Rebecca agreed to AI transcription - and was labeled a drug user: "I was in shock"
Rebecca Green agreed to have her medical appointment recorded by AI to save time, only to be stunned when a letter to her family doctor falsely claimed she used hallucinogenic mushrooms.
When Rebecca Green arrived for her first appointment with a urologist, she was asked if she agreed to have the meeting transcribed by artificial intelligence (AI). She agreed immediately, without hesitation. This decision caused her significant distress when the AI invented a completely false claim that she used illegal drugs.
Rebecca said the decision to allow the transcription was made voluntarily to save the doctor's time. "I didn't want to be that annoying patient who takes up too much of her time," she recalled in an interview with ABC Australia.
"I was stunned and in tears," Green said. "The software invented that I take hallucinogenic mushrooms. I have never done mushrooms in my life." Green discovered the error in March, after undergoing surgery for kidney stones, when she reviewed the surgical letter the urologist sent to her family doctor. The letter stated that Green consumed micro-doses of hallucinogenic mushrooms, suggesting this could be the cause of early bleeding around her kidneys.
For Green, the fear of the consequences was immediate, as she was receiving compensation under the Workers' Compensation Act at the time. She feared that any mention of illegal drugs in her medical file could harm her case. "This letter, which says something not based on any evidence but could have huge consequences for me, was quite scary," she said. "It felt very impersonal. I didn't get the level of care and closeness from the doctor that I had hoped for in such a sensitive situation."
After Green filed a complaint, the urologist sent an apology letter: "We take the accuracy of documentation very seriously and are very sorry that this happened." The doctor noted she could not determine how the claim about the mushrooms was entered, but it appeared to be an error during the dictation or transcription process. The urologist corrected the medical file and promised to re-examine the use of AI in her clinic, but refrained from answering further questions.
"When you check this letter, you should be able to correct it before it goes out," Green said angrily. "Why are you not synchronized with each of your patients?". Green admitted she feels lucky to have caught the error in time, but warns other patients: "I love AI, but the essential component is the human being. Ultimately, what happened here is mainly a lack of human involvement."
An expanding phenomenon: when AI "libels" in the clinic
The story of Green is not an isolated case. A new report by the privacy organization Digital Rights Watch (DRW) warns of the dangers of using medical AI writers due to their tendency to "hallucinate" symptoms and diseases. Examples revealed in the report include an AI confusing the right breast with the left during a cancer diagnosis and a case where a patient was mistakenly diagnosed with epilepsy.
Although regulators clarify that it is mandatory for doctors to check every AI output before sending it, experts point out that many doctors develop "over-trust" in the technology after a few weeks of use and stop critically evaluating the summaries.
Many doctors note that AI eases the bureaucratic burden and allows them to focus on the patient. However, experts recommend that patients not hesitate to ask pointed questions about the use of AI, ensure the doctor has reviewed the record, and remember they have the full right to refuse recording.
