Skip to content
Submit incident
AII-1002-56831
ReportedReported by fewer than three independent publishers, with no official finding.Supporting sources are pending verification.

ChatGPT recommends unsafe mountain hiking route to tourists in Poland

Three tourists in Poland's Tatra mountains asked ChatGPT for hiking directions and were directed to an unsafe route through Przełęcz Krzyżne that failed to account for winter conditions. The hikers encountered icy paths and blizzard conditions, requiring rescue by mountain rescue services.

January 2025 · OpenAI · Safety and wellbeing
January 2025
AII-1002-56831
Quick facts
Category
Safety and wellbeing
Country
Poland
Date
January 2025
Drafted by
Raidu Incident Detection

What happened

Three tourists hiking in Poland's Tatra mountains asked ChatGPT for directions from Hala Gąsienicowa to Dolina Pięciu Stawów. ChatGPT suggested the route through Przełęcz Krzyżne, which is difficult even in ideal summer conditions.

The hikers encountered treacherous winter conditions including limited visibility, snow blizzard, and icy rocks. Realizing they were in serious danger, the tourists called mountain rescue service TOPR for help and were safely brought down from the mountain.

Reported impact

The incident required intervention by mountain rescue services to safely extract the three hikers from dangerous conditions.

Index analysis

ChatGPT is a general-purpose tool that can generate inaccurate and unreliable information. The system failed to account for real-time weather conditions and seasonal hazards when recommending the hiking route.

According to a TOPR rescuer quoted in media reports, this was the first case rescue services had encountered of tourists following ChatGPT directions. The rescuer emphasized that artificial intelligence cannot replace careful planning and local knowledge.

Governance controls

OpenAI warns users that ChatGPT can make mistakes but does not disclose known limitations and risks in detail at the point of use, potentially leading users to believe the system is reliable for all situations.

Classification

Category
Safety and wellbeing
Industry
Travel, leisure and hospitality
Country
Poland
Harm status
Actual harm
Intent
Unintentional
Timing
Post-deployment
Harmed
Consumers
Failure modes
Agent misaction, Unsafe recommendation
AI systems
ChatGPT

Sources and evidence

Sources not yet published

This record was researched and written by the Index from original reporting. Its sources are still being verified by an editor, and they are listed here once that check is complete. Until then the record carries no reviewer.

How sources are verifiedSuggest a source for this incident