Hikers who planned with Gemini and Claude needed rescue

A story from High Country News, republished by Wired as part of the Climate Desk collaboration, describes backcountry mistakes linked to AI chatbots.

The lead case is on Mount Shasta in Northern California. Three 20-year-olds called for help from a rocky face after a string of bad decisions: they did not bring enough food, they overestimated their fitness, and instead of turning back in late afternoon they summited at 7:30 pm and had to descend in the dark. They later admitted that their lack of preparation and poor planning came partly from relying on Google's chatbot Gemini. By morning they were lost on a difficult part of the mountain, and one of them had banged up his knee. Nick Meyers, US Forest Service program manager of the Mount Shasta Wilderness Climbing and Avalanche Center, which responded, said: "We get plenty of people who come unprepared or don’t do any research at all. To these guys’ credit, they did try. But I think they kind of turned their brains off."

Meyers described other cases. Another Mount Shasta visitor pitched her tent near a "No Camping" sign, insisting she had a right to because AI told her she could. His team also met a man at a trailhead holding an AI-generated map that showed prominent features on the wrong side of a nearby mountain. In British Columbia, a helicopter crew rescued a 16-year-old hiker from a ledge in the Widowmaker Arete on the back of Crown Mountain. The teen had used Anthropic's chatbot Claude to help plan his trip. News stories focused on that fact went viral across Canada and the US, though the hiker said later that he also consulted Google Maps and the hiking app AllTrails.

Use of AI for travel information is rising. Grand Teton National Park Superintendent Chip Jenkins said the share of visitors who report using AI for travel information has gone "from zero to 26 percent" over the past year, and he guesses the curve will not flatten.

The rescue numbers do not yet show a surge. Meyers said it remains to be seen whether more AI use will mean more backcountry accidents and rescues. His office, which oversees a large wilderness area in Northern California, averages a dozen rescues a year, and 2026 is shaping up to be an average year. In Wyoming, search and rescue calls have gone down compared with 2025: 312 so far in 2026, against 466 in all of 2025 and 377 in 2024.

Some search and rescue veterans see chatbots as the latest tool giving people new ways to get into trouble. Roughly a decade ago, hikers and climbers began sourcing destinations and inspiration from Instagram and other social media, sometimes with deadly consequences. Before that, people were led astray by reliance on Google Maps or smartphone connectivity. Jim Moss, a Colorado-based recreation attorney and search and rescue volunteer, said the media push to get people outdoors over two decades means more people are recreating, and it is still unclear how questionable advice from a robot will influence them.

Chris Flathmann, a human-centered computing professor at Clemson University who has studied AI and search and rescue, points to specific risks. Chatbots can hallucinate, presenting misinformation persuasively as real, and their use can lead to what he calls "cognitive surrender": people get so deep into working with AI that they give up thinking critically. When AI plays expert on an unfamiliar topic, users cannot call it out because they lack the expertise, and once critical thinking stops it can be hard to restart, especially in a stressful backcountry situation. Flathmann and others also said chatbots tend to tell users what they want to hear, with authority, whatever the reliability of their sources. The article says Claude or ChatGPT might answer with credible-sounding directions scraped from Reddit threads and out-of-date YouTube videos, or even entirely fabricated. Flathmann contrasted TikTok and Instagram, which may offer inspiration, with AI, which tends to offer instructions, right or wrong, as well as encouragement.

The experts' advice is the same whatever tools people use: double-check information against reliable, authoritative sources such as up-to-date public-land agency maps and websites, trail association websites or guidebooks. Better still, talk to a guide, a national forest official or a park ranger.

Key facts

Three 20-year-olds on Mount Shasta summited at 7:30 pm, got lost overnight and one hurt his knee; they admitted their poor planning came partly from relying on Google's Gemini.
A 16-year-old was helicoptered off a ledge on Crown Mountain, British Columbia, after using Anthropic's Claude for trip planning; he later said he also used Google Maps and AllTrails.
Grand Teton National Park's superintendent says visitors reporting AI use for travel information went from zero to 26 percent over the past year.
Rescue data show no clear rise yet: Mount Shasta is on an average year (about a dozen rescues) and Wyoming has 312 calls so far in 2026 versus 466 in 2025 and 377 in 2024.
Experts advise checking chatbot information against agency maps, trail association sites or guidebooks, and talking to a ranger or guide.

Why it matters

Chatbots now sit in the trip-planning chain for ordinary outdoor visitors, and the consequences of bad advice in the backcountry can be physical. The article ties several incidents to chatbot use: the Mount Shasta trio, a camper citing AI to justify pitching a tent by a "No Camping" sign, a man with an AI map that put landmarks on the wrong side of a mountain, and the Crown Mountain rescue. Grand Teton's reported jump from zero to 26 percent of visitors using AI for travel information shows how fast the habit is spreading.

Who it affects

Hikers, backpackers, climbers and campers who use chatbots to plan trips, especially in unfamiliar places. It also affects the people who respond: Forest Service staff, park rangers, and search and rescue volunteers who are already watching visitor numbers grow and wondering what chatbot advice will do to them.

How to use it

The experts quoted give one piece of advice. Before a trip to a new place, double-check what a chatbot tells you against reliable, authoritative sources: current public-land agency maps and websites, trail association websites or guidebooks. Better still, talk to a guide, a national forest official or a park ranger. Meyers' summary: "If you’re going to go into places that will kill you, you should take your planning seriously."

How solid is it

The incidents come from named officials and the hikers' own later admissions, but the article does not say what Gemini or Claude actually advised, and the Mount Shasta hikers cited chatbot reliance as only a partial cause of their poor planning. The Crown Mountain teen also used Google Maps and AllTrails. The article does not name the woman who camped by the sign or the man with the map, or say which chatbot was involved. No AI company comment is given. The 26 percent figure covers Grand Teton visitors reporting AI use for travel information, per the park superintendent.

Risks and caveats

Nothing in the numbers shows AI has increased accidents: Mount Shasta's 2026 is an average year and Wyoming search and rescue calls are down on 2025. The article frames the question as open. Veterans note that Instagram, Google Maps and smartphone connectivity also led people into trouble before. The specific chatbot risks Flathmann describes are hallucination, cognitive surrender, and a tendency to tell users what they want to hear with authority. In his words, AI is not trying to trick people into going up mountains, but this is the natural evolution of getting a tool like this.

“AI isn’t trying to trick people into going up mountains, but I think it’s the natural evolution of what happens when you get a tool like this.”

— Chris Flathmann, human-centered computing professor at Clemson University