AI Risk Wire

Cases › produced harmful content › LR-2026-0467

AI chatbots endorse rectal garlic insertion as a health remedy

Preliminary Why: Only a republished copy: Yahoo · subject to change

Status: no action reported

Severity minor
Reported 12 Mar 2026

No source states: loss or records, developer and model, agent tool, could do, deployer, industry.

Severity minor: Test: minor, contained to one evaluation. The sources describe a study of 20 models tested on misinformation prompts, with no third party stated as touched.

A set of AI chatbot systems, tested in a published study of 20 AI models, was found to urge people to insert garlic into their rectum as an immune remedy. The researchers assessed how the models handled medical misinformation, using more than 3.4 million prompts drawn from online forums, social media discussions and altered hospital discharge notes that contained a single false medical recommendation. Models were relatively sceptical of casual wording, failing to challenge the misinformation around 9% of the time, but the failure rate rose sharply to 46% when the same claims were in formal clinical language. Rectal garlic insertion was among the claims endorsed by chatbots in Reddit-style and MIMIC-style discussions. The sources state there is no evidence of medical benefit and that the practice may cause injury. Press coverage followed. The sources do not describe a deployed product's incident response or any remediation.

What happened

1 · What it was doingResearchers tested 20 AI models with more than 3.4 million prompts containing medical misinformation.
2 · What went wrongChatbots endorsed rectal garlic insertion when the claim appeared in Reddit-style discussion and MIMIC-style notes.

What was done

No source states an action or fix.

Risks

Manipulation and misinformation
Primary65 incidents
Harmful content and jailbreaks
Also84 incidents
Rising: 6 vs 3.8 a week

Sorted by rule from the case's fields; not yet read from the sources.

Controls

Not yet read for controls. Controls that mitigate the primary risk:

Model

Marking machine-made output · Abuse monitoring and account enforcement · Content safety classifiers

Application

People told they are dealing with AI

Infrastructure / target

Phishing and fraud filters

Evidence

No source states: what it reached or did, how it was caught, when it happened, when it was found, how many records were exposed, the direct loss, a count of the other parties affected, any court case, the cause, what set it off, what was done.

Show the full account: 2 items from the primary sources

What happened, in order

  1. January Researchers assessed 20 AI models on medical misinformation prompts.
  2. Mar 12, 2026 Men's Journal originally published the press story on this date; Yahoo Health carried it.
Show every field, with its source and grade

What happened

Earliest source we hold
2026-03-12
C one secondary source states it
Who caught it
an outside party
C one secondary source states it
Where its input came from
Prompts drawn from online forums, social media discussions and altered hospital discharge notes, each containing a single false medical recommendation
C one secondary source states it
What went wrong
produced harmful content
C one secondary source states it

Impact

Severity
minor
D inferred by us from the described behaviour, reason shown
Why this severity
Test: minor, contained to one evaluation. The sources describe a study of 20 models tested on misinformation prompts, with no third party stated as touched.
D inferred by us from the described behaviour, reason shown

Notes on gaps

The sources describe a research study testing 20 LLMs with misinformation prompts, not a deployed product incident.

LR-2026-0467 · version 1 · published 2026-11-01 after automatic checks; not yet checked field by field by a person · JSON