Tourists stranded in mountains after ChatGPT advice

TVN24 journalist Maciej Stanisńsky detailed the incident, noting that the tourists relied on recommendations from a popular AI chatbot during their hike.

"This led them into a situation where they found themselves in very difficult, exposed climbing terrain. There was no way out — they had to call the TOPR center and ask rescuers for help," Stanisńsky said.

What happened

According to media reports, the travelers decided to take a shortcut to the Valley of the Five Polish Lakes and asked ChatGPT for the best route. The neural network suggested crossing via the Świnica Ledge (Świnicka Ławka).

The tourists veered off the official marked path, ended up on the north face of the Świnica peak, and became stranded below Niebieska Turnia. TOPR rescuers categorize this area as technical climbing terrain rather than a standard hiking trail.

The travelers were unable to extract themselves from the situation.

The weather that day was favorable, allowing rescuers to locate the tourists quickly. A helicopter was dispatched, and both were evacuated to Zakopane. No injuries were reported.

Why ChatGPT may have suggested such a route

Journalists attempted to investigate why the artificial intelligence suggested such a path in the first place.

It turns out that the Świnica Ledge does indeed exist. It is not a fictitious route, but a long rocky shelf on the north face of Świnica, at an altitude of about 350 meters. It is used by mountaineers for purposes such as traversing the face or exiting climbing routes.

In other words, ChatGPT did not "invent" a non-existent path. Most likely, the neural network mixed hiking and mountaineering sources. In climbing guides, the Świnica Ledge is mentioned as a relatively simple escape route. However, "simple" in this context means simple for trained climbers with safety gear, not for ordinary hikers.

During their investigation, journalists also received an admission of error from the chatbot.

"You are absolutely right. I should not have sent you through that pass," the message stated.

Advice from rescuers

Following the incident, rescuers reiterated their recommendation against using artificial intelligence as a primary source of information when planning mountain routes.

As Maciej Stanisńsky noted, nothing can replace experienced mountain guides or verified paper and specialized hiking guidebooks.

"Nothing replaces real Tatra guides who know how to navigate the Tatras and select routes according to one's abilities. Of course, it doesn't have to be a person — it can also be verified printed guides," the journalist noted.

According to TOPR data, rescuers assisted 55 tourists in the first week of July alone. The busiest day was July 4, when they were forced to conduct 25 separate rescue operations.

How to avoid similar situations

Rescuers recommend cross-referencing routes with official maps before a hike and mandatory checking of difficulty levels, direction of travel, and information regarding closed sections.

While specialized hiking apps can be used for navigation, routes should still be verified against official national park information before setting out.

Additionally, tourists in Poland are advised to install the RATUNEK app, which automatically transmits phone coordinates to rescuers when a call for help is made.

This is by no means the first instance where AI advice has led travelers into trouble.

Previously, a family of travelers reported that they could not fly to North Macedonia after ChatGPT incorrectly stated that Russians do not require a visa. At the airport, it was revealed that entry requires a document that must be processed in advance. Ultimately, the family had to decide whether to cancel the trip or find another country for their vacation.

Another similar case involved a Spanish couple heading to Puerto Rico. According to the tourists, ChatGPT told them that no additional permits were required for the trip. However, during check-in, it was discovered that while Spanish citizens do not need a visa, an ESTA electronic authorization must be obtained prior to travel. As a result, the travelers were barred from their original flight, though they eventually reached their destination.

These incidents further demonstrate that generative AI can make mistakes or overlook critical details. When planning travel, applying for visas, selecting routes, or addressing other matters related to safety or official requirements, it is recommended to verify information against official sources and data from relevant authorities.