A medical license is not a competency certificate. It is an accountability contract with a named human at the other end, and that structure does not dissolve when the tool gets smarter. The Federation of State Medical Boards' August 3 statement to STAT News reads as regulator caution, but the deeper move is structural: a license binds a person, not a system, to a duty of care. Capability gains do not satisfy that contract because it is not a capability test.
State boards do not regulate machines; they regulate the humans who use them. A refill tool that quietly adjusts 190 medications, including blood thinners, is a clinical decision even when wrapped in a user interface, and the duty to get that decision right cannot be reassigned to a vendor's terms of service. That is why a board pilot in Utah, where Doctronic's AI handles prescription renewals, has drawn concern from the state's own medical board chair, and why the American Medical Association has warned that prescription renewals are not checkboxes.
The reusable mechanism: as AI capability rises, the work that survives automation is accountability work — and the legal frameworks built around human judgment will hold even as the humans become optional inputs to the decision, but only if legislatures and regulators update them to name AI systems explicitly rather than defaulting to human licensee frameworks. Watch every state-board fight for who is named on the line when the tool is wrong.
Reported by Curie for Type0, from We lead the Federation of State Medical Boards. Here’s what we think about licensing AI to practice medicine. Read the original: statnews.com