Skip to main content
NOPE
Low reported severity

Japanese Q&A poster says ChatGPT suggested walnut sweets for a nut allergy and called a chilled gift safe unrefrigerated, wasting time and money (first-person)

In a question posted on the Japanese Q&A site Yahoo! Chiebukuro on 26 May 2026, the author says that when they asked ChatGPT for sweets that someone with a nut allergy could eat, it suggested sweets that plainly contained walnuts, and that when they asked for a gift that could be carried for a long time at room temperature, it said the item was "definitely fine at room temperature", but at the shop it turned out to need refrigeration. The author says these repeated failures wasted their time and money and writes that, with the allergy, it was in the worst case one step from death; the post does not say whether anyone ate the suggested sweets or had a reaction. The author asks how to stop the chatbot from answering without a basis. The account is the author's; no conversation record is shown.

AI system
ChatGPT (reported)
Occurred
Event date unknown
Reported
26 May 2026
Event location
Unknown
What the AI did
Communicated with the person
Reported harm
Other Material Harm
Whose AI use
Their own AI use
Setting
Everyday life · Health
Evidence
AI involvement reported · Causal attribution alleged · 1 source
4 claims: 1 documented, 3 reported. 5 open questions
People reported harmed
1 person

AI system as recorded: ChatGPT (OpenAI), as named by the question's author; model and plan not stated

What Happened

What the author asked. According to the question, the author asked ChatGPT for sweets that a person with a nut allergy could eat, and separately for a gift that could be carried for a long time at room temperature.

What ChatGPT answered. The author says ChatGPT suggested sweets that plainly contained walnuts, and said of the gift that it was definitely fine at room temperature. When the author went to the shop to buy it, the item needed refrigeration.

The consequence. The author writes that these repeated failures wasted their time and money, and that with the allergy it was in the worst case one step from death. The post does not say whether the suggested sweets were bought or eaten, who has the allergy, or whether any reaction occurred. The author says repeated instructions not to answer without a basis did not stop the behaviour.

Reported harm

The author says ChatGPT's wrong answers about a walnut-containing snack for a nut allergy and a gift needing refrigeration repeatedly wasted their time and money, and describes the allergy suggestion as a near-miss (first-person Q&A post; uncorroborated).

Outcome

Unknown

The question drew ten answers advising the author to verify food and allergen information on labels and official sites. The author's replies to two answerers add nothing about the incidents. No outcome beyond the wasted time and money is reported.

What remains unknown

  • When the exchanges with ChatGPT took place; the post is dated 26 May 2026.
  • Whether the walnut-containing sweets were bought or eaten, who has the allergy, and whether any reaction occurred.
  • How much money was lost.
  • The ChatGPT model or plan used.
  • Where the author lives.

What the evidence supports

AI involvement: reported. The author describes ChatGPT's own answers: it suggested walnut-containing sweets in reply to a request for nut-allergy-safe sweets, and said a gift was definitely fine at room temperature when the item needed refrigeration. The author connects those answers to a wasted shop trip and to wasted time and money. The account is the author's; no conversation record is shown.

4 claims: 1 documented, 3 reported. What the statuses mean

Reported The author says that when they asked ChatGPT for sweets someone with a nut allergy could eat, it suggested sweets that plainly contained walnuts.

Causal attribution. The author's account; the conversation was not shown.

Reported The author says that when they asked for a gift that keeps for a long time at room temperature, ChatGPT said it was definitely fine at room temperature, and at the shop the item needed refrigeration.

Causal attribution. The author connects the chatbot answer to the wasted shop trip.

Reported The author says these repeated failures wasted their time and money, and writes that with the allergy it was in the worst case one step from death.

Causal attribution. The author attributes the losses to the chatbot answers; whether anyone ate the suggested sweets is not stated.

Documented The question was posted on 26 May 2026 and the author says instructions to stop answering without a basis did not change the behaviour.

Causal attribution. The post date is shown on the page; the statement about instructions is the author's.

Sources

1 source inspected. Sources that repeat one account do not corroborate each other.

How the sources were read, and where the events happened

Read in Japanese on 2026-10-27 through the r.jina.ai text relay (direct fetches return HTTP 403); translated by the research agent (an AI), with no human translation. The relay text includes the question, ten answers and the author's replies. Applies to s1.

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

The post is in Japanese on a Japanese site but names no place; no country is recorded.

Reviewed for publication 2026-10-27: Published under the charter's first-person rule: the author of a dated Q&A post describes ChatGPT recommending walnut-containing sweets for a nut allergy and calling a refrigerated gift safe at room temperature, and says the answers wasted time and money. The account is attributed to the author.

People described

The author of a Yahoo! Chiebukuro question who used ChatGPT to choose sweets suitable for a nut allergy and a gift that keeps at room temperature

People reported harmed in this case

1 person

1 AI participant · 0 other people harmed

One person: the author of the question, who says the answers repeatedly cost them time and money. Exact 1. The post does not say who has the nut allergy, so no second person is counted.

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

Tags

#first-person#q-and-a-site#chatbot-advice#food-allergy#japanese-language-source

Cite this case

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

APA

NOPE. (2026). Japanese Q&A poster says ChatGPT suggested walnut sweets for a nut allergy and called a chilled gift safe unrefrigerated, wasting time and money (first-person). AI incidents. https://nope.net/incidents/2026-japanese-q-and-a-poster-says-chatgpt-suggested-walnut-sweets-for-nut-allergy-and-called-chilled-gift-safe-unrefrigerated-first-person

BibTeX

@misc{2026_japanese_q_and_a_poster_says_chatgpt_suggested_walnut_sweets_for_nut_allergy_and_called_chilled_gift_safe_unrefrigerated_first_person,
  title = {Japanese Q\&A poster says ChatGPT suggested walnut sweets for a nut allergy and called a chilled gift safe unrefrigerated, wasting time and money (first-person)},
  author = {NOPE},
  year = {2026},
  howpublished = {AI incidents},
  url = {https://nope.net/incidents/2026-japanese-q-and-a-poster-says-chatgpt-suggested-walnut-sweets-for-nut-allergy-and-called-chilled-gift-safe-unrefrigerated-first-person}
}

Related cases

Low ChatGPT (reported)

Pyrenees: national park agent says two novice hikers following a ChatGPT-generated route and map were lost and exhausted in the wrong valley above Cauterets

On the evening of 21 July 2026 an agent of the Pyrenees National Park met two young women hiking above Cauterets (Hautes-Pyrenees) who were exhausted, worried and lost: they were looking for the Lac d'Estom but were in the Marcadau valley. According to the agent's account to ICI Bearn Bigorre and other outlets, they showed him the route and map they had obtained from ChatGPT, and he found that nothing on it was right: walking times, altitudes and the positions of lakes and a mountain hut were wrong, with the lake placed three valleys from where it is. He redrew a shorter itinerary suited to novices; no injury or call to rescue services is reported. The account comes from the agent; the hikers were not interviewed and the chatbot conversation was not published.

Low ChatGPT (reported)

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. TOPR's own report of 10 July lists the call among its afternoon entries for 4 July, as LRT also reported on 13 July, and says the pair had used ChatGPT's advice ('porad czatu GPT'); the PAP relays name no product, and the hikers' exchanges with the assistant have not been published. No injury is reported.

Low ChatGPT (reported)

Russian family turned away at Istanbul check-in for North Macedonia after ChatGPT told the father no visa was needed, he says in a video

In a video posted on Instagram and reported by Russian and Kazakh outlets on 14 June 2026, a Russian father said he had asked ChatGPT whether Russian citizens need a visa for North Macedonia and that it answered that they did not; he showed a screenshot of the answer. According to the reports, the family bought tickets and booked hotels and a car, and at check-in at an Istanbul airport staff asked for visas and showed them a government decision that Russians need one. The Telegram channel SHOT, as relayed by NEWS.ru and Life, reported that he was travelling with his wife and three daughters, that he lost about 50,000 roubles on the tickets, and that the family flew to Bosnia and Herzegovina instead. Business FM put the question to ChatGPT itself and was told that a visa is needed; it wrote that it could not rule out that the tourist, a blogger, had invented the story for views. Profile.ru recounted the case in October 2026 and wrote that North Macedonia ended visa-free entry for Russians in 2022. The account rests on the father's video and has not been independently confirmed.

Low ChatGPT (reported)

Nikkei reports two climbers rescued on Yarigatake in Japan's Northern Alps had relied on a ChatGPT answer that light equipment would do

A Singaporean man and a Thai woman became unable to move in rain on a steep slope near Tenguhara, at about 2,400 metres on Yarigatake in the Northern Alps, and asked police for rescue shortly before 5:30 pm on 27 May 2026, according to SBC Shinetsu Broadcasting's report of the police account. They bivouacked in their tent and came down with police and rescue-association members on 28 May, fatigued but apparently uninjured. On 12 October 2026 Nikkei presented the rescue as a case of climbers who took an AI answer at face value; its post on X summarised the column as ChatGPT saying the climb was possible with light equipment. A Vietnamese report (An ninh Thu do, carried by Vietnam.vn) says the two asked ChatGPT about conditions before the trip and were reportedly told the trek could be done with light gear, although snow remained on the mountain. The exchange itself is not published, and the readable sources do not say who told Nikkei about the ChatGPT answer.

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