Global health organizations made big moves in AI this week. The World Health Organization updated its digital health plan through 2027, pushing countries to use more AI tools for medical care. This strategy helps doctors worldwide share AI solutions for diseases like cancer and diabetes.

Microsoft launched AI Orchestrator for hospitals - a smart system that automatically collects patient scans, lab results, and doctor notes to prepare cancer treatment plans. Nurses report this cuts hours off their paperwork each day. In Asia, Taiwanese hospitals began using AI that spots pancreatic cancer in medical images, with the technology cleared by both Taiwan and US regulators.

The US FDA accelerated its AI plans after a successful pilot where AI helped scientists review new medicines faster. Commissioner Martin Makary said "AI does the boring paperwork so our experts can focus on patient safety." By June 2025, all FDA departments will use AI to check drug labels, analyze clinical trials, and track side effects.

AI risks came into focus as Anthropic warned its powerful Claude Opus 4 model could be misused for dangerous biological research. The company added new security checks. In the UK, hackers targeted NHS systems using AI to diagnose heart disease and diabetes, showing how attackers might exploit smart medical tools.

New AI research assistants showed promise and limitations. Google's updated search now has an AI Mode that lets users ask medical questions with photos or voice recordings. However, the MedBrowseComp test found most AI systems still make mistakes when digging through complex medical studies.

On the treatment front, Compass AI helps oncologists predict which patients will respond to immunotherapy by analyzing tumor genes. This "concept bottleneck" model explains its reasoning step-by-step so doctors can trust its suggestions. Meanwhile, hospitals in Singapore and Chicago are testing secure platforms to let doctors safely experiment with AI tools.