The FDA recently relaxed requirements for medical decision-making systems with artificial intelligence, which will allow more AI tools to reach clinics without FDA approval.[1] At the same time, Utah launched a pilot project with Doctronic for autonomous medication refilling using AI, and OpenAI introduced ChatGPT Health, which adapts to medical records and data from patients' wearable devices.[1] These three developments have dramatically changed the landscape of healthcare AI, bringing the potential for better healthcare advice and easier access to medicines at no or minimal cost.[1] The safe deployment of AI in medicine requires three essential elements: clear rules and transparency, training and oversight.[2] It is important to distinguish between validated clinical software for healthcare professionals and general chatbots, which should only be taken as a suggestion, not as a final verdict.[2] AI in medicine remains a physician's assistant—it does not prescribe drugs, make decisions, and cannot perform a physical examination or measure vital signs.[2]