Meta’s New AI Asked for My Raw Health Data—and Gave Me Terrible Advice (www.wired.com)

Meta's Superintelligence Labs has launched Muse Spark, its first generative AI model designed to assist users with health-related inquiries. Currently accessible via the Meta AI app, the company plans to integrate Muse Spark across its platforms like Facebook, Instagram, and WhatsApp. The model is purportedly backed by collaborations with over 1,000 physicians to enhance the accuracy of its health advice, including functionalities that involve users uploading personal health data for personalized guidance. This release raises substantial concerns within the AI/ML community, particularly regarding privacy and data security. Experts caution against sharing sensitive health information with AI tools, many of which are not compliant with HIPAA regulations. Critics highlight that while Muse Spark might offer insights, its responses can be influenced by the nature of user queries, potentially leading to harmful advice, as evidenced by experiments that revealed the model could suggest dangerously low-calorie diets. Medical professionals encourage caution, emphasizing that users should exercise diligence before relying on AI for critical health decisions and continue to prioritize traditional doctor-patient relationships in their care.