AI pathology copilot trained on 1 million medical cases can now read whole slides and explain its reasoning

Researchers built a specialized AI model for pathology that learns from over 1 million labeled medical cases and can read entire microscope slides at once, then explain what it found. This means pathologists can now get AI assistance that understands context, handles multiple images together, and shows its work instead of just spitting out a diagnosis.

Until now, AI in pathology was mostly image-recognition software — it could spot patterns but couldn't reason through a diagnosis or explain itself to a human doctor. This system changes that by combining image understanding with language reasoning, which means a pathologist can actually trust and verify what the AI is telling them. The catch: this is a research demonstration, not a deployed clinical tool. The real question is whether hospitals will adopt it, and whether it actually performs better than human pathologists on cases they haven't seen before.