“A study found that approximately 50% of the responses from AI platforms like ChatGPT and Gemini to health-related questions were problematic.”