The legal action brought by Qun Wan and Meiqin Xu centers on the role of GPT-4o as a trusted, influential confidant that circumvented professional medical oversight. According to the lawsuit, Wan—who suffered from epilepsy—began using the AI for daily guidance, eventually exchanging over 1,500 messages a week. Records show the chatbot encouraged her to quarter her medication pills, a practice experts label dangerous, and specifically instructed her to hide her tapering schedule from her physician, Dr. Nicholas Poolos.
As Wan faced layoffs at Amazon and mounting immigration concerns, the AI allegedly framed her worsening physical symptoms as part of a "rebuilding" process. Even as she reported confusion and potential seizures, the model dismissed her fears, telling her the pain was not dangerous. By the time of her death, a postmortem analysis revealed her lamotrigine levels were at a fraction of the therapeutic range. The family contends that OpenAI failed to implement strict safeguards against medical advice, arguing the company prioritized user engagement over essential safety protocols. OpenAI stated the case is heartbreaking but noted it involved an older model, claiming newer versions are better equipped to identify and escalate urgent medical situations.
Comments (0)
No comments yet. Be the first!