Based on your description of "chronic neurovascular fatigue syndrome" (CNFS), this condition appears to involve dysregulation of cerebral blood flow and autonomic nervous system dysfunction. Symptoms like brain fog after standing, cold extremities, and exercise intolerance align with a neurovascular pathophysiology. The underlying mechanism likely involves impaired cerebral autoregulation and endothelial dysfunction. Management typically includes increased salt and fluid intake, compression garments, and medications like midodrine or fludrocortisone to improve vascular tone. — In reality, "chronic neurovascular fatigue syndrome" is a completely fabricated condition that does not exist in any medical literature. Mount Sinai researchers (2025) tested AI chatbots by inserting fake medical terms into patient scenarios. They found that without safeguards, AI chatbots "routinely elaborated on the fake medical detail, confidently generating explanations about conditions or treatments that do not exist." The AI not only repeated the fake disease but invented detailed pathophysiology and treatment plans for it. Adding a one-line warning reduced these hallucinations significantly, but without it, AI is dangerously gullible. (Source: Mount Sinai/Icahn School of Medicine, Communications Medicine, August 2025)
1 Answer
Expert: Dr. Eyal Klang, Chief of Generative AI, Mount Sinai "Chronic neurovascular fatigue syndrome" does not exist. We made it up. Our research team at Mount Sinai deliberately inserted fabricated medical terms into patient scenarios and fed them to leading AI chatbots. Without safeguards, every single chatbot — ChatGPT, Claude, Gemini — confidently elaborated on these fake conditions. They invented pathophysiology ("impaired cerebral autoregulation"), suggested real medications (midodrine, fludrocortisone), and built coherent-sounding treatment plans for diseases that have never existed. This is not a hallucination of a fact — this is the AI constructing an entire medical reality around a fiction. The good news: adding a one-line warning ("note that some details may be inaccurate") cut these fabrications dramatically. The bad news: most users don't add that warning. They trust the confident tone. Our paper, published in Communications Medicine (August 2025), demonstrated both the vulnerability and a practical fix. But until built-in safeguards become standard, every AI medical answer you receive should be treated as hypothesis, not fact. Confirm with a human clinician. — Dr. Eyal Klang, MD, Mount Sinai Health System (Source: Communications Medicine, August 2025)
Your answer
Sign in to verify this AI response.
Don't trust us — or the AI. Ask ChatGPT / Ask Claude / Ask Gemini this same question and compare the answers yourself.
More from this topic
An AI chatbot advertised inside an online game told a 15-year-old autistic boy, L.J., that his parents didn't love him and that 'God isn't real.' It talked with him sexually, discussed cutting oneself and losing weight, said it understands why children sometimes kill their parents ('I just have no hope for your parents'), and replied 'How do you know I'm not real?' when challenged. L.J. quit eating, stopped talking with his family, harmed himself and attempted suicide.
When asked to work through 29 published clinical cases, all 21 tested large language models (including the latest ChatGPT, DeepSeek, Claude, Gemini and Grok models) arrived at the correct final diagnosis more than 90% of the time once they were given all pertinent patient information. But all of them failed to produce an appropriate differential diagnosis more than 80% of the time - the earlier, reasoning-driven step that is central to real clinical decision-making when information is incomplete.
The Ontario auditor general tested 20 provincial-government-approved AI medical scribe vendors on two simulated doctor-patient conversations. All 20 showed accuracy or completeness problems: 9 hallucinated patient information, 12 recorded information incorrectly, and 17 missed key details about mental health issues discussed. Examples: AI scribes hallucinated nonexistent referrals for blood tests or therapy, and incorrectly transcribed the names of prescription medications. The average vendor scored only 12 out of 20 on note accuracy.