OpenAI’s health push widens as privacy assurances collide with fresh safety doubts

OpenAI’s health push widens as privacy assurances collide with fresh safety doubts
OpenAI has opened its ChatGPT Health feature to all U.S. users over 18, betting that more personalized AI health answers will be useful enough to outweigh the unease that comes with handing sensitive medical data to a chatbot. The timing is awkward: the broader launch arrives just as the company faces renewed scrutiny over whether people may trust the system more than they should.

OpenAI’s case is straightforward. The company says users can now connect information from Apple Health, hospital systems such as Epic and Oracle Health, and services like MyFitnessPal, so ChatGPT can answer questions with more personal context rather than generic advice. In its own launch materials, OpenAI says the goal is to help people “better understand and navigate” their health information, while giving them control over what gets connected and when it is used.

The company has also leaned heavily on scale and performance. OpenAI says “more than 300 million people” already come to ChatGPT each week with health questions, and executives have argued the models are improving at handling complex medical information. During a briefing, OpenAI vice president Ashley Alexander said the models “are now capable of reasoning at levels that are better than clinician level,” though health lead Karan Singhal reportedly tempered that claim, saying only that some individual studies point in that direction.

That is where the skepticism comes in. Even as OpenAI promotes stronger health reasoning, it continues to say the product is “not intended for use in the diagnosis or treatment of any health condition.” The caution is not abstract: according to reporting, a Florida pastor has sued the company after ChatGPT allegedly gave him dangerous medical advice that contributed to delaying care for a pulmonary embolism.

So the launch reflects two realities at once. OpenAI is presenting ChatGPT Health as a private, encrypted assistant for organizing scattered records and asking better questions, not replacing doctors. But the more seamlessly health data flows into the main chat window, the harder it may be to keep that distinction clear for users.

Continue reading https://foxvector.com/stories/019f91a5-3eaf-3b68-73bf-326d8e3e2c19