AI in Medical Billing Just Got Official: Inside the 10 New AI-Related CPT Codes
Artificial intelligence is no longer a side project in healthcare. It reads scans, flags risk, scores patients, and supports clinical decisions every day. Now the CPT code set is catching up. When the American Medical Association (AMA) released the CPT 2027 code set, it added 10 new AI-related codes, bringing the total number of AI-related CPT codes to 43. The AMA also updated its taxonomy for classifying AI in medical services.
For providers, coders, and revenue cycle teams, this is a practical change. Every AI-enabled service has to be documented, classified, and billed correctly. This guide explains what the 10 AI-related CPT codes 2027 mean, how the AI classification framework works, and what your practice should do before January 1.
Why CPT 2027 Marks a Turning Point for AI in Healthcare
For years, AI in healthcare lived mostly in pilot programs and vendor demos. CPT 2027 shows how far it has moved. With 43 AI-related codes now in the code set, AI-supported services have become part of routine, reportable care.
The AMA describes the CPT code set as more than a billing tool. It serves as the backbone of health data interoperability, supporting value-based care, quality measurement, research, and analytics. That is why AI coding matters beyond reimbursement. When AI services carry a standardized code, payers, researchers, and health systems can finally measure how they are used and what they deliver.
The connection to money is direct. A service without a clear code is hard to track, hard to contract for, and hard to get paid for. Each new code gives providers a defined way to report AI-supported work, and gives payers a defined way to review it.
What are the 10 AI-Related CPT Codes 2027?
The AMA’s release states that the CPT 2027 code set continues to expand its description of AI-related services that support or assist healthcare professionals. Ten new codes are added this year, which is part of a larger release containing 299 new codes, 74 revisions, and 80 deletions.
| CPT 2027 Snapshot | Figure |
|---|---|
| Total new codes | 299 |
| Revised codes | 74 |
| Deleted codes | 80 |
| Total editorial changes | 453 |
| New AI-related codes | 10 |
| Total AI-related CPT codes | 43 |
The codes take effect January 1, 2027. Exact descriptors and code numbers should be confirmed in the CPT 2027 codebook or the AMA’s data files, since coding decisions must always rely on the final published language. These AI healthcare CPT codes exist to capture services where software output contributes to diagnosing, treating, or managing a condition, and they reward accurate reporting because payers will look closely at how AI contributes to care.
Clinical Areas Impacted by the New AI Codes
AI-related codes typically touch several kinds of work. These are the areas where our coders expect the most questions:
| Clinical Area | How AI is Used |
|---|---|
| Diagnostic support | Software highlights findings or generates likelihood scores |
| Imaging interpretation | Algorithms analyze images and quantify results |
| Predictive analytics | Risk scores and prognostic indices from patient data |
| Clinical decision support | Recommendations to guide diagnosis or management |
| Automated workflows | Software that initiates or coordinates care actions |
Understanding AMA’s AI Classification Framework
The backbone of AI coding is CPT Appendix S: Taxonomy for Artificial Intelligence in Medical Services and Procedures. It helps applicants and other stakeholders choose the right terminology for CPT code descriptors by sorting AI-enabled services into one of three categories: assistive, augmentative, or autonomous.
Why does this matter? The AMA notes that no single term such as “AI” is enough to describe a service’s clinical purpose, so the taxonomy is based on the software’s output and the role that output plays in patient care. Appendix S was introduced in 2021, and at its May 2026 meeting the CPT Editorial Panel accepted revisions to clarify the differences between the three categories and to sharpen what counts as a clinically meaningful output.
Assistive AI
Assistive software provides clinically relevant data without deriving a score or category, and without generating an interpretation or conclusion. The physician or other qualified health professional still interprets the output and writes the report. Assistive tools can improve accuracy, precision, and consistency between readers, even if the primary service stays the same. Language such as “likelihood of” or “suggestive of” often appears in their output.
Augmentative AI in Healthcare
Augmentative software produces a quantitative or categorical result that is different in kind from its input, such as a clinical scale, index, risk score, or classification. It must offer more than adding, averaging, or measuring. It does not include an independent interpretation or conclusion. The AMA notes that physician work tied to augmentative services is typically captured by existing codes, and the software output might feed into an E/M service, surgical planning, or another service that includes interpretation. Documentation should show that the output contributes to patient management.
Autonomous AI in Healthcare
Autonomous software derives parameters and independently generates clinically meaningful interpretations without concurrent physician involvement. It has three levels:
| Autonomous Level | What the Software Does | Physician Role |
|---|---|---|
| Level I | Recommends a definitive diagnosis or specific management | Physician judgment to implement or reject |
| Level II | Recommends management based on its conclusions, with a chance to negate the action (for example, via an alert) | Opportunity to intervene before action |
| Level III | Automatically initiates management actions | Oversight and performance review over time |
Autonomous tools carry the highest regulatory and compliance weight, and reporting derived parameters is essential for transparency and explainability.
Also Read: AI Agents vs Traditional RCM Automation: What’s the Difference?
How AI Classification in Healthcare Affects CPT Reporting
Identifying the right category is not optional. It affects which code applies, what the code descriptor requires, and what evidence an auditor will expect. An assistive tool and an autonomous tool can look similar in a clinic workflow, but they are reported and reviewed very differently.
The biggest coding risk is unclear AI involvement. If a note says only “AI-assisted,” a coder cannot tell whether the output was a supporting data point, a risk score, or an independent conclusion. Provider oversight matters too. Where a physician interprets or acts on the output, that judgment should be visible in the record.
Our billing experts recommend a simple rule: the chart should answer three questions. What did the software do? What did the provider do with the output? And how did it affect patient management?
What These New CPT Codes Mean for Medical Billing Services
For a medical billing service, the new codes add review steps at almost every stage of the revenue cycle.
- Documentation review: Coders must confirm that AI use, output, and provider involvement are recorded.
- Claim submission: Claims must match the descriptor for the AI category and avoid duplicating physician work already captured in existing codes.
- Revenue cycle impact: New codes mean new fee schedule entries, payer edits, and denial patterns to monitor.
- Payer scrutiny: Early adoption periods usually bring inconsistent payer rules and heavier reviews.
Documentation Elements Coders Should Verify
| Element | What Our Coders Look For |
|---|---|
| Provider involvement | Who reviewed, interpreted, or acted on the output |
| AI tool usage | Name of the tool and what it produced |
| Clinical decision-making | How the output influenced diagnosis or management |
| Medical necessity | Why the AI-supported service was needed |
| Audit-ready records | Consistent, dated, retrievable documentation |
Compliance Challenges Practices Should Prepare For
New codes bring new exposure. Payer policies will keep evolving, so a service paid by one plan may be denied by another. Audit risk rises when documentation does not separate what software generated from what the provider decided. That distinction is central under the taxonomy, because the category depends on the role of the output.
Data governance matters as well. AI tools handle patient information, so practices must maintain HIPAA safeguards and vendor agreements that reflect how data is used. The HHS HIPAA resources are a useful official starting point. Our compliance-focused coders flag documentation gaps early and help clients build notes that hold up in an audit.
Opportunities Created by the New AI Healthcare CPT Codes
The new codes are not only a compliance burden. They create real advantages for prepared practices:
- Greater clarity: A defined code replaces guesswork about how to report AI-supported work.
- Better tracking: Reimbursement for AI services can finally be measured and compared.
- Recognition of new care models: Practices investing in technology can report that work properly.
- Measurement of outcomes: Standardized data helps show what AI-supported care delivers.
- Room to grow: With the AI code family now at 43 codes, further expansion looks likely.
How Healthcare Organizations Can Prepare for CPT 2027 AI Changes
- Review existing AI tools. Make an inventory of every AI-enabled tool in use, from imaging software to risk scoring, and note what each one outputs.
- Update coding policies. Add AI classification and reporting rules to your internal coding guidelines.
- Train providers and coders. Everyone should understand the assistive, augmentative, and autonomous categories and how they affect documentation.
- Strengthen documentation workflows. Build EHR prompts that capture tool name, output, and provider review consistently.
- Work with experienced medical billing services. This is where A3 Medical Billing comes in. Our team maps your AI tools to the right categories, updates charge masters, tests claims, and audits results after go-live. Explore our medical billing services or ask for a CPT 2027 readiness review.
AI is only one part of this year’s release. Our coders are also preparing clients for new maternity billing codes, ventricular assist device codes, urology prostate biopsy updates, radiology changes, and sleep study codes. For a full timeline, read CPT 2027: What Independent Practices Need to Do Before January 1.
The Future of AI Coding Beyond CPT 2027
Growth in AI codes is likely to continue. The AMA revised its AI taxonomy this year so it can keep pace with changing technology, which suggests more applications and more codes will follow. Reimbursement is a separate question. Coverage and payment decisions still sit with Medicare, Medicaid, and commercial payers, and those rules can move more slowly than coding. Practices should watch CMS and payer bulletins alongside the AMA. Staying current is how practices avoid missed revenue and audit surprises.
Conclusion
The 10 new AI-related CPT codes for 2027 confirm that AI is now a recognized part of patient care and billing. Success depends on three things: knowing whether a tool is assistive, augmentative, or autonomous; documenting the provider’s role clearly; and billing in line with each player’s rules. Practices that prepare early will avoid rework and denials in the new year.
FAQs
They are new codes added in the CPT 2027 code set to describe AI-related services that support or assist healthcare professionals. They bring the total number of AI-related CPT codes to 43 and take effect January 1, 2027.
It is the AMA’s method, found in CPT Appendix S, for sorting AI-enabled services into assistive, augmentative, or autonomous categories based on the software’s output and its role in care.
Augmentative software produces a score, index, or category that differs from its input and contributes to patient management. It does not provide an independent interpretation, and the related physician work is typically captured by existing codes.
Autonomous software derives parameters and generates clinically meaningful interpretations on its own, without concurrent physician involvement. It has three levels, from recommending a diagnosis (Level I) to automatically initiating management (Level III).
They add documentation, classification, and payer-review steps. Billing teams must confirm the AI category, the provider’s role, and medical necessity before submitting claims.
They give AI-supported services a standard way to be reported, tracked, and reviewed, which improves clarity and reduces guesswork. They also raise audit expectations, so accurate coding matters more.