AMA’s September CPT Panel Tightens AI Rules for New Lab Codes
The American Medical Association’s CPT Editorial Panel convenes September 17-19, 2026, in Minneapolis for its third quarterly meeting of the year, and the agenda includes a change that coders and lab compliance teams should watch closely: a substantially revised Code Change Application for laboratory and pathology codes (the application’s “Tab 94” section) that now asks applicants far more detailed questions about artificial intelligence and software validation before a new code can move forward.
What Changed in the Application
Per the AMA’s own CPT Editorial Panel meeting calendar, panel review of code change applications begins roughly a month before each quarterly meeting, which put the revised lab/pathology application in front of applicants over the summer. Health-policy researcher Bruce Quinn, writing on his Discoveries in Health Policy blog on September 6, 2026, flagged the update as unusually large, noting the application is now dense enough that reading the redlined version is difficult without AI-assisted analysis of the markup itself.
According to that analysis, the revision touches at least seven distinct areas of the application, including new questions tied to the FDA’s Laboratory Developed Test (LDT) framework, expanded documentation requirements across what the analysis describes as three separate AI-related pathways, a rewritten set of CLIA-related questions, and clarified language on evidence and clinical-utility claims that applicants must now substantiate.
Why AI Validation Is Now Front and Center
The FDA-LDT Connection
Laboratory developed tests have moved in and out of FDA’s direct oversight for years, and many of the newer tests seeking CPT codes rely on algorithmic or AI-assisted interpretation layered on top of the underlying assay. Tying the CPT application to FDA’s LDT framework means applicants proposing an AI-driven diagnostic test now need to document not just clinical validity, but how the software component itself was validated — a bar that plain assay-only tests didn’t previously have to clear in the same way.
A Compressed Timeline
The comment and submission deadlines are tight relative to the scale of the changes: pathology-specific comments closed in July 2026, the general (non-pathology) comment deadline was August 10, 2026, and the final request deadline for this cycle is September 16, 2026 — one day before the panel convenes, with a dedicated Path/Lab subcommittee session scheduled for 7 a.m. on September 17.
What This Means for Coders and Compliance Teams
Coders themselves don’t file CPT code change applications, but the ripple effects land squarely in their workflow once new lab and pathology codes emerge from this cycle:
- New AI-adjacent lab codes are likely to carry more detailed descriptor language tied to how the test’s algorithmic component was validated — coders will need that documentation on hand to support code selection, not just the lab report itself.
- CLIA-related documentation questions in the application suggest CMS and AMA are converging on a shared evidentiary bar between lab certification and code assignment, which could tighten what counts as adequate medical-necessity support for claims.
- Denial and audit risk rises for any AI-based lab test billed under a general or unlisted code while a specific code is pending — payers are increasingly scrutinizing algorithmic diagnostics, and RADV-style extrapolation exposure applies to risk-adjusting conditions identified through these tests just as it does to any other diagnosis code.
- Compliance teams tracking the AMA’s published Summary of Panel Actions after the September meeting should specifically watch for new or revised lab/pathology codes and read their full descriptors before building charge-capture rules around them.
The Broader Signal
This is a narrower, more technical story than a full code set release, but it fits a pattern this year: regulators and standards bodies are asking AI systems that touch clinical decisions — whether that’s an algorithmic diagnostic test or an agentic coding assistant — to show their validation work, not just their output. The same expectation increasingly applies inside RCM and coding departments themselves. As agentic AI tools take on more first-pass coding and CDI work, the documentation trail proving how a code was derived is becoming as important to compliance teams as the code itself.
Coding leaders don’t need to wait for the panel’s Summary of Panel Actions to start that conversation internally — reviewing how any AI-assisted coding tool documents its own reasoning, and whether that trail would satisfy an auditor, is a reasonable exercise to run in parallel with watching what comes out of Minneapolis this month. It’s also a useful prompt for evaluating vendors: any coding or CDI platform that can’t explain, in plain language, why it selected a given code is asking a compliance team to take the same leap of faith AMA is now asking lab-test applicants to stop taking for granted.
Medikode’s automated medical coding platform is built with that documentation trail in mind, giving coding and compliance teams visibility into how each code recommendation is derived — Medikode’s automated medical coding platform can help teams stay ahead of exactly this kind of scrutiny as AI-driven diagnostics and AI-driven coding both face closer review.