Fighting Back Against Artificial Intelligence-Driven Downcoding
Key Summary
- Attorney Jason Greis discusses how artificial intelligence (AI)-driven downcoding is affecting physician practices, highlighting repeated audits or downcoding of the same CPT codes as key warning signs.
- Greis cites an estimate that ~10% of claims are now affected by AI in some way, including automated downcoding and medical-necessity reviews, with commercial payers adopting AI aggressively.
- Tracking patterns, improving documentation and data analytics, and avoiding copy-and-paste documentation are recommended.
© 2026 HMP Global. All Rights Reserved.
Any views and opinions expressed are those of the author(s) and/or participants and do not necessarily reflect the views, policy, or position of Vascular Disease Management or HMP Global, their employees, and affiliates.
VASCULAR DISEASE MANAGEMENT. 2026;23(8):E125-E126
Benesch Law, Chicago, Illinois
As commercial payers increasingly turn to artificial intelligence (AI) to review claims and adjust reimbursement, physician practices are facing a new set of financial, operational, and legal challenges. Vascular Disease Management spoke with Jason Greis, partner at Benesch Law in Chicago, about his presentation at the 2026 Amputation Prevention Symposium in Boston on AI-driven downcoding. He discusses how practices can recognize patterns of AI-driven downcoding, respond strategically to payer audits, and build the documentation and data infrastructure needed to challenge inappropriate payment reductions. He also explains why practices should treat repeated downcoding as a broader revenue-protection issue rather than isolated claim disputes.
What are some warning signs that a payer might be relying too heavily on AI when downcoding claims?
When you are consistently getting audits for the same type of issue over and over and over again, whether it’s for a specific CPT code and you are seeing downcoding in levels, that's without a doubt the biggest red flag.
Another red flag is getting a letter from the payer’s investigations unit. If you are getting very generic audit letters that don't say a whole lot in terms of, are you being investigated due to technical issues, or things as simple as missing documentation, missing clinical signatures, or that don't provide information that they are looking at a clinical issue, that's another one.
If a practice notices a pattern of repeated downcoding, at what point should it consider escalating the issue beyond just individual appeals?
As soon as possible. If you think there’s something going on, reach out to them first if they have a person in the payer's claims department, though those people can be difficult to get ahold of. The next best source of access would be to reach out to the medical director to say, hey, what is going on here? We are starting to get a lot of these; are they clinical issues? Are they technical issues? Are you using AI? Is this a result of AI to figure out what’s going on? Once you know you're dealing with an AI issue, depending upon the state you’re in, you may have some recourse. Some states have always had laws prohibiting the use of AI, and a growing number of states are also adopting new laws along those lines.
How much have downcoding issues increased since AI came into being? Have they grown a lot?
Exponentially. Originally, Medicare said they were going to adopt AI but were going to look at it slowly. Well, they haven’t done it slowly. They’ve rolled out a number of different initiatives at the Medicare level, and they are continuing to roll them out. The commercial payers have been much more aggressive in terms of rolling AI out. They tend to move first and then ask for forgiveness later.
The statistic that I heard last was roughly 10% of claims are now impacted in one way or another by AI to some extent. It could be automated downcoding, or it could be AI taking a look at medical necessity. It’s growing quickly.
When responding to a commercial payer audit, what can a practice do to put itself in the strongest position to negotiate?
Number one, if it is verified that there’s an AI issue, having the data to support that there is an AI issue. Number two is making the argument that it is inappropriate to use AI in a certain capacity. There is supposed to be a physician involved looking at all these claims, deciding whether downcoding is appropriate. If they have removed the medical director from that process, or if they are including the medical director but it’s a matter of them batching together 300, 400, 500 claims and the physician is just clicking one button without spending a lot of time reviewing the claims, that’s inappropriate.
Also important is working with trade organizations to develop a toolkit. At least one trade organization in the nephrology space, the Renal Physicians Association, has developed a toolkit and letters for physicians to send out to say, hey, we’ve noticed that you’re using AI and, if you think it's inappropriate, stop it. Some practices have had some degree of success in getting the payers to discontinue the practice, at least for now.
As AI becomes more common in payer decision-making, what should physician practices be doing now to protect themselves?
Have better data analytics, which can be difficult. Do the best that you can in terms of making sure you are describing accurately the acuity of the disease state, that you are being as descriptive as possible, and you are avoiding simple cut-and-paste from one patient to another. That tends to trigger a lot of AI-related issues. Many small practices don’t have a whole lot in the way of data analytics, but it’s going to become more important for groups to have that capability. n
