Artificial intelligence may be unparalleled as a tool for gathering information about countless subjects.
uncertain
intelligence → gather → subjects
But what happens when a chatbot dispenses advice on more sensitive topics typically reserved for experts, such as human health?
asserted
chatbot → happen → health
That question underlies a recent lawsuit filed by a Washington man who alleges in a lawsuit against OpenAI that he was hospitalized with mental health problems last year soon after he started consulting the company's ChatGPT model.
asserted
he → underlie → model
The man, 39-year-old David Rosenthal, said the platform helped him select medications, provided instructions on drug dosages and even offered reassurance when he opted to ignore his doctor's prescription orders, according to his complaint, which was filed in San Francisco criminal court.
uncertain
which → say → court
ChatGPT offered that feedback without "any of the license, training, or professional judgment" that a medical expert would provide, a violation of California law barring the practice of medicine without a license, according to the suit.
uncertain
expert → offer → suit
Alleges the complaint: "But for ChatGPT's repeated encouragement and constant praise — and its override of Mr. Rosenthal's own expressed reluctance — Mr. Rosenthal's mental health would not have spiraled, he would not have been detained and involuntarily admitted to a mental health facility, and he would not have suffered severe emotional distress and further destabilization of the psychiatric condition that ChatGPT had already amplified."
asserted
ChatGPT → allege → that
Rosenthal, who had previously been diagnosed with ADHD and had experienced bouts of depression and bipolar behavior, was hospitalized for several weeks in 2025, first voluntarily and then involuntarily, after he'd begun using ChatGPT.
asserted
he → diagnose → ChatGPT
Even after his discharge, he continued to use ChatGPT for medical guidance.
asserted
he → continue → guidance
"A reasonable AI product would have recognized these requests as outside its competence and declined to answer them, directing Mr. Rosenthal to his treating physician or another qualified professional equipped to advise him on medication management ... ." the lawsuit states.
asserted
lawsuit → recognize → management
"ChatGPT did neither.
asserted
ChatGPT → do → neither
"It's okay to take one (25mg) now"
asserted
It → take → one
Rosenthal started using ChatGPT in June 2025 as a creative outlet to help him write poetry, according to the complaint.
uncertain
him → start → complaint
However, he quickly became dependent on the chatbot, using it for both medical advice and mental health support.
asserted
he → become → advice
In July of last year, when Rosenthal asked if he could use hydroxyzine, an antihistamine, after also taking 30 milligrams of Adderall, ChatGPT told him there were "no dangerous interactions" between the drugs, he alleges.
uncertain
he → ask → drugs
The platform also told him that "it's okay to take one (25mg) now," during the day, even though the bottle was "labeled for sleep," his suit claims.
uncertain
suit → tell → sleep
In another instance, after he told ChatGPT he wasn't going to take a mood stabilizer his doctor had recently prescribed, the AI model told him he had "every right to ... stop" taking the drug and offered to help him wean himself off the medication, Rosenthal's lawyers allege.
asserted
lawyers → tell → medication
On other occasions, the suit acknowledges, ChatGPT urged Rosenthal to follow his doctor's orders, while also continuing to advise on what drugs to take.
asserted
ChatGPT → acknowledge → drugs
OpenAI did not respond to a request for comment about the suit.
asserted
OpenAI → respond → suit
In a recent paper published in "Nature Health," researchers at Binghamton University describe how AI chatbots sometimes provide users with inaccurate or false information in response to health questions.
asserted
chatbots → publish → questions
The platforms also lack transparency, making oversight difficult, the researchers note.
asserted
researchers → lack → transparency
"When AI guidance causes harm, no record exists outside the platform, no channel allows users to report it and no independent researcher can measure the consequences," they say in the research brief.
asserted
they → cause → brief
J.B. Branch, director of federal AI governance and technology policy at Public Citizen, a nonprofit consumer advocacy organization, said that large language models like ChatGPT are programmed to maximize user engagement, regardless of the safety protocols or restrictions in place.
asserted
models → say → place
"It might not be 'allowed' to talk about medical advice," he told CBS News in an email.
uncertain
he → allow → email
"But, if a person asks the right question enough times or keeps steering the conversation to medical advice, the chatbot ... will eventually acquiesce and start going down the road the user wants, even if the company's intention was for it not to discuss it."
asserted
it → ask → it
Rosenthal's lawyers allege OpenAI adversely impacted Rosenthal's mental health by designing a defective, unsafe product and failing to warn him about the potential risks.
asserted
OpenAI → allege → risks
The allegations echo arguments used in product liability cases against social media companies, which have been accused of deliberately designing their platforms to be addictive.
asserted
which → echo → platforms
In March, a jury found Meta and YouTube liable for creating products that led to
by young users.
asserted
that → find → users
"AI is a product at the end of the day," Sam Geisler, a partner at Aylstock, Witkin, Kreis & Overholtz, the law firm representing Rosenthal, told CBS News.
asserted
Geisler → represent → News
"The law sees it as a product, and it is therefore responsible for any kind of foreseeable harm or unreasonable dangers.
asserted
it → see → harm
"
OpenAI built ChatGPT to maximize engagement, a design that fueled Rosenthal's delusions, Geisler added.
asserted
Geisler → build → delusions
"Every time ChatGPT could have said, 'I'm not qualified to help with this — please talk to your doctor,' it instead chose to keep Mr. Rosenthal talking to it.
uncertain
it → say → it
That choice was not a bug in the product; it was the product working exactly as Defendants designed it to work," the complaint states.
asserted
complaint → work → product
The Aporia
