AI Hiking Advice Is Sending Some Trekkers Into Danger, WIRED Reports

AI chatbots are now shaping some hikers’ route plans in the backcountry, and according to WIRED, that is contributing to bad decisions and more dangerous rescues in wilderness areas. The reporting points to recent incidents at Mount Shasta, Grand Teton and Crown Mountain as evidence that fluent chatbot answers can push people into trouble when they replace maps, ranger guidance and basic trip planning.

What happened on Mount Shasta?

WIRED reports that three hikers in their 20s called for help after a Mount Shasta outing went wrong in ways rescuers linked partly to Google Gemini use. The hikers summited at 7:30 p.m., then had to descend in the dark, and by morning they were lost, with one member of the group injured in the knee.

The Mount Shasta Wilderness Climbing and Avalanche Center’s Nick Meyers told WIRED that hikers often show up unprepared, but he said this case stood out for how much the group appeared to rely on chatbot output instead of their own judgment.

“But I think they kind of turned their brains off.”

That comment captures the central concern running through the reporting: when people outsource backcountry decisions to an AI system that sounds confident, they may stop checking obvious red flags, including the time of day, the difficulty of the route, and whether they have enough daylight left to descend safely.

Why it matters

In the backcountry, small mistakes compound quickly. A wrong turn, a poor estimate of daylight, or a misunderstanding about where camping is allowed can become a rescue call, an injury or a life-threatening overnight stay.

WIRED’s reporting suggests AI may be different from older forms of bad internet advice. Social media can inspire risky outings, but a chatbot can look and feel like a planning tool. That makes an error more dangerous because users may treat it as instruction rather than opinion.

The stakes are especially high because wilderness travel depends on current information: trail conditions, weather, access restrictions, closures and local safety rules. A chatbot can produce a polished answer without actually knowing whether it is current, correct or even grounded in reality.

Why AI errors can be more consequential outdoors

Researchers cited by WIRED say this is not just a question of one bad recommendation. Chatbots can fabricate details and still sound authoritative, which makes it harder for users to notice when they are being misled.

Chris Flathmann, a human-centered computing professor at Clemson University, described a broader behavioral problem that can emerge when people lean on AI for decisions.

“When humans work with AI systems, they can get so deep into it they give up thinking critically for themselves.”

That dynamic is especially risky in places where there is no reliable phone service, no quick way to verify a claim on the spot and little margin for error after sunset.

How did AI become part of hiking decisions?

AI did not arrive in the backcountry all at once. The broader pattern, according to WIRED, mirrors the way social media became a source of risky route inspiration roughly a decade ago. The difference now is that chatbots are not merely showing people a dramatic place to go; they are helping construct the trip itself.

That shift matters because route planning is more than finding a destination. Hikers need to understand distance, elevation gain, weather windows, terrain, water availability, permit rules and escape options. If any of those details are wrong, a plan that looks sensible on a screen can become unsafe on the mountain.

From inspiration to instruction

WIRED’s reporting suggests the danger is not just that AI can be wrong, but that it can be wrong in a more actionable way than social posts. A photo can inspire a hike. A chatbot can tell someone exactly where to go, when to leave and how to camp. If the advice is false, the user may follow it as if it were authoritative.

The result is a new kind of backcountry risk: not just overconfidence in a beautiful place, but misplaced confidence in a machine that presents itself as helpful and specific.

Other rescues tied to AI planning

The Mount Shasta case was not isolated, according to WIRED. Rescuers in other places have also encountered hikers whose plans were shaped by AI-generated guidance.

What happened near Crown Mountain?

WIRED reports that a 16-year-old in British Columbia was rescued from Crown Mountain after using Anthropic’s Claude for trip planning. The story adds another example of a chatbot-generated plan colliding with reality in a setting where route knowledge and conditions matter.

Another incident involved a man found at a trailhead with an AI-made map that misplaced major landmarks, according to the report. That kind of error is not minor. If the landmarks are wrong, the whole route logic can collapse.

WIRED also reported a visitor camping near Mount Shasta next to a “No Camping” sign who said AI had told him it was permitted. That case underscores how chatbot errors can spill beyond navigation into land-use compliance and safety rules.

What are land managers seeing?

Park and rescue officials are beginning to notice the pattern. At Grand Teton National Park, superintendent Chip Jenkins told WIRED that reports of visitors using AI for travel information have risen sharply over the past year.

“It went from zero to 26 percent. And I’m guessing that curve is not going to flatten.”

That figure is striking because it suggests not only adoption, but fast adoption. According to WIRED, the park had previously reported 0 percent, and the reported share is now 26 percent. Even if the number reflects self-reported use rather than a full census, the shift points to a meaningful change in how some visitors gather information before heading into remote areas.

By the numbers

Are rescues actually rising because of AI?

Not necessarily. WIRED notes that Wyoming search-and-rescue call volume is down in 2026 compared with prior years, even as concerns about AI use grow. The state had 312 calls so far in 2026, below 466 in 2025 and 377 in 2024.

That means the current evidence does not support a simple claim that AI is driving a flood of new rescues. The clearer point is narrower and still important: rescuers are seeing AI show up in more incident narratives, and the mistakes can be vivid, avoidable and dangerous.

There is also an important distinction between total call volume and incident quality. A single rescue triggered by a bad chatbot recommendation can still consume time, volunteers, equipment and personnel, even if the overall annual total declines.

Background: how we got here

Backcountry travel has always depended on reliable, local information. People heading into mountains, forests or alpine terrain need accurate maps, up-to-date weather, trail conditions and knowledge of local restrictions. That has long been true whether the source is a paper map, a ranger station, an app or a friend’s advice.

What has changed is the speed and confidence with which AI systems can answer questions. A chatbot can generate a route summary, a packing list or a camping recommendation in seconds. But as WIRED’s reporting makes clear, speed is not the same as accuracy.

That gap matters because the outdoor context is unforgiving. Search-and-rescue teams routinely deal with hikers who underestimate terrain, fitness, timing and food. AI adds another layer: a tool that can sound exact while being wrong about something critical.

Why official sources still matter

The standard safety advice in the WIRED report is simple: verify AI guidance against official maps, agency websites, trail groups or human experts.

That advice is not old-fashioned. It is a recognition that no single tool should be treated as final authority in a setting where conditions change and errors have immediate physical consequences. Official park and forest guidance is designed for exactly the sort of real-world questions chatbots can mishandle.

What the experts are warning about

Experts quoted in the WIRED report frame the problem as a mix of technical failure and human psychology. The machines can be wrong. Users can also become less skeptical once the machine seems helpful.

Flathmann’s phrase, “cognitive surrender,” describes that second problem: users stop interrogating the answer because the system appears to have done the work for them. In the outdoors, where conditions are dynamic and danger can arrive quickly, that surrender can be costly.

Rescue officials are concerned for another reason too: AI advice can be more instruction-like than social media inspiration. That means the user is not just reading about an adventure; they may be following a step-by-step plan that feels tailored and trustworthy. When the plan is wrong, the failure is more direct.

What happens next?

The biggest open question is whether AI-assisted route planning is actually increasing rescues overall, or merely becoming more visible within rescues that would have happened anyway. WIRED’s reporting does not settle that question, but it shows why land managers are paying attention.

Another question is whether parks, forests and trail organizations will begin issuing formal warnings about chatbot use. The reporting suggests this may become more common if incidents continue to stack up, especially where people are relying on AI for route selection, camping permission or timing decisions.

Officials will also be watching whether incident reports start documenting AI as a factor more routinely. If that happens, the public record could become clearer about how often chatbots are involved in mistakes that end in rescue.

What hikers should watch for

• Whether official park and forest agencies begin adding AI-specific safety advice.
• Whether rescue teams report more cases involving chatbot-generated maps or route plans.
• Whether visitors increasingly cite AI tools in incident reports.
• Whether local trail groups start offering clearer warnings about verification before departure.

The bottom line

WIRED’s reporting points to a new kind of outdoor risk: hikers using AI chatbots for route planning, camping advice and landmark navigation, then discovering too late that the answers were wrong. The evidence so far does not prove that AI is driving rescue totals higher across the board, but it does show that bad chatbot advice is already reaching the trailhead, the summit and the rescue call.

For anyone heading into remote terrain, the lesson is plain. A fluent answer is not the same as a safe one, and in the mountains, confidence alone is not a plan.