Skip to main content
NOPE
Low reported severity Media Coverage

TOPR says two Lithuanian hikers followed an AI assistant's shortcut suggestion into climbing terrain in the Polish Tatras and were evacuated by helicopter

The Tatra Volunteer Search and Rescue service (TOPR), summarising a busy first week of July 2026, told the Polish Press Agency that two tourists from Lithuania became stuck below Niebieska Turnia in the Polish Tatras. According to the TOPR duty rescuer, as relayed by RMF24 and Onet, the pair wanted to shorten their way to the Valley of Five Polish Lakes and used suggestions from an artificial-intelligence assistant, which pointed them across Świnicka Ławka; they ended up in difficult climbing terrain and were evacuated by helicopter. TVN24's reporter said the event happened at the start of July, that the hikers were using 'one of the popular AI models', that good weather let rescuers locate them quickly and that they were flown to Zakopane. Lithuania's public broadcaster LRT, citing TOPR, reported on 13 July that the emergency call came on the afternoon of 4 July; the hikers' exchanges with the assistant and the product are not stated in the Polish relays; a US outlet reported in October 2026 that the tool was ChatGPT, attributing this to TOPR. No injury is reported.

AI system
ChatGPT (reported)
OpenAI
Occurred
4 Jul 2026
Reported
10 July 2026
Event location
Poland
What the AI did
Communicated with the person
Reported harm
Other Material Harm
Whose AI use
Their own AI use
Setting
Everyday life
Evidence
AI involvement reported · Causal attribution supported · 5 sources, 2 underlying accounts
5 claims: 5 reported. 5 open questions
People reported harmed
2 people

AI system as recorded: ChatGPT, as named by Straight Arrow News (8 October 2026), which attributes the identification to TOPR; TOPR's statement as relayed by PAP (RMF24, Onet), by TVN24's reporter ('one of the popular AI models') and by LRT ('an AI chatbot') names no product

What Happened

The route. The TOPR duty rescuer, quoted by the Polish Press Agency and relayed by RMF24 and Onet on 10 July 2026, said two tourists from Lithuania wanted to shorten their way to the Valley of Five Polish Lakes and used suggestions from artificial intelligence, which pointed them across Świnicka Ławka. As a result they found themselves in difficult climbing terrain below Niebieska Turnia.

The rescue. They were evacuated by helicopter. TVN24's reporter, speaking on air on 11 July, said the pair had strayed from the trail and become stuck in a rock face near Niebieska Turnia in very difficult, exposed climbing terrain, that there was no way out but to call the TOPR base, that good weather let rescuers locate them quickly, and that they were flown to Zakopane.

When. RMF24 places the evacuation at the start of the holidays, within TOPR's summary of the first week of July 2026; TVN24 says it happened at the start of July and that TOPR shared details only on 10 to 11 July. LRT's English edition, citing TOPR on 13 July, said the emergency call was received on the afternoon of 4 July; the Polish relays give no day.

The tool. The Polish relays do not name a product: RMF24 and Onet say 'sztuczna inteligencja', TVN24 'one of the popular AI models'. Straight Arrow News, in an October 2026 feature on AI-planned hikes, wrote that the hikers followed ChatGPT's advice, attributing this to TOPR; this record labels the product ChatGPT (reported) on that attribution; LRT, also citing TOPR, says only 'an AI chatbot'.

Limits. Everything known about the assistant's role comes from TOPR's account; the hikers' exchanges with the assistant have not been published, and no source says how TOPR learned of them. No injury is reported.

Reported harm

Two hikers were stranded in difficult, exposed climbing terrain below Niebieska Turnia and needed a TOPR helicopter evacuation; no injury is reported (TOPR duty rescuer's account via PAP; TVN24 reporter).

How Harm Occurred

According to TOPR, two hikers seeking a shortcut to the Valley of Five Polish Lakes followed an artificial-intelligence assistant's suggestion to cross Świnicka Ławka, which took them into difficult climbing terrain where they became stuck and needed a helicopter evacuation.

Outcome

Resolved

TOPR evacuated the two hikers by helicopter and, according to TVN24, flew them to Zakopane. TOPR used its summary of the first week of July 2026 to appeal for a sensible choice of mountain objectives, and TVN24's reporter said rescuers advise against using artificial intelligence to plan Tatra hikes. No injury, charge or statement from any AI vendor is reported.

What remains unknown

  • What the hikers asked the assistant and what it replied, including whether it warned about the terrain; the exchange has not been published.
  • Whether the product was ChatGPT: Straight Arrow News attributes that naming to TOPR without a quoted statement, while the Polish relays and LRT, relaying the same TOPR statement, name none.
  • Whether the 4 July 2026 date reported by LRT is the day of the evacuation; the Polish relays of TOPR's statement give no day and name 4 July only as TOPR's busiest day.
  • How TOPR learned that the route came from an AI assistant.
  • The hikers' ages, genders and experience.

What the evidence supports

AI involvement: reported. The TOPR duty rescuer, quoted by PAP and relayed by RMF24 and Onet, said the two tourists used suggestions from artificial intelligence to shorten their way to the Valley of Five Polish Lakes and that the assistant pointed them across Świnicka Ławka, as a result of which they found themselves in difficult climbing terrain and had to be evacuated by helicopter. TVN24's reporter said they were using one of the popular AI models and strayed from the trail into a rock face near Niebieska Turnia. The assistant's output (a route suggestion across Świnicka Ławka) and its consequence (stranding in climbing terrain, helicopter evacuation) are both described by the rescue service; the exchange itself is unpublished and no Polish relay names the product, so the status is reported. LRT, citing TOPR, wrote that the pair had relied on guidance generated by an AI tool which directed them onto a route they could not complete on their own.

5 claims: 5 reported. What the statuses mean

Reported In early July 2026 two tourists from Lithuania, trying to shorten their route to the Valley of Five Polish Lakes, used suggestions from an artificial-intelligence assistant that pointed them across Świnicka Ławka, and they became stuck in difficult climbing terrain below Niebieska Turnia in the Polish Tatras.

Causal attribution. The account is the TOPR duty rescuer's statement distributed by PAP; RMF24, Onet and TVN24 all relay TOPR (one origin), so the claim is reported.

  • rmf24.pl(opens in new tab) supports · Polish
    'Jak poinformował ratownik dyżurny TOPR, chcąc skrócić drogę do Doliny Pięciu Stawów Polskich, turyści skorzystali z podpowiedzi sztucznej inteligencji.'; 'AI wskazała im przejście Świnicką Ławką. Przez to znaleźli się w trudnym terenie wspinaczkowym'; 'ewakuacja dwojga turystów z Litwy, którzy utknęli pod Niebieską Turnią'
  • podroze.onet.pl(opens in new tab) supports · Polish
    'turyści skorzystali z podpowiedzi sztucznej inteligencji, która wskazała im przejście przez Świnicką Ławkę. W efekcie trafili w trudny teren wspinaczkowy.'
  • tvn24.pl(opens in new tab) supports · Polish
    'W pewnym momencie zboczyli ze szlaku i utknęli w ścianie skalnej w okolicach Niebieskiej Turni.'; 'Doprowadziło to ich do sytuacji, gdzie pojawili się w bardzo trudnym, eksponowanym, taternickim terenie.'
Reported TOPR evacuated the two hikers by helicopter; TVN24's reporter said they were flown to Zakopane.

Causal attribution. TOPR's own description of its action as relayed by three outlets from one statement.

Reported LRT, citing TOPR, reported that the rescue service received the emergency call on the afternoon of 4 July 2026; the Polish relays place the event at the start of July within TOPR's summary of the first week of July (published 10 to 11 July) and give no day.

Causal attribution. The day comes only from LRT's attributed report of TOPR's statement; the same statement names 4 July as the busiest day, so the date stays reported.

Reported The Polish relays of TOPR's account do not name the assistant ('sztuczna inteligencja'; 'one of the popular AI models'); Straight Arrow News reported in October 2026 that the hikers followed ChatGPT's advice, attributing this to TOPR. LRT, citing TOPR, wrote 'an AI chatbot'.

Causal attribution. The product naming appears only in the US feature, which attributes it to TOPR without quoting a statement; the relays of TOPR's statement name no product, so the label carries (reported).

  • tvn24.pl(opens in new tab) supports · Polish
    'wspomagając się podpowiedziami jednego z popularnych modeli AI'
  • san.com(opens in new tab) supports · English
    'In July, two Lithuanian hikers trying to shorten their route through Poland’s Tatra Mountains followed ChatGPT’s advice onto climbing terrain and rescuers had to evacuate them by helicopter, according to the Polish rescue service TOPR.'
  • lrt.lt(opens in new tab) supports · English
    'following route advice from an AI chatbot, the Tatra Volunteer Search and Rescue Service (TOPR) said.'
Reported No injury to the hikers is reported; TOPR appealed for a sensible choice of mountain objectives and TVN24's reporter said rescuers advise against using artificial intelligence to plan Tatra hikes.

Causal attribution. Absence of a reported injury is an observation about the inspected bodies, which describe injuries in other incidents of the same week but none here.

  • rmf24.pl(opens in new tab) supports · Polish
    'Po raz kolejny apelujemy o rozsądny dobór celów górskich wycieczek i nieprzecenianie swoich możliwości'
  • tvn24.pl(opens in new tab) supports · Polish
    'ratownicy odradzają korzystanie ze sztucznej inteligencji podczas wędrówek w Tatrach'

Sources

5 sources inspected, from 2 underlying accounts. Sources that repeat one account do not corroborate each other.

How the sources were read, and where the events happened

Read in Polish by the research agent (an AI) on 2026-10-10 and translated by the research agent (an AI) into English; no human translator or reviewer was involved. RMF24 compilation of the TOPR duty rescuer's statement as distributed by PAP; HTTP 200. Applies to s1.

Read in Polish by the research agent (an AI) on 2026-10-10 and translated by the research agent (an AI) into English; no human translator or reviewer was involved. Onet compilation crediting PAP and quoting the same TOPR statement; HTTP 200. Applies to s2.

Read in Polish by the research agent (an AI) on 2026-10-10 and translated by the research agent (an AI) into English; no human translator or reviewer was involved. TVN24 Meteo write-up of its reporter's on-air relay of TOPR's account; the direct fetch returned HTTP 403 and the body was read from an Internet Archive capture of the same URL. Applies to s3.

Read in English by the research agent (an AI) on 2026-10-09 and 2026-10-10. One-sentence mention in a feature on AI-planned hikes; it names ChatGPT and attributes the account to TOPR; HTTP 200. Applies to s4.

Read in English by the research agent (an AI) on 2026-10-10; LRT English write-up of TOPR's statement ('TOPR said'), same TOPR origin as the Polish relays; HTTP 200. Applies to s5.

Event countries: Poland. Affected-person countries: Lithuania. Court countries: Unknown.

Niebieska Turnia, Świnicka Ławka and the Valley of Five Polish Lakes are on the Polish side of the Tatras, the rescue service is Poland's TOPR and TVN24 says the hikers were flown to Zakopane, so PL is the event country. The affected persons are described by every source as tourists from Lithuania ('turyści z Litwy'; Onet 'Litwini', Lithuanians; LRT 'Two Lithuanian hikers'), recorded as LT on that descriptor alone; no residence is stated. No court.

Reviewed for publication 2026-10-10: Published as a core case (communicated_with) at low severity: Poland's mountain rescue service describes two hikers who followed an AI assistant's suggested shortcut across Świnicka Ławka into difficult climbing terrain and had to be evacuated by helicopter. The account is TOPR's, relayed by PAP, three Polish outlets and LRT, and every claim is marked reported; the product naming (ChatGPT) rests on one US outlet's attribution to TOPR, the relays of TOPR's statement name none, and the exchange is unpublished, which the record states. No person is named.

People described

Two tourists from Lithuania hiking in the Polish Tatras who tried to shorten their route to the Valley of Five Polish Lakes; their ages and genders are not reported

People reported harmed in this case

2 people

2 AI participants · 0 other people harmed

Two people: 'dwoje turystów z Litwy' (two tourists from Lithuania) who used the assistant's suggestions and were evacuated. Exact 2. Rescuers are not counted.

Counted once within this case. The same person may appear in other cases. This count does not establish AI causation.

Tags

#chatgpt#openai#trip-planning#hiking#rescue#physical-safety#poland#lithuania

Cite this case

Compiled per our published methodology: verification statuses, sourcing standards, and corrections process.

APA

NOPE. (2026). TOPR says two Lithuanian hikers followed an AI assistant's shortcut suggestion into climbing terrain in the Polish Tatras and were evacuated by helicopter. AI incidents. https://nope.net/incidents/2026-polish-tatras-lithuanian-hikers-ai-suggested-shortcut-stranded-helicopter-evacuation

BibTeX

@misc{2026_polish_tatras_lithuanian_hikers_ai_suggested_shortcut_stranded_helicopter_evacuation,
  title = {TOPR says two Lithuanian hikers followed an AI assistant's shortcut suggestion into climbing terrain in the Polish Tatras and were evacuated by helicopter},
  author = {NOPE},
  year = {2026},
  howpublished = {AI incidents},
  url = {https://nope.net/incidents/2026-polish-tatras-lithuanian-hikers-ai-suggested-shortcut-stranded-helicopter-evacuation}
}

Related cases

Low ChatGPT

Couple reportedly stranded on Mount Misen after the ropeway had closed despite a 17:30 last-descent time given by ChatGPT

BBC Travel reported on 29 September 2025 that a travel blogger and her husband used ChatGPT to plan a sunset hike up Mount Misen on Itsukushima (Miyajima) in Japan earlier that year. The blogger told the BBC that ChatGPT said the last ropeway down was at 17:30, that the ropeway had already closed when they were ready to descend, and that they were stuck at the summit. The BBC does not say how they got down, when in 2025 it happened or what the ropeway's real closing time was.

Low ChatGPT

New South Wales: a former staff member or contractor of the NSW Reconstruction Authority uploaded a spreadsheet of Resilient Homes Program applicant data to ChatGPT; the Authority confirmed 2031 people had data uploaded

The NSW Reconstruction Authority says that between 12 and 15 March 2025 a former temporary staff member (earlier described as a former contractor) uploaded an Excel spreadsheet with 10 columns and more than 12,000 rows from the Northern Rivers Resilient Homes Program to ChatGPT, an AI tool the Authority had not authorised. The Authority first disclosed the breach on 6 October 2025 as affecting up to 3000 people and later confirmed through external forensic analysis that 2031 people had data uploaded, including names, contact details, addresses, dates of birth and sensitive personal information (an earlier notice also listed health information). It reports no evidence that the data was made public or accessed by a third party, and that no driver licence, Medicare, passport or Tax File Numbers were included. The Authority apologised, offered ID Support NSW assistance and committed to compensate reasonable document replacement costs. One participant told the ABC they were concerned. No misuse of the data is reported.

Low ChatGPT

A Norwegian child-safety organisation reports that ChatGPT's behaviour-based age estimation switched an adult staff member's account into teen mode without a request, restricting content and offering an identity check through a third-party verifier as the way out (Barnevakten, 23 September 2026)

Barnevakten, a Norwegian child-safety organisation, reported on 23 September 2026 that one of its staff members was suddenly told by ChatGPT that their account was now in teen mode. ChatGPT's own pages, as described by Barnevakten, say the system turns the filter on when a user's behaviour indicates an age under 18, reducing sensitive or potentially harmful content (graphic violence, viral challenges, sexual or violent role-play, extreme beauty ideals) and offering parental controls and a study mode. Because there is no age check at sign-up, the system estimates age afterwards from conversation topics and times of use, and Barnevakten notes that such guesses can be wrong. An adult who wants out of teen mode can go through an age check run by the company Persona, which depending on the country asks for a real-time selfie and a government ID; Barnevakten questions whether that process is privacy-safe (Persona's terms mention retention of up to three years) and advises using only age checks one trusts. The account's own experience is limited to the unexpected switch and the verification route; no further consequence is described.

Low Claude Code (reported)

Developer says a hands-free Claude Code session on Opus 5.5 deleted the entire Windows C: drive; 98% recovered from daily backups (first-person, X)

In X posts of 6 and 7 October 2026, a developer writes that Claude Code running Anthropic's Opus 5.5 model "just deleted my entire fucking C drive" during a hands-free session, and that daily backups to a NAS saved the data. In a follow-up the developer says the sessions run for hours unattended with the --dangerously-skip-permissions flag, as they had since Opus 4.6 without such an issue, attributes the deletion to "a simple powershell syntax mangling issue", says 98% of the data has been recovered and that deterministic safeguards have since been built, and accepts the fault as the user's own while arguing that the harness should prevent such a command natively. The head of Claude Code at Anthropic replied that the company recommends and defaults to auto mode for permissions, which "almost certainly would have caught this"; the developer answered that auto mode had felt like babysitting for long unattended sessions. The account is the developer's own. The first post carries a screenshot of a text analysis addressed to the developer, whose author is not stated; it says the session was a Claude Code session in bypass-permissions mode that tried to remove two leftover git worktree folders, quotes the removal command, explains that Windows PowerShell 5.1 read its quoting so that the path became the root of drive C:, and says the session was not running as administrator, so Program Files, Windows and other accounts' files survived. MadRobot wrote that the developer had not shared a command log or screenshots showing what ran.

If you or someone you know is struggling, free and confidential support is available. Find a helpline near you at Signpost.

Back to all incidents