88 percent deploying AI.
86 percent not ready for it.
McKinsey.
10,000 senior leaders.
15 countries.
State of Organizations 2026.
One in six said nobody owns it.
In enterprise AI
that is an operations problem.
In clinical AI
that is a patient problem.
Most health systems govern one layer.
Almost none govern all five.
The Rules Layer.
↳ Classical AI.
↳ The rule is outdated.
↳ Nobody updated it.
The Pattern Layer.
↳ Machine learning.
↳ The pattern drifted.
↳ Nobody validated it.
The Imaging Layer.
↳ Deep learning.
↳ It misclassified.
↳ Nobody owned the call.
The Documentation Layer.
↳ Generative AI.
↳ The record is wrong.
↳ It is already in the chart.
The Action Layer.
↳ Agentic AI.
↳ It acted.
↳ Nobody designed accountability before it moved.
Five layers.
Five failure modes.
Most institutions are governing the wrong mix.
The Governance Owner names
what each layer is authorized to do.
The Decision Owner holds
what each layer produces at the bedside.
The Handoff between them
is where the gap either closes
or compounds.
The Accountability Gap™ (TAG™)
does not close at the layer level.
It closes when both seats are named
at every layer
before the failure mode arrives.
You already know which layer
in your institution
has no named owner right now.
That layer is not a gap.
It is a liability.
Which layer in your clinical AI portfolio
would fail the named owner test today?
Mo Johnson, MD MBA is a cardiothoracic surgeon and the founder of GPe Research. Field Notes are short dispatches from the clinical AI accountability frontier, published alongside the MedicoVigilance™ newsletter at medicovigilance.org.
Follow the work on LinkedIn: linkedin.com/in/mo-johnson