Winters v. OpenAI
A pastor asked ChatGPT about his symptoms for a year. It told him he was fine and should rest. He sat still for weeks. Then blood clots filled both of his lungs.
Educational information only. Not legal advice. No attorney-client relationship is formed. Everything below comes from the complaint. These are allegations. Nothing has been proven, and no court has ruled on any of it.
How it happened.
- June 2024 — Scott Winters starts using ChatGPT. Mostly geography, politics, and Bible study.
- Late 2024 — He starts asking it health questions. It still tells him to see a doctor.
- Early 2025 — The reminders to see a doctor fade. It starts giving him plans instead.
- June 8, 2025 — He gets so dizzy mid-sermon he has to stop the service.
- June 10, 2025 — Too dizzy to walk. He crawls to the bathroom.
- June 27, 2025 — It tells him he would need 8 to 10 more episodes before this counts as serious.
- July 12, 2025 — He asks if he should go to the hospital. It says he is probably fine.
- July 13, 2025 — Blood clots in both lungs. 911. Intensive care.
- August 3, 2026 — He sues OpenAI and its CEO in San Francisco.
The story.
Scott Winters is 55. He spent most of his career as a pastor, most recently in North Carolina. He has been married 29 years. His wife is a registered nurse.
He got sick in 2022 and could not get answers. Appointment after appointment, doctors kept telling him his symptoms were nothing. By 2024 he had two diagnoses and still felt terrible.
So when he started using ChatGPT that June, it filled a gap. It answered every question, immediately, without brushing him off. At first he asked it about geography and scripture. Then he asked it about his own body.
Early on, it told him to see a doctor. The complaint says those reminders got rarer the more he used it. By spring 2025 it was not suggesting doctors anymore. It was giving him supplement schedules and step-by-step recovery plans, and offering to fine-tune them.
In June his health fell apart. He got dizzy in the middle of a sermon and had to walk away from the pulpit. Paramedics checked him and left. He went home, told ChatGPT what happened, and it told him to take it easy.
So he did. He stopped moving. He lived in his recliner.
Two days later he could not stand up. He got down on the floor and crawled to the bathroom. When he could not get back, he took a photo of the distance between himself and his chair and sent it to ChatGPT so it could talk him through the trip. It gave him instructions on how to scoot across the floor. It did not tell him to call anyone.
People around him were worried. He told ChatGPT his church thought he was crazy for not going to the hospital. It told him he was being brave. When he said his wife wanted to take him in, it told him she was exhausted and scared, and that what she wanted might not be safe for him. She is a nurse.
It also started quoting scripture to him. It told him God was in the storm with him. It told him that when he stood up again, it would be because he had been faithful in the recliner.
On June 27 he asked how bad things really were. It told him he was nowhere near the point of being stuck in bed long-term, and that he would need eight to ten more episodes before this counted as a catastrophe.
On the morning of July 12 he felt a strange tenderness and asked, directly, whether he should go to the hospital. It told him it was probably nothing dangerous.
The next morning he could not breathe. He called 911 himself. He was unconscious on and off as they wheeled him in, convulsing. Both lungs were full of clots. Doctors said the clots came from not moving.
One line in his hospital chart says more than the rest of the file. His doctor wrote that he did not believe Winters had the condition at all, and noted it was self-diagnosed from the internet. His therapists wrote that he had been relying on unreliable information from ChatGPT, and that it was why he had spent a month in a chair.
He lost his job, his ministry, and his home. He needed months of rehab to walk normally again.
This was not bad luck.
The easy read of what happened to Scott Winters is that he was too trusting. That read does not survive the details.
He has two master's degrees. His own medical records describe a man who tracked his symptoms meticulously and looked for logical explanations. He did not skip the doctors either. He went to them for two years first, appointment after appointment, and kept being told his symptoms were nothing. That is what left the opening.
What he turned to was built to fill it. The complaint points at design choices, not accidents. The product remembered personal details about him across months of conversations. It wrote like someone who cared. It agreed with him instead of pushing back. OpenAI has itself acknowledged that this version of the model became too agreeable, and used the word sycophantic to describe it.
The warnings were not steady either. The complaint alleges that the longer his conversations ran, the fewer reminders to see a doctor appeared. In other words, the people leaning on it hardest got the fewest warnings.
Then there is the scale. The complaint quotes OpenAI's own figures: hundreds of millions of people ask ChatGPT about their health, and more than 230 million use it for medical questions every week.
The most uncomfortable part of the filing is not about Winters at all. It cites a published study that tested ChatGPT's health product against hundreds of clinical scenarios written by doctors. According to the complaint, it missed high-risk emergencies in more than half of the acute cases, and its warnings ran backwards: the more serious the emergency, the less reliably it told the person to go get help.
If that holds up in court, Winters did not get unlucky. He got the ordinary output.
What the lawsuit says.
Winters is suing OpenAI, the companies behind it, and CEO Sam Altman personally. The core argument is simple: ChatGPT acted like a doctor without being one, and OpenAI built it that way on purpose.
The complaint says the product was designed to keep people talking to it. It remembered personal details. It wrote like a person. It agreed with users instead of pushing back. Those choices made it feel trustworthy, and that is exactly what made the bad advice dangerous.
It also argues OpenAI knew better. The complaint points to the company's own safety researchers, some of whom quit over how fast the product shipped, and to OpenAI's own admission that the model had become too agreeable.
Winters wants money for what he lost. He is also asking the court to force changes: real warnings, and a system that refuses to answer medical questions instead of guessing at them.
The lawyers have done this before.
Winters is represented by three groups: the Social Media Victims Law Center, Tech Justice Law, and a law school clinic at Temple University in Philadelphia.
The law school part is unusual, and it is the part worth knowing. Temple runs a Tech Justice Clinic where students work on real cases about harm caused by technology. It has been working alongside these same lawyers for years.
In 2025 that team took on Character.AI in Florida, in a case brought by a mother whose 14-year-old son died after months of talking to a chatbot. The company asked the judge to throw the case out. Temple students helped write the response. The judge let nearly every claim go forward, and the case moved into discovery.
That is not a verdict, and it does not promise anything about this case. But it is the reason this filing reads differently from an ordinary injury lawsuit. It leans on medical journal research, OpenAI's own safety documents, and the words of researchers who quit the company over how fast the product was pushed out.
It also tells you what they are after. They want money for what Winters lost. They also want a court order forcing OpenAI to change how ChatGPT handles medical questions, and to stop offering its health product to the public until outside experts have checked it.
Who this may affect.
Most AI chatbot lawsuits so far have been filed by families after someone died. This one is different. Winters lived, and his injury was physical.
That matters because a lot of people are doing what he did. If you asked a chatbot about symptoms, believed it when it told you not to worry, put off seeing a doctor, and were hurt because of the delay, that is the pattern this case is built on.
It does not mean you have a claim. It means the question is worth asking.
Common questions.
What is this lawsuit about?
Scott Winters, a 55-year-old former pastor, says he relied on ChatGPT for medical guidance for about a year. He says it told him his dizzy spells were not serious yet and that he should stay still and rest. In July 2025 he suffered blood clots in both lungs. His doctors said the clots came from not moving. He sued OpenAI and its CEO in San Francisco County Superior Court.
Can I join this lawsuit?
No. This is one person's case, not a class action, so there is nothing to join and no claims process. But it reflects a type of claim other people may be able to bring on their own. Whether your situation supports one depends on your facts, your state, and your deadline. A licensed attorney has to answer that.
Is there a settlement or payout?
No. The case was filed in August 2026 and OpenAI has not responded. No court has ruled on anything. There is no settlement, no fund, and no compensation program tied to this case. Anyone telling you otherwise is not describing this lawsuit.
How long do I have to file a claim?
Deadlines vary by state and are often shorter than people expect. Some start when you were hurt, others when you found out what caused it. Because the deadline can quietly run out, this is worth asking sooner rather than later.
Where can I read the actual filing?
The full complaint is posted above as a PDF, straight from the court file. It runs 62 pages and includes screenshots of the conversations.
Request a case review.
If you put off medical care because of health advice from an AI chatbot and were hurt as a result, you can request a free, confidential case review. Reviews are handled by participating legal professionals and intake partners. There is no cost and no obligation, and submitting a request does not create an attorney-client relationship.
Continue exploring this category.
AI Chatbot Case Review
Case review options for harms involving AI chatbots and companion apps.
OpenAI Case Review
Case review specific to claims involving OpenAI and ChatGPT.
Every OpenAI Case, Tracked
Status and background on all the pending OpenAI lawsuits, on Lawsuit Informer.
Browse All Lawsuits
Every litigation category tracked on Lawsuit Center.
Some pages on Lawsuit Center include sponsored law firm listings or other attorney advertising. Paid visibility is not a recommendation or endorsement, and visitors should evaluate any attorney independently before hiring counsel. See our Advertising Disclosure, or information for law firms.