In real cases, people don’t usually “find AI” the way they find a diagnosis. Instead, they notice it indirectly—through:
- Generated or auto-populated summaries in the chart
- References to decision-support systems in imaging or perioperative notes
- Documentation that doesn’t line up with what they were told or what symptoms followed
- Software-based outputs that appear in records without clear confirmation steps
After surgery-related harm, the question isn’t whether AI exists in healthcare—it’s how it was used and whether the clinical team followed safe processes around it.


