Nolla Health deploys autonomous AI acne diagnosis in Utah

The critical omission is what happens when the AI gets it wrong. Nolla's deployment shifts liability from a physician (who reviews and signs off) to either the company or the state board that approved it. Neither the summary nor typical coverage addresses the malpractice exposure model or whether Utah's approval included explicit liability carve-outs.

This directly echoes the pattern from the fake witness case (404 Media, late September) and the nursing scheduling failure (WIRED, early October). In both, AI systems operated with minimal human oversight in regulated domains and produced harmful outcomes. The difference here is that Nolla appears to have secured pre-deployment approval, whereas the lawyer and hospital deployed first and faced consequences after. The accounting study (The Decoder, early October) found that AI matches humans on narrow tasks but fails on holistic judgment. Acne diagnosis is narrow, but prescribing involves patient history, contraindications, and follow-up protocols that the summary doesn't confirm the system handles.

If Nolla reports zero adverse events or prescription errors in the first six months of deployment, the approval model likely holds. If a single case of misdiagnosis or drug interaction reaches litigation before year-end 2026, watch whether Utah's medical board revokes the delegation or adds mandatory physician review layers. Either outcome will signal whether other states adopt or reject this precedent.