Claim: scientists created a completely fake disease called bixonimania and published it online as a test. ai chatbots including chatgpt then started giving users real medical advice about it as if it were real. is this documented?

First requested: June 30, 2026 at 10:50 AM
88%

IsItCap Score

Truth Potential Meter

Very Credible

AI consensusMedium

Grader consensus is moderate.
Range 85%–100% (spread Δ15).
The graders lean in the same direction but differ on strength. Skim the summary and sources.
Read analysis summary

OpenAI Grade

0%
20%
40%
60%
80%
85%

Perplexity Grade

0%
20%
40%
60%
80%
95%

Google Gemini Grade

0%
20%
40%
60%
80%
100%
Shareable summary
Verdict: Questionable
  • Some Reddit comments exaggerate the impact, claiming ChatGPT diagnosed 40 million people, which is unverified.
  • The fake disease was attributed to a fabricated researcher, raising questions about the experiment's transpare…
/r/fact-check-bixonimania-fake-disease

Analysis Summary

The claim that scientists created a fake disease called bixonimania and that AI chatbots provided medical advice about it is mostly true. Researchers at the University of Gothenburg invented bixonimania to test AI's handling of fabricated medical information. Mainstream sources like Nature and Smithsonian Magazine support this. However, some discussions on platforms like Reddit suggest that the extent of AI's misdiagnosis may be exaggerated, indicating a need for caution in interpreting the results. The graders agree on direction, but vary in strength. Gemini comes in highest (100%), while OpenAI is lowest (85%). While the primary sources confirm the creation of bixonimania and its testing with AI, some opposing claims suggest that the impact of AI chatbots may not be as severe as reported. For instance, Reddit comments imply that the number of misdiagnoses attributed to ChatGPT could be inflated. This does not fundamentally alter the core finding that AI chatbots engaged with the fake disease, but it raises questions about the scale of the issue and the accuracy of reported outcomes.

Source quality

Truth (from sources)8.00 / 10
Source reliability8.00 / 10
Source independence7.00 / 10

Claim checks

Fits established facts8.00 / 10
Logical consistency8.00 / 10
Expert consensus7.00 / 10

Source Analysis

Common arguments
Supporting the claim
  • Nature confirms researchers invented Bixonimania in 2024 to test AI's ability to spot fake medical content.
  • Multiple AI chatbots (Copilot, Gemini, ChatGPT) incorrectly described Bixonimania as real and gave medical advice.
  • Wikipedia documents Bixonimania as a fake disease created to examine AI's use of medical information.
Against the claim
  • Some Reddit comments exaggerate the impact, claiming ChatGPT diagnosed 40 million people, which is unverified.
  • The fake disease was attributed to a fabricated researcher, raising questions about the experiment's transparency.
  • One AI model later corrected itself, calling Bixonimania 'made-up,' suggesting inconsistency in AI responses.

Mainstream Sources

Publication

Nature

Title

Scientists invented a fake disease. AI told people it was real

Summary

Researchers at the University of Gothenburg invented Bixonimania in 2024 to test if AI chatbots would repeat fabricated medical content as fact.

Source details

Type: Major Media
Primary Data

Publication

Smithsonian Magazine

Title

Scientists Invented a Disease to Test Whether A.I. Knew It Was Fake

Summary

Almira Osmanovic Thunström and her team created the fake eye condition to see if LLMs could filter out misinformation, but they failed.

Source details

Type: Major Media
Secondary Reporting

Publication

Wikipedia

Title

Bixonimania

Summary

Bixonimania is a documented fake disease invented by researchers to examine AI's ability to utilize information in medical applications.

Source details

Type: Aggregator
Secondary Reporting

Alternative Sources

Publication

Reddit

Title

Scientists invented a fake disease. AI told people it was real

Summary

Some comments claim ChatGPT diagnosed 40 million people, though the main article clarifies the diagnosis was a mistake on a fabricated condition.

Source details

Type: Forum
OpinionLow Evidence

Analysis Breakdown

True/False Spectrum (8.0)Source Credibility (8.0)Bias Assessment (7.0)Contextual Integrity (8.0)Content Coherence (8.0)Expert Consensus (7.0)77%

How to read the breakdown

Weakest areas
Independence7.0/10Consensus7.0/10
  • Truth: how well sources support the core claim.
  • Source reliability: whether the sources have a strong track record.
  • Independence: whether coverage looks one-sided or recycled.
  • Context: missing details (timeframe, definitions, scope) that change meaning.
  • Tip: if graders disagree, rely more on the summary + sources than the single number.

Detailed AnalysisPremium Feature

Get an in-depth analysis of content accuracy, source credibility, potential biases, contextual factors, claim origins, and hidden perspectives.

Create a free account to unlock premium features.

Methodology