top of page
Sirius Solutions Global website header with navigation menu: Home, Services, Specialties, Our Expertise, Resources, and Contact Us.
"Sirius Solutions Global Logo"

2027 CPT AI Coding Updates: Understanding the 43 AI-Related Codes and Their Billing Implications

1 day ago
6 min read
Sirius Solutions Global banner about 2027 CPT AI coding updates, with a clinician using a laptop in a blue tech setting.

A practice adds software that flags suspicious findings on scans. Is that a billable service, a tool that supports an existing service, or something with its own code? CPT 2027 gives billing teams more to work with. On September 9, 2026, the AMA released the code set with 10 new AI-related codes, bringing the total to 43, along with a revised taxonomy for classifying AI services.

The distinction matters because code selection, documentation, payer policy, and reimbursement all depend on what the software output actually is and what the clinician does with it. Not every AI tool has a code, and a code does not guarantee payment. This guide focuses on what is verified, what is not public, and how to prepare. The 43 is the AMA’s reported total, and not all 43 are new.

CPT 2027 AI Update: Quick Facts

CPT 2027 update

Verified information

Total AI-related CPT codes

43

Additional AI-related codes in 2027

10 (so 33 before the update, by subtraction)

Effective date

January 1, 2027

Updated AI classification framework

Appendix S, revised by the CPT Editorial Panel in May 2026

AI application categories

Assistive, augmentative, autonomous

Understanding the Three AI Taxonomy Categories

Per the AMA, the taxonomy guides code descriptors and classifies software by its output and the output’s role in clinical care. The output must be useful in diagnosis, treatment, or prevention. The AMA also notes the term “AI” is neither sufficient nor necessary to describe a service.

Infographic with arrow for increasing software independence: Assistive, Augmentative, and Autonomous Levels I–III panels with text.

Classification

AI function

Clinical output

Professional involvement

Coding consideration

Assistive

Provides clinically relevant data

No derived parameter, interpretation, or conclusion

Physician or QHP interprets and reports

Improves performance; the primary service may be unchanged

Augmentative

Derives a parameter that differs from the input

Score, index, category, or risk metric; more than a sum or average

May interact with software; no independent interpretation

Physician work typically captured by existing codes (e.g., within E/M or surgical planning)

Autonomous

Derives parameters and generates interpretations without concurrent involvement

Definitive diagnosis or specific management recommendation

Level I: judgment to implement or reject. Level II: chance to negate. Level III: oversight of automatic actions

Review code descriptor and level carefully

Definitions are paraphrased from the AMA’s Appendix S, and the Appendix S table is available there. Use the licensed CPT text for exact language.

What Do the 43 AI-Related CPT Codes Actually Represent?

A limit we want to be upfront about

The AMA publicly reports 10 new codes and a total of 43. The individual code numbers for the 10 new codes are not on the AMA’s free pages, so we do not list them, and we have not independently verified the full inventory of 43. We will not guess numbers. Use the CPT 2027 codebook or data files to build your list.

For any code you evaluate, confirm: the clinical service, the software’s function, how its output is classified, the clinician’s role, whether the code describes the software-enabled service or a different service, and any extra reporting rules.

What Changed in 2027? Previous Approach vs. 2027 Updates

Area

Before

CPT 2027

AI-related codes

33 (43 minus 10)

43

Taxonomy

Appendix S since 2021

Revised May 2026 for clearer category distinctions

Clinically meaningful output

Less precisely defined

Defined more precisely

Billing implication

Check the code and payer

Same, with sharper category language

This does not mean every AI tool now has a billable code.

How Could the 2027 AI CPT Updates Affect Medical Billing?

•      Code selection: identify the actual clinical service and the software’s classification before choosing a code.

•      Documentation: record the service, software involvement, outputs, and the clinician’s interpretation.

•      Medical necessity: using AI does not by itself justify a separately reportable service.

•      Coverage and payment: a code’s existence does not guarantee Medicare or commercial payment. Check Medicare policies, the fee schedule where applicable, commercial and Medicaid policies, and contracts.

•      Category III codes: temporary codes for emerging technology. Payment varies by payer, so they are neither universally paid nor universally denied. Confirm individual payer policy.

•      Duplicate billing: do not report work already included in another service where CPT rules prohibit it.

•      Denials: risks include wrong code selection, thin documentation, missing authorization, and unsupported payment assumptions.

Which AI Category Describes the Clinical Application?

Educational classification exercise, not an official coding determination. Compare the software output with each row.

If the software…

Category

Professional involvement

Coding consideration

Gives clinically relevant data without a score or conclusion

Assistive

Clinician interprets and reports

Often supports an existing service

Produces a validated score, index, or category that is more than a summary

Augmentative

Possible interaction; no independent interpretation

Work often captured by existing codes

Independently reaches a diagnosis or management recommendation

Autonomous

Depends on Level I, II, or III

Verify code, level, and payer policy

Real-World Billing Scenarios

Hypothetical examples. We assign no CPT codes because none have been verified for these cases.

Scenario

Software and clinician roles

Likely category and review

1. AI-assisted information

Software highlights regions of interest. The physician interprets and reports.

Assistive. Verify that the report reflects physician interpretation and that no separate code is assumed.

2. AI-generated parameter

Software converts inputs into a risk score the physician uses in management.

Augmentative. Check whether the work belongs in an existing service and whether validation is documented.

3. Autonomous interpretation

Software independently reports a diagnostic conclusion; a clinician oversees the workflow.

Autonomous. Confirm the level, the code descriptor, payer policy, and documentation of oversight.

Is Your Billing Team Ready for CPT 2027 AI Coding Updates?

☐ Reviewed the 2027 CPT changes

☐ Updated relevant coding references

☐ Reviewed Appendix S

☐ Understand the three AI taxonomy categories

☐ Identified AI-enabled services used by providers

☐ Reviewed clinical documentation requirements

☐ Checked applicable payer reimbursement policies

☐ Reviewed Category III reporting requirements

☐ Evaluated potential duplicate billing risks

☐ Updated billing team training

☐ Reviewed claim edits and denial workflows

☐ Established an ongoing CPT update review process

Completed: ____ of 12 (divide by 12 for your percentage).

How AI CPT Updates May Affect Revenue Cycle Management

RCM area

Potential challenge

Recommended response

Coding

Incorrect classification

Coding review

Documentation

Insufficient clinical detail

Documentation review

Claims

Incorrect code reporting

Pre-submission validation

Eligibility

Coverage-related issues

Eligibility verification

Denials

Coding or policy-related denials

Root-cause analysis

A/R

Unresolved balances

Structured follow-up

Compliance

Inconsistent reporting

Periodic coding audits

Reporting

Limited billing visibility

RCM performance monitoring

These updates need coordination among clinical teams, coders, billing staff, and payer-policy reviewers.

Preparing Your Medical Billing Operations for AI-Enabled Healthcare

Sirius Solutions Global is a medical billing and revenue cycle management partner offering medical billing, medical coding, advanced eligibility verification, denial management, prior authorization, credentialing and re-credentialing, A/R management, billing audits, revenue leakage assessments, and claims management. Technology-assisted workflows work best with human verification, because classification and payer questions need judgment. An experienced team can review coding workflows, find inconsistencies, improve claim accuracy, and monitor performance. We do not offer an AI coding product here, and we cannot guarantee reimbursement for any AI-related service.

How Prepared Is Your Practice for AI-Related Coding Changes?

Question

Yes

No

Has your coding team reviewed the CPT 2027 AI updates?

☐

☐

Have you identified AI-enabled clinical services used in your practice?

☐

☐

Are you familiar with the three Appendix S categories?

☐

☐

Have you reviewed payer policies for applicable services?

☐

☐

Are clinical documentation workflows being reviewed?

☐

☐

Are coding and billing staff receiving updated training?

☐

☐

Do you regularly audit claims for coding inconsistencies?

☐

☐

 

Yes answers

Educational interpretation

0–2

A structured review of your AI coding readiness may help

3–5

Consider reviewing specific coding and policy areas

6–7

Several review processes appear to be in place; keep monitoring

Educational only, not a compliance certification or guarantee. Request a billing workflow review.

Frequently Asked Questions

▸ What are the 43 AI-related CPT codes?

The AMA reports 43 AI-related codes after CPT 2027. The full list is in the CPT codebook and data files, not on free AMA pages.

▸ How many new AI-related codes were introduced in 2027?

Ten.

▸ When do the updates take effect?

January 1, 2027.

▸ What is CPT Appendix S?

The AMA’s taxonomy for classifying AI medical services as assistive, augmentative, or autonomous, to guide code descriptors.

▸ What is the difference between assistive and augmentative AI?

Assistive software gives clinically relevant data without a derived parameter. Augmentative software derives a clinically meaningful score or category.

▸ What does autonomous AI mean?

Software that derives parameters and independently generates interpretations, at three levels of professional involvement.

▸ Are AI-related CPT codes reimbursable by Medicare?

It depends on the specific code and coverage policy. We did not verify Medicare coverage for the new codes.

▸ Does an AI-related code guarantee payment?

No. Payment depends on payer policy, documentation, and contracts.

▸ How do the updates affect medical billing?

Mainly through code selection, documentation, and payer-policy checks.

▸ What documentation matters for AI-enabled services?

The service performed, the software’s output, the clinician’s role, and medical necessity.

▸ Are Category III AI codes automatically reimbursed?

No. Payment varies by payer.

▸ How can billing companies prepare?

Update code references, review payer policies, train staff, and audit claims.

Editorial note: CPT descriptors and payer policies change. Verify details in the licensed CPT 2027 codebook and each payer’s policy. Technical content like this benefits from review by a certified coder.

Disclaimer: This article is educational, published by Sirius Solutions Global, and is not coding, legal, or compliance advice. CPT® is a registered trademark of the AMA, and Appendix S text here is paraphrased. Scenarios are hypothetical. Sirius Solutions Global is not affiliated with the AMA or CMS. Nothing here guarantees coverage, reimbursement, or audit outcomes.

bottom of page