Large Language Models hallucinate when removing patient info from EPR, finds study

Artificial intelligence (AI) tools sometimes produce hallucinations when asked to remove personal patient information from electronic patient records (EPRs), a study has found.

Researchers from the University of Oxford evaluated the ability of large language models (LLMs) and purpose-built software tools to detect and remove patient names, dates, medical record numbers, and other identifiers from real-world records, without altering clinical content.

The study, published by iScience on 9 December 2025, found that smaller LLMs frequently over-redacted or produced hallucinatory content, in which erroneous text not present in the original record was shown, or occasionally introducing fabricated medical details.

Read full article.

Source: Digital Health, 18 December 2025.

Related reading

• 2025: A turning point for digital patient safety
• Artificial intelligence and patient safety in healthcare: Insights and recommendations from HETT 2025 roundtable