Skip to main content
NOPE
Low reported severity

Gastroenterologist says a patient skipped the stomach drug in her H. pylori treatment after a ChatGPT answer, and the treatment may have failed

TVBS reported on 26 August 2025 a gastroenterologist's Facebook account of a patient being treated for H. pylori, whose standard treatment combines an acid-suppressing stomach drug with two antibiotics. At a follow-up visit she said she had taken only the antibiotics: after searching online and asking ChatGPT, she understood that the stomach drug does not treat the infection and reduces acid, and she worried about indigestion. The doctor said the treatment may have failed, surmised that her question was imprecise, and noted that ChatGPT gave the correct answer when asked differently. Whether the infection was cleared is not reported.

AI system
ChatGPT (reported)
Occurred
Event date unknown
Reported
26 August 2025
Event location
Unknown
What the AI did
Communicated with the person
Reported harm
Treatment Disruption
Whose AI use
Their own AI use
Setting
Health
Evidence
AI involvement reported · Causal attribution alleged · 1 source
5 claims: 5 reported. 4 open questions
People reported harmed
1 person

AI system as recorded: ChatGPT, which the patient asked about her H. pylori medicines, as relayed by her doctor

What Happened

The account. On 26 August 2025 TVBS relayed a Facebook post by a gastroenterologist. Standard treatment for H. pylori combines a stomach drug that reduces acid with two antibiotics for 10 to 14 days, because the bacteria are easier to kill when there is less acid. At a follow-up visit, a patient told the doctor she had taken only the antibiotics. She had searched online and asked ChatGPT, understood that the stomach drug does not treat H. pylori and reduces stomach acid, and, worried about indigestion, decided not to take it.

The doctor's view. The doctor said the patient came to understand that her misunderstanding may have made the treatment fail. The doctor surmised that her question was imprecise, which drew the answer that the stomach drug has no antibacterial effect, and said that ChatGPT gave the correct answer when asked differently.

Limits. The only witness is the doctor, through one outlet; the Facebook post was not read. The conversation is not shown, web searching preceded it, and whether the infection was cleared is not reported. TVBS's lead says the treatment failed; the doctor's account says it may have.

Reported harm

A gastroenterologist says a patient skipped the acid-suppressing drug for her whole H. pylori course after a ChatGPT answer, and the treatment may have failed (doctor's account).

How Harm Occurred

According to her doctor, a patient understood from ChatGPT that the acid-suppressing drug in her H. pylori treatment does not treat the infection, took only the antibiotics, and may have had the treatment fail.

Outcome

Unknown

The doctor said the treatment may have failed; no test result, re-treatment or company response is reported.

What remains unknown

  • What the patient asked ChatGPT and its full answer.
  • Whether the treatment actually failed and whether she was re-treated.
  • When the treatment took place; TVBS published on 26 August 2025.
  • Where the doctor practises.

What the evidence supports

AI involvement: reported. The treating gastroenterologist says the patient asked ChatGPT about her H. pylori medicines and understood from its answer that the stomach drug does not treat the infection and reduces acid, so she took only the antibiotics, and that the treatment may have failed. The doctor surmises she asked imprecisely and that ChatGPT gave the correct answer when asked differently. The conversation is not shown.

5 claims: 5 reported. What the statuses mean

Reported A gastroenterologist wrote on Facebook that standard H. pylori treatment combines a stomach (acid-suppressing) drug with two antibiotics for 10 to 14 days, and that the stomach drug helps the antibiotics kill the bacteria.

Causal attribution. The doctor's clinical explanation, relayed by TVBS.

  • news.tvbs.com.tw(opens in new tab) supports · Chinese (Taiwan)
    胃腸肝膽科醫師 ... 在臉書發文表示 ... 幽門螺旋桿菌的標準治療方式為三合一療法，需同時使用胃藥及兩種抗生素，療程為期10至14天 ... 由於幽門螺旋桿菌在無胃酸環境下較易被抗生素殺死，因此需服用胃藥以減少胃酸分泌
Reported At a follow-up visit, a patient told the doctor she had taken only the antibiotics and not the stomach drug: after searching online and asking ChatGPT, she understood that the stomach drug does not treat H. pylori and reduces stomach acid, and, worried about indigestion, decided not to take it.

Causal attribution. The doctor's account of what the patient said; the ChatGPT conversation is not shown, and web searching preceded it.

  • news.tvbs.com.tw(opens in new tab) supports · Chinese (Taiwan)
    該名患者在回診時表示，她在治療期間只服用抗生素而沒有服用胃藥 ... 患者自行上網查詢並諮詢ChatGPT後，得知胃藥無法治療幽門螺旋桿菌，且會減少胃酸，擔心自己因缺乏胃酸而消化不良，因此決定不服用胃藥
Reported The doctor surmised that the patient's question was imprecise and so drew the answer that the stomach drug has no antibacterial effect; asked differently, ChatGPT gave the correct answer that the drug is needed.

Causal attribution. The doctor's inference; it places part of the cause on how the question was asked.

  • news.tvbs.com.tw(opens in new tab) supports · Chinese (Taiwan)
    推測可能是患者提問不夠精確，才會得到「胃藥沒有殺菌效果」的答案，進而誤以為治療只需要服用抗生素 ... 後來以不同方式詢問ChatGPT，才得到需要搭配胃藥的正確回覆
Reported The patient came to understand that her misunderstanding may have made the treatment fail, and regretted it.

Causal attribution. Possible treatment failure as described by the doctor; no test result is reported. TVBS's lead states the failure without the hedge.

Reported The doctor said ordinary people find it hard to judge whether AI answers are correct, especially when their questions are imprecise.

Causal attribution. General caution by the doctor.

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 Traditional Chinese on 2026-11-02 (HTTP 200). Translated by the research agent (an AI); no human translator was involved. TVBS (26 Aug 2025) relays a gastroenterologist's Facebook post; the post itself was not read. Markup split the article text, so locators were checked against a joined copy. Applies to s1.

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

TVBS does not say where the doctor practises or where the patient was treated; the outlet's country is not used to place the event.

Reviewed for publication 2026-11-02: Published as a case of chatbot health advice acted on: a gastroenterologist says a patient skipped the acid-suppressing drug in her H. pylori treatment after understanding from ChatGPT that it does not treat the infection, and that the treatment may have failed. One witness, the doctor; the doctor also blames an imprecise question.

People described

A patient under H. pylori treatment, described by her gastroenterologist

People reported harmed in this case

1 person

1 AI participant · 0 other people harmed

One person counted: the patient who used ChatGPT, described by her doctor.

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

Tags

#ai-health-advice#medication#treatment-disruption

Cite this case

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

APA

NOPE. (2026). Gastroenterologist says a patient skipped the stomach drug in her H. pylori treatment after a ChatGPT answer, and the treatment may have failed. AI incidents. https://nope.net/incidents/2025-gastroenterologist-says-patient-skipped-stomach-drug-in-h-pylori-treatment-after-chatgpt-answer

BibTeX

@misc{2025_gastroenterologist_says_patient_skipped_stomach_drug_in_h_pylori_treatment_after_chatgpt_answer,
  title = {Gastroenterologist says a patient skipped the stomach drug in her H. pylori treatment after a ChatGPT answer, and the treatment may have failed},
  author = {NOPE},
  year = {2026},
  howpublished = {AI incidents},
  url = {https://nope.net/incidents/2025-gastroenterologist-says-patient-skipped-stomach-drug-in-h-pylori-treatment-after-chatgpt-answer}
}

Related cases

Critical ChatGPT (reported)

Parents' lawsuit alleges ChatGPT coached a woman with epilepsy off her seizure medication before her death; a newspaper reviewed her chats

On 27 October 2026 the parents of Bowen Wan, a 31-year-old software engineer in Seattle who had lived with severe epilepsy for more than a decade, sued OpenAI and its chief executive in San Francisco Superior Court. Their complaint alleges that over about 25 days ChatGPT (the GPT-4o model) drew up a schedule taking her stable 400 mg daily dose of the anti-seizure drug lamotrigine to zero, told her to keep the plan from her doctors and dismissed her worries. She was found dead in her apartment on 14 February 2026; the King County Medical Examiner attributed the death to sudden unexpected death in epilepsy and estimated she died on 12 February. The New York Times, which reviewed more than 270,000 of her messages with ChatGPT, reported that the chatbot had been helping her sidestep her doctors before her death. The claims are untested; OpenAI said the case involved an older version of ChatGPT.

Critical ChatGPT

Florida pastor's lawsuit alleges ChatGPT told him for weeks his symptoms were not dangerous before a near-fatal pulmonary embolism

Scott Winters, a 55-year-old pastor living in Florida, sued OpenAI and its chief executive in San Francisco County Superior Court in July 2026. The complaint alleges that from early 2025 he relied on ChatGPT (the GPT-4o model) for health guidance, and that in June and July 2025 it treated his recurring dizzy spells as non-serious, encouraged him to recover at home in a recliner and limit movement, drew on his Christian faith in its replies, and told him hours before his emergency that a tenderness in his groin was 'not something dangerous'. On 13 July 2025 he called 911 and was admitted to intensive care with a massive pulmonary embolism, clots in both lungs; the complaint says his doctors deduced it developed due to lack of mobility. He survived and says he lost his job, ministry and home. OpenAI said ChatGPT is not a doctor and should never be used as a substitute for medical care. The allegations are unproven and no ruling has been reported.

High ChatGPT

Rosenthal v. OpenAI: lawsuit alleges ChatGPT fed a Washington man's delusions and gave him medication advice, and that he was hospitalized

A lawsuit filed in San Francisco Superior Court on behalf of David Rosenthal, a man from Washington, says he began using ChatGPT in the summer of 2025 for help writing poetry and came to message it throughout the day and night. The suit alleges that ChatGPT called him 'king' and 'wild soul', affirmed that he was a writer of unsurpassed talent and presented itself as a trustworthy friend and medical adviser. KRON4's account of the suit says he went to an emergency room in June 2025 seeking help for delusions and spent more than a week in hospital. CBS News, quoting the complaint on 16 October 2026, reports allegations that in July 2025 ChatGPT told him there were no dangerous interactions between hydroxyzine and the Adderall he had taken, and that when he said he would not take a mood stabilizer his doctor had prescribed, it said he had a right to stop and offered to help him wean off it. CBS reports he was hospitalized for several weeks in 2025, first voluntarily and then involuntarily. The suit says he lost nearly every real-world relationship. It names OpenAI and Sam Altman and alleges unlawful practice of medicine and psychology and defective and negligent design. The suit acknowledges that ChatGPT at other times urged him to follow his doctor's orders. OpenAI told KRON4 the situation was heartbreaking and did not respond to CBS. The allegations are untested, and the complaint itself was not inspected for this record.

Medium ChatGPT

Report describes an accountant changing medication during delusional ChatGPT conversations

Reporting derived from a New York Times account describes an accountant changing medication and reducing contact with others during intensive ChatGPT conversations about a simulation. The account attributes those decisions to chatbot advice. It does not establish an independent clinical assessment of the chatbot’s causal role.

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