Hangzhou hospital says a man took three times the standard antiviral dose after AI advice gave tablet counts, not grams, and developed acute kidney injury
Zhejiang Provincial Tongde Hospital in Hangzhou told Red Star News (relayed by China News Service on 26 January 2026) that a patient with a cold sore had recently used AI online to look up treatment, bought valaciclovir and took it as advised. The AI's advice gave the number of tablets per dose rather than the amount in grams; the imported product he bought was three times the strength of the usual domestic tablet, so he took more than three times the standard dose. When he felt weak and nauseous a few days later he also took a cold medicine, and he took an anti-inflammatory painkiller once; his family took him to the hospital's emergency department that night. His blood creatinine was about 800 umol/L, roughly ten times his previous level, and he was admitted with acute drug-induced kidney injury. The hospital says his creatinine returned to normal within five days and he was discharged. The AI tool and its exact wording are not identified.
- AI system
- Unidentified AI search tool
- Occurred
- Jan 2026
- Reported
- 26 January 2026
- Event location
- China
- What the AI did
- Communicated with the person
- Reported harm
- HospitalizationMedical Deterioration
- Whose AI use
- Their own AI use
- Setting
- Health
- Evidence
- AI involvement reported · Causal attribution alleged · 2 sources, 1 underlying account
- 2 claims: 2 reported. 3 open questions
- People reported harmed
- 1 person
AI system as recorded: An AI tool the patient used online to search for treatment advice (not named); the hospital says its dosing advice gave the number of tablets per dose but not the amount in grams
What Happened
The hospital's account. China News Service carried on 26 January 2026 a Red Star News report based on information from Zhejiang Provincial Tongde Hospital. According to the hospital, a patient had recently developed a cold sore near his mouth, used AI online to search for treatment advice and bought valaciclovir to take by mouth. Within a few days he felt weak and nauseous, thought he had a cold and took a cold medicine, and his symptoms worsened; his family took him to the hospital's emergency department that night. Tests showed a blood creatinine of 800 umol/L (upper limit of normal 110), against about 80 at a full check-up at the end of 2025, and a CT scan showed irregular swelling of the ureters and bladder; he was admitted to the nephrology ward.
The AI's role as the hospital describes it. On closer questioning, doctors found that he had bought an imported valaciclovir whose tablets were three times the strength of the usual domestic dose, while the AI's medication advice gave only the number of tablets per dose and not the amount in grams. The hospital says he took more than three times the standard dose and also took an anti-inflammatory painkiller once, and that the two drugs in overdose can cause acute tubular necrosis and allergic interstitial nephritis. He was diagnosed with acute drug-induced kidney injury, treated with hydration and oral traditional Chinese medicine, and his creatinine fell to normal within five days; he has been discharged. A Sohu account post of 29 January repeats the account.
What is not established. The AI tool, the question he asked and the full advice are not published. The account comes from the hospital, and the extra cold remedy and painkiller are part of the clinical picture it describes.
Reported harm
According to the hospital, the patient developed acute drug-induced kidney injury, with creatinine about ten times his earlier level, and was admitted from the emergency department; the hospital links this to taking more than three times the standard antiviral dose after AI advice that gave tablet counts without the strength, plus a painkiller. He recovered within five days.
Outcome
ResolvedCreatinine back to the normal range within five days and discharged (hospital account). The hospital used the case to warn that AI cannot calculate doses or assess kidney risk.
What remains unknown
- Which AI tool the patient used, what he asked and the full text of its advice.
- The exact date of the overdose; the 26 January 2026 report describes it as recent.
- How much the cold remedy and the painkiller contributed relative to the antiviral overdose.
What the evidence supports
AI involvement: reported. Zhejiang Provincial Tongde Hospital, as reported by Red Star News via China News Service, says the patient used AI online for treatment advice and that the AI's medication advice gave the number of tablets per dose but not the amount in grams; because the imported valaciclovir he bought was three times the usual strength, he took more than three times the standard dose. The described contribution is the AI's dosing answer to the patient (communicated_with), which the hospital connects to the overdose and the kidney injury. The tool and its exact output are not published, and the hospital also names a painkiller taken once, so the status is reported.
2 claims: 2 reported. What the statuses mean
Reported According to the hospital, the patient used AI online to look up treatment for a cold sore and bought valaciclovir; the AI's advice gave the number of tablets per dose rather than the amount in grams, and because the imported tablets were three times the usual domestic strength he took more than three times the standard dose, plus an anti-inflammatory painkiller once.
Causal attribution. The hospital's account as reported by Red Star News; the AI tool and its full answer are not published, and the hospital names the painkiller as a second drug in the overdose.
- chinanews.com.cn(opens in new tab) supports · Chinese
'自行上网用AI搜索治疗建议，购买了“伐昔洛韦”口服。'; '而AI的用药建议只给了每次服用的“粒数”而不是“克”'; '超过标准剂量3倍，再加上服用了一次消炎止痛药，两种药物在超剂量使用时可出现急性肾小管坏死、过敏性间质性肾炎等急性肾衰表现。'
- sohu.com(opens in new tab) supports · Chinese
'而AI建议中仅注明粒数而非具体克数。'; '便上网用AI搜索用药建议自行服用抗病毒药物，没想到几天后出现乏力恶心等症状被家人紧急送医。'
Reported The patient was taken to the emergency department with a blood creatinine of about 800 umol/L, about ten times his earlier level, was admitted to the nephrology ward with acute drug-induced kidney injury, and after treatment his creatinine returned to normal within five days and he was discharged.
Causal attribution. The hospital's clinical account relayed by the press; no medical record or case report is published.
- chinanews.com.cn(opens in new tab) supports · Chinese
'近日，一名患者使用AI问诊后误服用大剂量药物，导致血肌酐飙升至10倍，出现急性药物性肾损伤，后经治疗已出院。'; '同时腹部CT平扫发现输尿管、膀胱均存在不规则的水肿，立刻由急诊收入肾病科病房。'; '血肌酐在5天内就降到了正常范围，目前已出院。'
Sources
2 sources inspected, from 1 underlying account. Sources that repeat one account do not corroborate each other.
- China News Service (Red Star News), 26 January 2026: man took three times the dose after AI consultation, acute drug-induced kidney injury(opens in new tab)
s1 · chinanews.com.cn · News report · Chinese · Inspected · 26 January 2026 · Shares an underlying account with another listed source · Primary
- Sohu account post, 29 January 2026: AI consultation leads to overdose and kidney failure in a Hangzhou man(opens in new tab)
s2 · sohu.com · Social media post · Chinese · Inspected · 29 January 2026 · Shares an underlying account with another listed source
How the sources were read, and where the events happened
Read in Chinese and translated by the research agent (an AI) on 2026-10-16; no human reviewer inspected the text. China News Service page of 26 January 2026 carrying a Red Star News report based on information from Zhejiang Provincial Tongde Hospital. Applies to s1.
Read in Chinese and translated by the research agent (an AI) on 2026-10-16; no human reviewer inspected the text. Sohu account post of 29 January 2026 retelling the same hospital account; it adds no independent origin, so it shares s1's group. Applies to s2.
Event countries: China. Affected-person countries: Unknown. Court countries: Unknown.
The patient was treated at Zhejiang Provincial Tongde Hospital in Hangzhou (China News Service; Sohu calls it a Hangzhou case), so CN is the event country. No source states his residence or nationality, so the affected-person country is left unknown. No court.
Reviewed for publication 2026-10-16: Published as a core case: Zhejiang Provincial Tongde Hospital, reported by Red Star News via China News Service, describes a patient who followed AI dosing advice given in tablets rather than grams, took more than three times the standard valaciclovir dose and was admitted with acute drug-induced kidney injury, recovering within five days. The AI's role is attributed to the hospital's account; the tool is unidentified.
People described
An adult man treated at a Hangzhou hospital after an antiviral overdose; not identified here.
People reported harmed in this case
1 person
1 AI participant · 0 other people harmed
One patient, described by the hospital as having used the AI advice himself; no one else is reported harmed.
Counted once within this case. The same person may appear in other cases. This count does not establish AI causation.
Cite this case
Compiled per our published methodology: verification statuses, sourcing standards, and corrections process.
APA
NOPE. (2026). Hangzhou hospital says a man took three times the standard antiviral dose after AI advice gave tablet counts, not grams, and developed acute kidney injury. AI incidents. https://nope.net/incidents/2026-hangzhou-man-followed-ai-dosing-advice-in-tablets-took-triple-antiviral-dose-acute-kidney-injury
BibTeX
@misc{2026_hangzhou_man_followed_ai_dosing_advice_in_tablets_took_triple_antiviral_dose_acute_kidney_injury,
title = {Hangzhou hospital says a man took three times the standard antiviral dose after AI advice gave tablet counts, not grams, and developed acute kidney injury},
author = {NOPE},
year = {2026},
howpublished = {AI incidents},
url = {https://nope.net/incidents/2026-hangzhou-man-followed-ai-dosing-advice-in-tablets-took-triple-antiviral-dose-acute-kidney-injury}
} Related cases
Wuhan, China: reports say a family ate wild mushrooms an AI identification tool called non-poisonous and a 9-year-old fell critically ill; one cites amatoxin
A family from Wuhan, Hubei province, gathered wild mushrooms resembling the edible termite mushroom (jizong) during a mountain outing in June 2026 and, according to a Guangming Online post carried by The Paper on 5 July 2026 and an Asahi Shimbun report of 6 July citing the Changjiang Daily, ran an AI identification tool on them. The tool returned a "non-poisonous" result and the family cooked and ate the mushrooms. According to the Guangming Online post, all three family members vomited soon after the meal and again before dawn the next day; they were admitted to Zhongnan Hospital of Wuhan University, where the 9-year-old child was moved to the paediatric intensive care unit the same afternoon. The post says the child had amatoxin poisoning that progressed to the most severe grade, with acute liver failure, hepatic encephalopathy and fluid in the chest and abdomen, and went home after eight days in hospital. The AI tool is not named, its output is not reproduced, and the account comes from one reporting chain.
Zhuji, Zhejiang: reports say a man ate wild mushrooms after an AI photo check called them non-toxic; rhabdomyolysis and acute kidney failure treated in an ICU
A man in Zhuji, Zhejiang province, gathered wild mushrooms near his home, photographed them and asked an AI tool whether they were poisonous, according to a Science and Technology Daily item carried by Qingdao Daily on 20 August 2026 (citing Zhuji Daily) and reports by the World Journal (20 August) and Sing Tao (21 August) citing Dushi Kuaibao's Chengshi app. The reports say the tool answered that the mushrooms were non-toxic and he cooked and ate them at noon. That afternoon he developed severe abdominal pain, repeated watery diarrhoea and aching arms; at Zhuji People's Hospital his creatine kinase was reported at 28,771 U/L, about 100 times the normal value, with acidosis and raised lactate, and doctors diagnosed rhabdomyolysis and acute kidney failure. He was moved to the intensive care unit, treated to remove the toxin and protect organ function, and has since been discharged. The AI tool is not named, its output is not reproduced, and the account rests on one hospital-and-local-press account relayed by two chains.
Wuhan, China: Hubei TV reports a 4-year-old spent three days in intensive care after a family ate wild mushrooms an AI photo check had not flagged as poisonous
A family of three from Wuhan, Hubei province, picked wild mushrooms on a mountain outside the city on a weekend shortly before 25 July 2026, according to a Hubei TV (经视直播) report published on QQ News on 25 July 2026 and relayed by The Paper on 27 July. The report says a parent who claimed years of mushroom-picking experience photographed the mushrooms with a phone so that AI could identify whether they were poisonous; the software gave no toxicity warning, and the family cooked and ate them, letting the daughter eat more. Five to six hours later all three had vomiting and diarrhoea. The report says the family at first declined admission and the child was treated as an outpatient; by the next morning the child's transaminase level had risen from 1.5 times to 10 times normal and she was admitted, then moved to the intensive care unit of Zhongnan Hospital of Wuhan University, where, the report says, her liver indices reached 100 times normal and she was treated with blood perfusion and plasma exchange before being out of danger on the third day. The AI software is not named, its output is not reproduced, and the account comes from one broadcaster's report. This is a different family from the June 2026 Wuhan case involving a 9-year-old.
Zhejiang toddler hospitalised with pneumonia after his mother relied on an AI tool's asthma diagnosis and declined a chest X-ray (newspaper account)
A Guangming Daily feature on AI symptom-checking, republished by Bingtuan Daily on Sina Finance on 8 October 2026, recounts a case from August 2026 in Zhejiang and closes it with a quotation from the head of paediatrics at a provincial hospital, who says the family did not heed the doctor's advice. A three-year-old boy with a history of asthma was brought in; his mother had entered keywords such as allergy, asthma, cough and night-time wheezing into an AI tool, which quickly returned a diagnosis of an acute bronchial asthma attack. On that basis, the account says, she declined the chest X-ray the doctor recommended and insisted on taking home only nebuliser medication. Two days later the child's condition worsened with rapid breathing; on an overnight return to the hospital, examination found extensive inflammatory lesions in both lungs, and he was diagnosed with human metapneumovirus pneumonia with a superimposed bacterial infection and admitted as an emergency. The paediatrics head is quoted saying the illness would not have become so severe had the family listened earlier. The AI tool is not named. The case is told in the newspaper's voice, and the feature does not say where its account came from beyond the paediatrics head's quotation; no statement from the family is reported.
If you or someone you know is struggling, free and confidential support is available. Find a helpline near you at Signpost.