CPT 2026 AI Codes: What Medical Coders Need to Know

The AMA released the CPT 2026 code set with 288 new entries, and a quiet but significant shift sits in the middle of it: AI-augmented services now have their own permanent home in Category I. For coders watching how AI gets billed in real-world workflows, the CPT 2026 AI codes mark a turning point. Augmentative software is no longer a Category III experiment. It is a recognized clinical service.

A New Category for AI-Augmented Services

The CPT 2026 update, effective January 1, 2026, includes the first set of Category I codes designated for augmented intelligence services. The AMA's framing is deliberate. Rather than treat AI as a separate diagnostic modality, the new codes describe services that augment a physician's interpretation. Software that quantifies, characterizes, or surfaces data the physician would otherwise have to derive manually.

The Category I designation matters. It means these services have demonstrated clinical efficacy and sufficient provider use to clear the same bar set for any established procedure. The bar is not low, and clearing it changes how the service gets paid, audited, and reviewed for the rest of its life as a billable code.

Code 75577: From Category III to Category I

The headline change is CPT 75577, which covers quantification and characterization of coronary atherosclerotic plaque derived from a coronary CT angiography (CCTA) data set. Since 2021, this work was reported with the temporary Category III codes 0623T-0626T. As of January 1, 2026, those temporary codes are deleted.

CMS finalized a 2026 Physician Fee Schedule payment of more than $1,000 for the office and imaging-center setting, with the Hospital Outpatient Prospective Payment System rate set at $950.50. Major commercial payers including Aetna, UnitedHealthcare, Cigna, and Humana already cover the analysis. Coders should expect the volume on 75577 to climb quickly through the year.

The shift from Category III to Category I is not just bureaucratic. Category III codes carry no assigned RVU, and reimbursement is left to local payer discretion. Category I codes are priced into the fee schedule, audited by CMS, and incorporated into claim edits. Every claim with this code now lands in a much more structured review pipeline.

How Coders Should Approach the New Codes

Coders in cardiology, radiology, and imaging-heavy specialties should expect questions from clinicians and billers as the codes go live. A few practical points are worth surfacing now.

Documentation requirements

The descriptor language is precise. 75577 requires both quantification and characterization of plaque, derived from augmentative software analysis of a CCTA dataset, with interpretation and report by a physician or other qualified health care professional. A radiologist's standard CCTA read is not enough on its own. The AI software output must be incorporated into the report, and the physician must review and sign off on the AI-derived findings.

Avoiding common mistakes

• Don't bundle 75577 with the underlying CCTA code without confirming separate medical necessity. The AI analysis is a distinct service.

• Don't carry over the old Category III codes into 2026 dates of service. Claims using 0623T-0626T after January 1 will be denied.

• Watch for payer-specific coverage policies. Category I status doesn't always remove prior authorization requirements.

• Confirm the software used is FDA-cleared for the specific indication. Some payer policies tie reimbursement to a list of approved tools.

• Train front-end teams to capture the AI tool name and version in the report. It is the cleanest defense in an audit.

What This Signals for the Future of Medical Coding

75577 is the first AI-augmented Category I code in cardiac imaging, but it will not be the last. The CPT 2026 set also adds augmented-intelligence codes for perivascular fat analysis and additional imaging quantification work, and the AMA has signaled a broader framework for how augmentative software gets evaluated over time.

The pattern is clear. As AI tools accumulate clinical evidence, they migrate from Category III to Category I. Coding teams need processes ready to handle that transition without lapses in revenue or compliance. The teams that get ahead of this won't be the ones with the most coders. They'll be the ones with the most adaptable workflows.

For coders, the takeaway is twofold. First, mapping the right code to the right service is now bound up with understanding what the AI did inside the workflow. Second, AI is not just a tool that helps coders. It is increasingly a service that coders bill for. Both shifts are in play in the same code set.

Staying Ready for the Next Update

Keeping pace with code-set updates is exactly the kind of work that benefits from automated review. Medikode's automated medical coding platform helps coding teams stay current with new CPT and ICD-10 entries, flag claims that risk denial under updated guidelines, and reduce the manual lift of audit prep. As CPT 2026 AI codes start showing up on claims throughout the year, having a system that learns the new rules with you is the difference between catching issues at submission and chasing them on the back end.

Source: AMA releases CPT 2026 code set.