It is August 2026. Three novice hikers thought they were prepared to ascend Mount Shasta. Google Gemini had generated a kit list for them and estimated how much food and water they would need for the climb. But by the end of the day, they found themselves in a dangerous situation, spending an unplanned night on the mountain awaiting rescue with no suitable provisions.
AI tools such as ChatGPT and Copilot have become a trusted source of instant information for millions of people. In a matter of seconds, these tools can answer questions about almost any mountain challenge you care to name. However, AI can only work with the information available to it.
In a risky mountain environment, where weather, terrain and conditions can change rapidly, inaccurate or outdated advice can have serious consequences. Blindly trusting every answer can leave you in situations that you are unprepared for.
When AI planning falls short
Mount Shasta rises to 14,179 feet (4,322 m) and is one of North America’s most popular glaciated climbing peaks. Thousands of people hike this trail every year. However, this is not a place to be unprepared. Popularity can create a false sense of security for many researching the hike. In reality, it remains a challenging mountain expedition where climbers can encounter high winds, snow, ice, rockfall and rapidly changing conditions.
Three hikers left base camp carrying provisions for what they expected to be an eight-hour climb. In the end, it was closer to double this by the time they were rescued.
They ignored officials advice during the hike, instead relying heavily on AI-generated recommendations, as well as the AllTrails app for navigation. After one hiker injured their knee in a fall at around 11,500 feet, the group were forced to spend an unplanned night in Mud Creek Canyon with limited supplies and no reliable means of communication. USFS Climbing Rangers eventually found and rescued them the next morning.
The confidence trap
In just a few years, tools such as ChatGPT, Microsoft Copilot, and Gemini have evolved from niche technologies into everyday assistants. AI is designed to communicate clearly, confidently and persuasively. As a result, many users are placing increasing trust in AI-generated information, including advice on route planning, equipment selection and expedition preparation.
But the mountain doesn’t care what AI told you yesterday. Conditions can change in an instant, and a route that was safe yesterday may be dangerous today. Severe weather, route closures, avalanche risk, landslips and rescue alerts may not be reflected in an AI-generated response. Not to mention the fact that AI hallucinates and often confidently and convincingly generates incomplete or inaccurate information.
Trust your own human judgement
Just a few days before this particular incident in Mount Shasta, Google announced a partnership with YouTube influencer Jimmy Donaldson, aka MrBeast, promoting how to “survive harsh environments with Gemini”. The first creator to reach half a billion followers, MrBeast is well-known for his endurance challenges.
In his first collaboration video with Google, Jimmy and his crew attempt to survive as they race each other through some of the most extreme and treacherous climates on the planet. Each team faced different challenges in the jungle, the desert, and the Arctic. Meanwhile, they advertised how they used Gemini to help them identify dangers and survive brutal weather changes.
Despite how seamlessly these tools are marketed, relying on AI in the wilderness carries severe risks. A basic level of training and experience is essential for staying safe. Human survival skills like map reading, compass work, and sound judgement cannot be substituted by AI. These abilities may seem less important when everything is going to plan, but in difficult conditions they can be the difference between getting home safely and requiring rescue.
“AI makes planning feel easier and more certain than it really is,” explained Davide Guzzardi, CEO at Exped Tribe, an expedition trip and mountain guide company from central Switzerland. “An AI gives you a confident answer about a route, but it doesn’t know that there’s still snow in the north-facing gully this week, or that a section was damaged by rockfall. The information can be outdated or just wrong, and it sounds convincing. It doesn’t replace local knowledge, your own judgment, or personal skills either”.
Behind the perfect picture
The recent Mount Shasta incident highlights the dangers of treating AI-generated advice as expert guidance. The hikers had little margin for error, limited mountain experience and did not fully use local sources such as avalanche and conditions reports. This is just one occurrence in a growing string of cases of people venturing out unprepared following advice from AI or social media videos, and ending up needing to be rescued.
“Social media shows us the “highlight”, the summit, not the effort to get there. You see one beautiful photo from a ridge, taken on a perfect day, and it looks easy and accessible. You don’t see the six hours before it, the exposed section, or the fact that the person has been doing this for ten years,” shared Guzzardi.
According to data from Mountain Rescue England and Wales, the number of rescues required jumped by 24% between 2019 and 2024. Analysed by Ordnance Survey, the report insights revealed that the demographic requiring the most amount of rescue services now is 18-24 year olds, as opposed to 50-54-year olds in 2019.
“It’s hard for us to give a definitive reason for the age group change. But we know from incident reports that more and more people are tempted into risky locations by Instagram posts and the navigation apps being used aren’t always suitable for an outdoor environment. Add to that a tendency to follow popular routes online without knowing the hazards and context, and it’s not surprising that those likely to be relying most on their phones are the ones getting into difficulties,” said Mike Park, CEO of Mountain Rescue England and Wales (MREW).
A harsh lesson in the limits of AI
Many incidents attended by the MREW rescue teams share common themes: poor planning, inadequate equipment, navigation errors, a lack of local knowledge, and people underestimating the challenges they face. These are exactly the types of mistakes that can be made worse when individuals place too much trust in technology.
The lesson is simple: use AI as an assistant, not an authority. Ask questions, challenge its answers and verify critical information through trusted sources. Because when things go wrong on a mountain, it is not an AI that faces the consequences. It is you.
Read more related stories
Support independent, honest journalism
At STERNA, we believe travel journalism should provoke new thought, inspire positive change, and highlight the purpose-led and sustainable businesses protecting our planet and local communities.
True, independent storytelling takes intention and deep research. We operate without clickbait, sensationalism, or corporate bias.
By contributing to STERNA, you directly fund independent journalism and help us bring more stories of positive environmental and cultural impact to light.
If our work has inspired you, consider supporting our mission.
Contribute to support independent journalism.
At STERNA, we believe travel journalism should provoke new thought, inspire positive change, and highlight the purpose-led and sustainable businesses protecting our planet and local communities.
True, independent storytelling takes intention and deep research. We operate without clickbait, sensationalism, or corporate bias.
By contributing to STERNA, you directly fund independent journalism and help us bring more stories of positive environmental and cultural impact to light.
If our work has inspired you, consider supporting our mission.
Contribute to support independent journalism.
Choose an amount:
Or enter a custom amount:
Your contribution is appreciated, and helps to keep our site ad-free.
Your support is appreciated, and helps to keep our site ad-free.
