In this episode, we cover three major AI advancements shaping healthcare, open-source tech, and AI safety. First, we explore UCLA's groundbreaking AI model, SLIViT, which achieves expert-level accuracy in analyzing 3D medical scans at 5,000 times the speed of human specialists—potentially transforming diagnostic practices. Next, we delve into Meta’s latest AI releases, including Spirit LM for cross-modal language generation and SAM 2.1 for advanced image segmentation, signaling a shift toward more accessible open-source AI tools. Finally, we examine Anthropic's new evaluations aimed at identifying AI sabotage risks, shedding light on the growing need for AI safety measures.