Key Insights on the Use of AI in the Audit and Compliance Industry

What’s happening in healthcare today regarding artificial intelligence (AI) and auditing?

What I am seeing across the auditing landscape is the expectation that AI will become a primary auditing method and tool in the future.

With AI, we can expect less manual data entry, the ability to audit more claims or larger statistically valid samples, and more robust review processes with less effort. We are also hoping to search medical records more effectively without relying entirely on manual chart review.

Ideally, AI-enabled software will help determine audit results by answering questions such as: Are we meeting local and national coverage determinations? Does the chart note support medical necessity? Are all required elements present for behavioral health programs under Medicaid? Will the language in the note translate to a clean claim?

We have reached the point where it is clear that AI will fundamentally change how we do our work. That is exciting, but it also creates understandable uncertainty, as employees ask, “What happens to me?”

1. Compliance and Auditing When Reviewing More (or all) Claims Before Submission

I agree that AI will significantly expand our ability to perform front-end review before claims are submitted. Imagine being able to validate medical necessity within the chart note prior to submission. This could help reduce denials, prevent reduced payments, and identify financial opportunities.

At the same time, this work will still depend on experienced auditors to validate AI-generated results. As AI “learns,” its outputs must be verified. Terminology varies across the country, and inconsistencies in clinical documentation could have a greater impact in an AI-driven process. A single phrase or word could send an AI auditing program in different directions, and one unintentional error could potentially flag thousands of claims.

For that reason, setup, oversight, and ongoing monitoring will be critical. I do not necessarily expect compliance standards themselves to change – either something is compliant, or it is not – but I do expect AI to uncover more issues. The larger question will be how organizations manage what they find.

2. What are Organizations Underestimating?

I expect AI to be especially useful for auditing specific areas of risk, including disease implications, procedures, clinical indicators, and patterns of consistency or inconsistency that can support both population health and compliance efforts.

However, I do worry about unintentional bias related to demographics or culture. How well will AI be able to rationalize clinical complexity? For example, literacy is often associated with better health outcomes, but that assumption may not hold when other factors, such as a mental health diagnosis or a congenital condition, are involved. We will need to ensure that AI does not make recommendations or decisions based on isolated data points or flawed interpretations of clinical documentation.

Clinical judgment remains central to coding and auditing. The interpretations coders and auditors make in assigning codes are a defining part of their role, and a valuable skill set. The key question is whether those nuances can be captured accurately through AI. If we make an exception for a clinical concept in one audit, will that same concept be treated appropriately in another?

Healthcare coding and auditing are not black-and-white, which is why the phrase “trust but verify” will absolutely apply here.

My additional concerns include:

• Frequent updates: Codes, coverage requirements, and Federal Register changes must be validated regularly, sometimes annually and sometimes quarterly, and software must be audited to confirm those updates are applied correctly.
• Cost and time: keeping AI tools accurate and current requires ongoing investment.
• Privacy and security: Whether organizations access federal, state, and commercial payer information databases externally or integrate them into internal systems, they must protect sensitive patient information.
• Fraud and abuse risk: AI-driven data aggregation could create new opportunities for fraud, abuse, or data breaches if controls are not strong enough.
• Governance requirements: Encryption, audit trails, and role-based access must be thoroughly tested and consistently maintained.
• Operational readiness: If AI uncovers widespread issues, organizations must be prepared to respond quickly and effectively.

3. What Information am I Sharing with my Team, and How am I Helping them Manage the Shift?

It goes without saying that AI will change resource needs and required skill sets across coding, billing, and auditing roles. However, I also see this as an opportunity for employees to develop higher-level capabilities. Hospitals should consider offering skills-based education for individuals who want to learn, adapt, and grow.

Turnover is costly, and we want to retain employees who understand not only the revenue cycle, but also organizational culture. Because validation of healthcare data will become an increasingly important function in an AI-enabled environment, I have encouraged my staff to strengthen their critical-thinking skills. They are also continuing management coursework, keeping up with required continuing education, and contributing to projects and initiatives across the organization.

We are actively discussing what we want our day-to-day work to look like as AI projects expand. Most of us already use AI as a search tool in some capacity. Auditors who are comfortable reviewing services through a work queue may find that expectations regarding productivity, performance, and audit scope will shift. I expect we will be asked to do more with less, and that will require broader, stronger skill sets. In my view, that can be a win-win.

4. What Do I Hope Hospitals, Coders, and Auditors Take Away from Today’s Conversation?

For auditors, I would encourage you to see this as an opportunity to raise your game. This shift appears to create significant opportunities for professionals interested in data analysis, risk assessment, provider education, and denial management.

For hospitals and practice leaders, my caution is simple: do not be dazzled by the sales pitch. Involve your coding, billing, audit, education, and compliance teams in evaluating the advantages and disadvantages of any AI product. Give those teams enough time to test AI-driven software thoroughly.

Because AI will affect workforce structure, hospitals should assess resource changes, evolving skill requirements, and return on investment. They should also establish clear policies and procedures to protect privacy and maintain compliance, and determine whether legal counsel should help assess liability related to AI misinterpretation.

Overall, AI has the potential to be tremendously helpful for healthcare auditing professionals, and the opportunities appear nearly endless. At the same time, it will not lessen our commitment to compliance; in many ways, it will require us to think about compliance differently. This shift is exciting, transformative, and full of possibility, and I am eager to see what comes next.