United States: A US medical journal cautioned against using ChatGPT regarding health information after one man developed a rare condition after reportedly having an interaction with the chatbot about cutting table salt out of his diet.
A study published in the Annals of Internal Medicine discusses a case of bromism or bromide poisoning that a 60-year-old man developed after accessing ChatGPT.
More about the news
According to the article, bromism was considered a well-known syndrome during the early 20th century, and it was believed that it was one of the causes of nearly one out of every 10 psychiatric hospitalisations during the same period.
According to doctors, the patient informed them that after reading about the adverse consequences of sodium chloride, or table salt, he used ChatGPT to research removing chloride in his diet and began using the latter, sodium bromide, within three months.
A 60-year-old man accidentally poisoned himself after replacing salt (sodium chloride) with sodium bromide after taking ChatGPT's dietary advice.https://t.co/fxGVJjkq7j
— WLNS Newsroom (@WLNS) August 12, 2025
Moreover, this was despite reading that “chloride can be swapped with bromide, though likely for other purposes, such as cleaning,” as the Guardian reported.
Sodium bromide is a sedative that was used at the beginning of the 20th century.
According to the article’s authors, from the University of Washington in Seattle, the case highlighted “how the use of artificial intelligence can potentially contribute to the development of preventable adverse health outcomes”.
They also specified that due to their inability to view the ChatGPT conversation log of the patient, they were unable to identify what advice had been received by the man.
However, when the authors asked ChatGPT directly what chloride can be substituted with, it also suggested bromide, without providing a reasoned explanation of any particular health risk, and failing to prompt the writers on their reasons for inquiring, either committing to the role of a medical professional, the Guardian reported.
The company introduced an upgrade of the chatbot a week ago and said health was one of its largest capabilities.
It claimed that ChatGPT, which will be GPT-5-based, will be more effective in providing answers to health-related questions and also be more proactive in terms of flagging potential concerns, e.g., serious physical or mental illness.