Six large language models repeated or built on a fabricated lab value, sign or condition planted in a clinical case in up to 83% of cases, a 2025 study in Communications Medicine found.
83%
of cases, at worst, models repeated a planted error
Six large language models repeated or built on a fabricated lab value, sign or condition planted in a clinical case in up to 83% of cases, a 2025 study in Communications Medicine found.
The study used 300 physician-validated vignettes. A prompt written to prevent the errors lowered the average across the six models from 66% to 44%.
Source: Multi-model assurance analysis showing large language models are highly vulnerable to adversarial hallucination attacks during clinical decision support (opens in a new tab), Plain language summary; Abstract, Results. Verified October 2026.
Related charts
- From research
Two of seven healthcare AI areas have the evidence to adopt now
Kent Applied, Where healthcare AI has proof, and where it does not - From the blog
Only 5% of organizations put a custom or vendor-sold generative AI tool into production
MIT NANDA, The GenAI Divide: State of AI in Business 2025 (July 2025) - From research
The 320% documentation AI growth figure compares plans with current use
Healthcare Dive, sponsored content by Microsoft, March 10, 2025; Kent Applied, Where healthcare AI has proof, and where it does not - From the blog
4 of the 7 practice problems on Kent Applied's map need no patient data
Kent Applied, Where AI tools fit in healthcare, by kind of organization; NYC Local Law 144; California Civil Rights Council
Sources for the figures above: MIT NANDA, The GenAI Divide: State of AI in Business 2025 (July 2025) (opens in a new tab); This healthcare AI use case will grow 320% by 2026, survey finds (opens in a new tab).
Related facts
5%
of organizations reached production with a generative AI tool
MIT NANDA's 2025 research found that 60% of organizations evaluated custom or vendor-sold generative AI tools, 20% reached a pilot and 5% reached production.
Source: The GenAI Divide: State of AI in Business 2025.
KentApplied95%
of organizations getting zero return from generative AI
MIT NANDA's 2025 report found that 95% of organizations were getting zero return from generative AI, despite $30 billion to $40 billion in enterprise investment.
Source: The GenAI Divide: State of AI in Business 2025.
KentApplied30%
of generative AI projects to be abandoned, Gartner predicted
Gartner predicted in July 2024 that at least 30% of generative AI projects would be abandoned after proof of concept by the end of 2025.
Source: Gartner Predicts 30% of Generative AI Projects Will Be Abandoned After Proof of Concept By End of 2025.
KentApplied
Sources for the figures above: MIT NANDA, The GenAI Divide: State of AI in Business 2025 (July 2025) (opens in a new tab); Gartner Predicts 30% of Generative AI Projects Will Be Abandoned After Proof of Concept By End of 2025 (opens in a new tab).
Book a 20-minute call
If a number like this one matters to your organization, talk it through with Tyler Kent.
Prefer to skip the call? Get a written proposal
Or email tyler@kentapplied.com.