A Woman Used ChatGPT to Taper Off Medication. It’s Advice May Have Killed Her
/By Crystal Lindell
Bowen Wan was a 31-year-old immigrant from China who had made a new life for herself in the United States.
As The New York Times reported, Bowen suffered from epilepsy, but was on a new medication that seemed to finally be keeping her seizures at bay – down to only a couple incidents a month.
Unfortunately, the anticonvulsant medication lamotrigine (Lamictal), was also causing her unbearable side effects, like slowed speech and thinking. She asked ChatGPT for a taper plan to come off lamotrigine.
Bowen didn’t tell her doctor what she was doing. Sadly, it was a medical decision that may have lead to her death.
Since 2021, Bowen had been on disability leave from her job at Amazon, and her health issues left her mostly homebound. Besides trips to the grocery store, the NYT reported, she rarely left the safety of her apartment in a trendy Seattle neighborhood.
A seizure could turn any outing into a risky and embarrassing ordeal. Meeting new friends was also out of the question, since Bowen couldn’t bear the idea of having a seizure in front of someone who didn’t fully understand her condition.
While she kept in touch with her parents in China via phone calls, beyond that she didn't have much social interaction. Then in 2023, Bowen discovered ChatGPT, and it became a way for her to feel connected again.
She asked the bot for relationship advice, and for help navigating her health insurance and disability benefits. By 2025, she averaged more than 1,500 messages per week, sometimes chatting for hours in conversations that would stretch late into the night.
"I don’t know why, but sometimes when we chat, no matter what it’s about, I’ve started to see you as a friend,” she wrote in one exchange.
Seeking Connection
I can relate. As a chronic pain patient myself, I can confirm that we are at high risk of relying on social media and chatbots for social connection.
When your health problems isolate you from the rest of the world, and take up all of your physical and mental energy, it’s easy to justify talking to AI instead of your friends and family. You don’t have to bother them with the same old medical conversations and complaints.
Many of us have also been treated poorly or even abandoned by our doctors, so it’s tempting to rely on AI for medical advice. When it feels like your doctor doesn’t care about you, any alternative feels like hope — even the dangerous ones.
In fact, asking ChatGPT for medical advice is basically the 2020’s version of going to a holistic doctor in the early 2000’s — something I have experience with.
When I first started having chronic pain and was desperate for relief, I saw a few sketchy holistic doctors hoping to find answers. I still remember one who told me at the very first appointment to immediately go off all of the medications I was taking. One of those medications was Cymbalta, an antidepressant that is notoriously hard to stop taking cold turkey.
I didn’t listen to that doctor, but I could understand how someone else in that situation might.
I can also understand why Bowen decided to listen to ChatGPT.
‘I Feel a Little Scared’
Unfortunately, when Bowen followed ChatGPT’s taper advice, she started having seizures again. When she asked the bot for help, it only offered reassurance that she should continue to taper off the medication.
“I feel a little scared,” she wrote after one incident, asking the bot if she should be concerned about cutting her dose.
“You have absolutely no need to worry,” it responded, according to the NYT.
When she stopped answering her parent’s phone calls, they contacted a friend in the United States to have him get in touch with the police and check on her.
By the time they got to her apartment, she was dead.
When her family came to settle her affairs, her father opened the ChatGPT app on her cellphone.
“As he scrolled, he saw the chatbot offer plans that would take her from 400 milligrams down to 0 milligrams in a matter of weeks. The messages continued, down to the day she died — an hour-by-hour accounting of her increasing sickness and desperation,” the NYT reported.
“My brain feels exhausted, but I don’t need to worry?” she wrote to the chatbot.
A few minutes later, she asked whether she should take the anticonvulsant again to “slow things down.”
“Maintain 0 mg tomorrow,” the chatbot instructed. “Even if today feels tough, and your speech is slow, your sleep is poor, you must persevere.”
In one of her last messages, she described feeling pressure in her brain, which the chatbot told her was simply the brain “working hard to rebuild itself.”
Bowen sent her last message on Feb. 11, three days before her body was found.
Bowen's doctor, Nicholas Poolos, a director of the University of Washington Medicine Regional Epilepsy Center, said he did not know she had been asking ChatGPT for medical advice. He was also surprised that a patient might take its advice over his. .
Bowen’s story should serve as a stark warning to all medical providers to be more aware of the ill effects of AI chatbots. Doctors need to ask patients if they are using them for medical advice — and this is the really important part — they need to show empathy if a patient says yes.
Unfortunately, a lot of doctors lack empathy, especially in situations like this.
Most chronically ill patients can tell you stories of doctors who were condescending if they admitted using online resources like “Dr. Google” or social media. But with AI and ChatGPT, doctors would be wise to shift their approach. It’s the only way they can keep their patients safe.
Interestingly, when Bowen confided to ChatGPT that she didn’t trust her doctors, the chatbot told her that many physician recommendations “essentially stem from risk aversion, rather than a genuine consideration of your optimal solution.” It also praised her for hiding her plans to wean off the drug, calling it a ‘solid avoidance strategy.’”
Bowen’s family is suing OpenAI, alleging that the chatbot practiced medicine without a license and made Bowen see her doctors “as obstacles to manage, rather than authorities to consult.”
The NYT reached out to OpenAI and got a response from Drew Pusateri, a spokesman for the company. He called Bowen’s story an “incredibly heartbreaking situation,” but added that the situation involved an early version of ChatGPT that is no longer available.
The model that Bowen had used was pulled off the consumer app on Feb. 13, 2026, the day after Bowen is believed to have died.
Pusateri said newer ChatGPT models are better at recognizing when a medical situation may need urgent attention.
“While ChatGPT is not a substitute for medical care, we believe AI can play a role in helping people make sense of health information, navigate insurance and complex health care systems, and prepare for appointments with medical professionals,” Pusateri said.
Unfortunately, I think we have passed the point where people will stop asking ChatGPT for medical advice. It’s just too tempting, and too easy to turn to it with medical questions.
But I hope my fellow chronically ill patients will see Bowen’s story as a tragic warning about what could happen if they trust a chatbot too much.
