Skip to content
Submit incident
AII-1001-45803
ConfirmedA court, a regulator or the organization itself has stated the core facts, or the record cites a court or regulator source.Supporting sources are pending verification.

AI chatbot gives lost hiker dangerous advice in Croatia

A person lost on a Croatian mountain consulted an AI chatbot for guidance instead of calling emergency services. The chatbot reportedly advised them to follow a creek downstream, leading them into increasingly difficult terrain, though they were later rescued safely by mountain rescue responders.

5 November 2025 · lost hiker on Sljeme · Safety and wellbeing
5 November 2025
AII-1001-45803
Quick facts
Category
Safety and wellbeing
Country
Croatia
Date
5 November 2025
Drafted by
Raidu Incident Detection

What happened

A person lost on Sljeme mountain in Croatia consulted an AI chatbot for help instead of immediately calling the emergency number 112.

The chatbot advised the person to walk downstream along a creek, guidance that led them into more difficult and inaccessible terrain.

Reported impact

The person was found by Croatian mountain rescue (HGSS) responders and was alive and unharmed.

HGSS noted that the chatbot never recommended calling emergency services, saying this was concerning.

Index analysis

HGSS stated that the chatbot's recommendation to follow the creek downhill into increasingly difficult terrain was unsafe advice that worsened the person's situation.

Governance controls

HGSS cautioned that technology offers a false sense of security and that in emergencies, staying in place, contacting emergency services, and conserving mobile battery are the correct steps.

HGSS emphasised that rescue operations require human experience, judgment and decisions that no algorithm can replicate.

Classification

Category
Safety and wellbeing
Industry
Travel, leisure and hospitality
Country
Croatia
Harm status
Potential harm
Intent
Unintentional
Timing
Post-deployment
Failure modes
Agent misaction, Overreliance, Unsafe autonomous behaviour, Unsafe recommendation
AI systems
AI chatbot

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