#Chatbots

Man took diet advice from ChatGPT, ended up hospitalized with hallucinations – USA Today

Welcome to the forefront of conversational AI as we explore the fascinating world of AI chatbots in our dedicated blog series. Discover the latest advancements, applications, and strategies that propel the evolution of chatbot technology. From enhancing customer interactions to streamlining business processes, these articles delve into the innovative ways artificial intelligence is shaping the landscape of automated conversational agents. Whether you’re a business owner, developer, or simply intrigued by the future of interactive technology, join us on this journey to unravel the transformative power and endless possibilities of AI chatbots.
A man was hospitalized for weeks and suffered from hallucinations after poisoning himself based on dietary advice from ChatGPT.
A case study published on Aug. 5 in the Annals of Internal Medicine, an academic journal, states the 60-year-old man decided he wanted to eliminate salt from his diet completely. To do so, he asked ChatGPT for an alternative to salt, or sodium chloride, to which the AI chatbot suggested sodium bromide, a compound historically used in pharmaceuticals and manufacturing.
While the journal noted that doctors were unable to review the original AI chat logs and that the bot likely suggested the substitution for another purpose, such as cleaning, the man purchased sodium bromide and used it in place of table salt for three months.
As a result, he ended up in the ER with paranoid delusions, despite having no history of mental health problems. Convinced that his neighbor was poisoning him, the man was reluctant to even accept water from the hospital, despite reporting extreme thirst. He continued to experience increased paranoia, as well as auditory and visual hallucinations, eventually landing him an involuntary psychiatric hold after he tried to escape during treatment.
Doctors determined that the man was suffering from bromide toxicity (or bromism), which can result in neurological and psychiatric symptoms, as well as others experienced by the man, including acne and cherry angiomas (bumps on the skin), fatigue, insomnia, subtle ataxia (clumsiness) and polydipsia (excessive thirst).
Other symptoms of bromism can include nausea and vomiting, diarrhea, tremors or seizures, drowsiness, headache, weakness, weight loss, kidney damage, respiratory failure and coma, according to iCliniq.
Bromism was once far more common due to bromide salts having been used in everyday products. In the early 20th Century, it was used in over-the-counter medications, often resulting in neuropsychiatric and dermatological symptoms, according to the study’s authors. Incidents of such poisoning saw a sharp decline when the Food and Drug Administration phased out the use of bromides in pharmaceuticals between the mid 70s and late 1980s.
The man was treated at the hospital for three weeks, over which time his symptoms progressively improved.
USA TODAY reached out to OpenAI, the maker of ChatGPT, for comment on Wednesday, Aug. 13, but has not received a response.
The company provided Fox News Digital with a statement, saying, “Our terms say that ChatGPT is not intended for use in the treatment of any health condition, and is not a substitute for professional advice. We have safety teams working on reducing risks and have trained our AI systems to encourage people to seek professional guidance.”
Doctors involved in the case study said they suspected that the patient had used ChatGPT version 3.5 or 4.0, the former of which they tested in an attempt to replicate the answers the man received. While the study’s authors noted that they couldn’t know exactly what the man was told without the original chat log, they did receive a suggestion for bromide as a replacement for chloride in their tests.
“Though the reply stated that context matters, it did not provide a specific health warning, nor did it inquire about why we wanted to know, as we presume a medical professional would do,” said study authors Dr. Audrey Eichenberger, Dr. Stephen Thielke and Dr. Adam Van Buskirk.
AI carries the risk of providing information like this without context, according to the doctors. For example, it is unlikely that a medical expert would have mentioned sodium bromide at all if a patient asked for a salt substitute.
“Thus, it is important to consider that ChatGPT and other AI systems can generate scientific inaccuracies, lack the ability to critically discuss results, and ultimately fuel the spread of misinformation,” according to the study.

source

Man took diet advice from ChatGPT, ended up hospitalized with hallucinations – USA Today

OpenAI Designed GPT-5 to Be Safer. It

Man took diet advice from ChatGPT, ended up hospitalized with hallucinations – USA Today

The Complete Guide to Using AI as