78 percent cannot.
Black Book Research.
182 hospitals.
One question.
Can you deliver a complete auditable
clinical AI explanation
to a regulator within 30 days.
78 percent said no.
Most health system leaders know
they have a clinical AI governance problem.
Most do not know where to start.
Not the policy.
Not the committee.
Not the vendor contract.
The deployment.
That is where the accountability gap lives.
That is where the audit starts.
Here is the five step process
I use with every institution.
Step 1. Map the deployment.
Not the technology.
The deployment.
Which clinical AI tools are active right now.
Which workflows they touch.
Which patients they reach.
You cannot name an owner
for something you have not mapped.
Step 2. Name the seats.
Two seats.
Two names.
No exceptions.
↳ The Governance Owner.
Charter. Commission. Cover.
↳ The Decision Owner.
Decide. Document. Defend.
If either seat is empty
the accountability gap is open.
Step 3. Interview the Decision Owner.
Not a survey.
A direct conversation
with the clinician holding the call.
What does this deployment recommend.
What does the chart record when you act on it.
What does the chart record when you do not.
The Decision Owner already knows
where the accountability breaks.
Most institutions never ask them.
Step 4. Build the Handoff map.
The Handoff is the documented junction
between the Governance Owner
and the Decision Owner.
Most charters name both seats.
Most leave the junction undocumented.
The Handoff map is the audit trail.
It is what your institution produces
when a regulator asks
who knew what and when.
Step 5. Get the sign off.
Both seats sign.
In writing.
Before the next deployment goes live.
Governance Owner signs
what the agent is authorized to do.
Decision Owner signs
what they are accountable to hold.
The sign off is the moment
the accountability gap closes.
The Accountability Gap™ (TAG™)
does not close with a framework.
It closes when two seats are named,
the Handoff is documented,
and both owners sign
before the deployment goes live.
The carousel below maps the full process.
You already know which deployment
in your institution
has no named owner right now.
That deployment is not a gap.
It is a liability.
Which of the five steps
has your institution not completed yet
for your active clinical AI deployments?
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