A patient needs an MRI. The provider’s system says ‘yes.’ The payer’s system says ‘no.’ Two AIs, two outcomes, zero coordination.
The healthcare industry is racing to modernize utilization management (UM), but in many ways, it’s heading in the wrong direction. Instead of fostering collaboration between payers and providers, emerging AI models are being deployed in ways that reinforce historic silos.
Some technologies are positioning AI as a representative of one party over another; accelerating approvals for providers or enforcing controls for payers. For example, certain EHR-integrated solutions are trained to interpret medical policies through a provider-centric lens, while payer-side systems may develop AI to flag high-risk or high-cost requests. The result? A looming AI arms race that risks turning prior authorization into an adversarial battleground that will delay patient care, increase provider burnout, and fuel distrust in a system already under strain.
The risk of an AI arms race
What does an AI arms race look like in healthcare? Imagine a scenario where a provider-side AI rapidly approves requests based on loosely interpreted clinical data, while a payer-side AI simultaneously flags those requests for potential denial, fraud, or overutilization. The two systems, designed with different incentives and decision logic, enter a continuous loop of escalation, audits, and appeals. The result is a system that prioritizes algorithmic gamesmanship over aligned, patient-centered outcomes.
This risk looms large as the industry works toward ambitious modernization goals. The Centers for Medicare & Medicaid Services Interoperability and Prior Authorization Final Rule (CMS-0057-F) and the pledge between AHIP and the Blue Cross Blue Shield Association to improve the prior authorization process both call for a future where authorizations are real-time, transparent, and digital.
CMS-0057-F mandates Fast Healthcare Interoperability Resource (FHIR)-based APIs and timely responses from payers by 2027, while the AHIP pledge represents a collective industry commitment to improving provider experience and patient access. But technology isn’t enough. If the underlying systems are adversarial, automation will only accelerate the dysfunction.
Designing AI that collaborates
There is a better way. Rather than using AI to compete, we must design AI to collaborate.
That begins with transparency. AI systems that support UM must show their work. Recommendations should be traceable to clinical criteria, aligned with codified medical policy, and auditable by clinicians on both sides of the transaction. CMS and AHIP have both emphasized clarity in decision rationale and real-time reporting. In this environment, black-box models present not only a clinical risk but also a compliance liability.
Second, AI must be interoperable in more than just the technical sense. APIs and standards like FHIR are essential, but insufficient. True interoperability means that AI can operate within the workflows of providers, payers, and vendors alike, guided by shared governance and structured policy alignment. More than just API connectivity, interoperability requires data normalization, contextual policy application, and a willingness to participate in federated systems where control is balanced with collaboration, without favoring one over the other.
Third, AI must be embedded, not bolted on. That means placing intelligent decisioning directly within the point-of-care and administrative workflows where UM happens. For example, a provider using an EHR should be able to submit, track, and receive authorization decisions without ever leaving the clinical interface. This level of integration reduces friction, accelerates decisions, and promotes adoption across all stakeholders.
Health plans must lead with clarity
If health plans don’t take the lead, they risk being disintermediated or relegated to passive participants in decisions they should be guiding. EHRs and third-party vendors are already embedding policy-based AI into their platforms, creating fast-track approval paths that may bypass the payer entirely. This shift threatens payer oversight, data ownership, and their ability to enforce evidence-based standards across care delivery. In a fragmented landscape, disintermediation fractures accountability. Decisions made within payer UM teams could increasingly occur inside opaque clinical workflows, where payers have limited visibility, traceability, or recourse.
To prevent this, health plans should control and insist on transparency while retaining ownership of their medical policies and determine how those policies are applied, even when implemented through third-party platforms. Without retaining governance over how medical policies are applied, health plans risk inconsistent determinations, increased provider abrasion, and diminished trust in their UM processes.
As the regulatory landscape evolves and industry pledges push for real-time, electronic, and transparent authorizations, the imperative is clear: technology must bridge the payer-provider divide, not widen it. The next era of UM will not be won by the fastest algorithm or the flashiest AI. It will be led by those who prioritize interoperability, transparency, and shared accountability.
Health plans should start by auditing their current UM technology stack for transparency, investing in shared decision frameworks, and partnering with platforms that support bi-directional integration, ensuring they remain active stewards of their clinical and operational standards.
The future of utilization management isn’t just faster; it’s fairer, clearer, and smarter. It’s a future where decisions are clinically sound, auditable, and trusted by all sides.
Photo: Mironov Konstantin, Getty Images
This post appears through the MedCity Influencers program. Anyone can publish their perspective on business and innovation in healthcare on MedCity News through MedCity Influencers. Click here to find out how.