Three in four U.S. health systems are using or planning to deploy AI - but fewer than one in five have the governance structures to manage what these tools do after go-live. The gap is no longer a planning oversight. It is now a documented structural failure, according to a February 2026 survey of 120 health systems by Eliciting Insights.
The survey puts the AI adoption rate at 75%, a 27% year-over-year increase. Meanwhile, only 18% of organizations reported mature AI governance - defined as a documented strategy backed by a formal governance group. "Health systems are embracing AI to address workforce constraints and financial pressures, moving beyond pilots into strategic deployment," said Trish Rivard, CEO of Eliciting Insights.
AI runs live - while oversight stays on the planning deck
Ambient documentation tools sit in exam rooms, revenue cycle models process live claims, and clinical decision support is embedded across specialties. At 42% of health systems, there is neither a governance strategy nor the organizational structure to enforce one. These organizations operate AI tools with no formal accountability framework.
The rapid scaling compounds the risk. The number of health systems running three or more AI tools grew 67% year over year, based on data from the HFMA/Eliciting Insights Health System Readiness report. Yet 35% reported having no formal AI data policy. The report concluded that "health systems are investing in AI solutions before establishing internal governance, data policies, or adequate resources for responsible operation."
Vendor reliance becomes the default governance
When health systems don't build internal oversight, they default to vendor-defined rules. Seventy percent of programs without mature governance rely on vendors to define how AI is used, according to HFMA data. That effectively shifts the governance function - and the liability for AI errors - outside the organization.
Regulatory pressure is also tightening. A 2026 HIPAA Security Rule overhaul brings AI systems that handle protected health information into explicit compliance scope. Revised FDA Clinical Decision Support Software guidance, published in January 2026, narrows the exemptions that previously kept many deployed AI tools outside regulatory review.
Health systems that cannot staff internal AI monitoring teams are starting to look outward. The governance gap creates an operational opening - sharpest in revenue cycle management, where AI-generated outputs demand continuous validation. AI for Medical Billers training prepares revenue cycle staff to monitor AI outputs, spot errors, and maintain the audit trails that regulators will increasingly require.
Why this matters for healthcare leaders
The deployment-and-forget approach is no longer viable. Every AI tool running without a governance body means the organization is answering for outputs it never reviewed. For revenue cycle leaders, the immediate priority is building the internal skill to audit AI-generated claims and coding decisions - or contracting with partners who can do it under a clear accountability framework. For clinical leaders, the FDA's narrowed exemptions mean decision support tools that once needed no oversight may now fall into regulated territory. The deployment is done. The responsibility for what AI does next is still open - and health systems that establish mature governance before regulators force their hand will control the terms.
Your membership also unlocks:
