that they give up their own critical thinking," Flathmann explained. "When AI pretends to be an expert on an unfamiliar topic, you can't catch it because you don't understand it clearly. Once you're lazy about critical thinking, it's very difficult to reactivate your brain, especially when you're in a panic situation in the wilderness."

Flathmann also warns that chatbots tend to say what users want to hear. They always answer very assertively and confidently, regardless of whether the reference source is trustworthy or not. For example, Claude or ChatGPT can provide detailed instructions and advice that sound extremely reputable, but are actually gleaned from Reddit discussions, old YouTube videos, or even made up themselves.

While TikTok and Instagram only provide inspiration, AI often gives specific instructions (which may or may not be correct) with encouragement. "AI does not intentionally trick people into climbing, but this is an inevitable trend when we use such a tool," Flathmann told Wired.

According to AFAR, chatbots can misunderstand the real world in disturbing ways. Catherine Jackson Moynihan, an executive living in Charlottesville, Virginia, used ChatGPT to plan a campervan trip from the Outer Banks in North Carolina to the Boundary Waters region of Minnesota, with the requirement not to drive more than six hours per day. Moynihan shared that the chatbox worked well until she reached Lake Michigan, when it suggested she drive straight across.

Regardless of the source of information used, experts recommend that people double-check information with reputable sources when traveling to a completely new place, such as up-to-date maps and websites of public lands management agencies, climbing association websites, or travel guidebooks. Visitors can also talk to local guides or forest rangers.

An Khang summary