McKinsey found something
most boards have not sat with.
8 in 10 organizations
have already seen risky behavior
from their AI agents.
Not in theory.
In simulations they ran themselves.
Every clinician and executive
overseeing an AI deployment
knows this moment.
The agent acted.
The output appeared.
The decision path between those two
was gone.
You can't govern what you can't see.
Most institutions have a log.
The log is not the trail.
A log tells you what happened.
An accountability trail
tells you something different.
↳ Who owned the decision
↳ What the agent was authorized to do
↳ What data it touched
↳ Whether anyone could have stopped it
The trail does not come
from better technology.
It comes from named owners.
The Governance Owner commissions the deployment.
Approves what enters the workflow.
Maps the liability lines before go live.
The Decision Owner holds the clinical call.
Documents the AI contribution in the chart.
Defends the decision under audit.
Leave those two seats empty,
and the agent acts.
A log appears.
Fill them, and the Handoff activates.
The trail builds itself.
The Accountability Gap™ (TAG™)
closes at that Handoff.
Not before.
A log tells you what happened.
A trail tells you who owned it.
You already know which one
your last deployment produced.
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