Source: Nong P, Neprash HT. Unintended Consequences of Using Ambient Artificial Intelligence Scribes for Billing. JAMA Health Forum. 2026;7(1):e255771.
The Numbers You Need to Know
Ambient AI scribes (the tools that listen to your doctor’s visit and write the note automatically) cost $300 to $500 per month per clinician. Nearly one-third of US hospitals adopted generative AI integrated with their electronic health records in 2024. That adoption is accelerating.
The promise: less time typing, more time looking at you. The problem: vendors are now selling these tools not just for documentation but for billing. And that changes the math for patients.
A $500/month scribe pays for itself with roughly 4 additional level-4 Medicare office visits per month. That is a low bar. Once the tool is revenue-positive, the incentive shifts from “help the doctor” to “bill more.”
Medicare Advantage plans already receive a mandatory 5.9% downward payment adjustment from CMS because upcoding in that system is so well-documented. AI scribes threaten to accelerate the same pattern across all payers.
What This Means (The Evidence)
This JAMA Health Forum viewpoint identifies three ways AI scribes increase spending when used for billing:
Upcoding visits. The scribe can convert a preventive visit to a problem-based visit, or code a level 5 evaluation and management service instead of a level 3, based on what the patient mentions during conversation. In ObGyn, think about what happens during a routine annual well-woman exam. A patient mentions irregular periods, some pelvic pain, a question about contraception. An AI scribe optimizing for billing could reclassify that straightforward preventive visit as a complex problem-based encounter. Same conversation. Higher bill. No additional care delivered.
Maximizing diagnoses. The scribe can tag every mentioned symptom as a billable diagnosis, inflating the severity profile. In prenatal care, where patients are seen frequently and discuss multiple symptoms at every visit (nausea, back pain, fatigue, anxiety), the opportunity for diagnosis stacking is enormous. Each added diagnosis can increase payments, especially in value-based contracts and Medicare Advantage.
Identifying “care gaps.” The scribe flags overdue screenings (mammograms, colonoscopies, Pap smears) that generate revenue if performed in-house. This could be genuinely helpful, or it could push unnecessary testing to generate revenue.
What Gets Lost
The authors make a point that matters for every ObGyn patient: the time AI scribes save on documentation may not go back to you. Organizations may respond to the cost of the technology by increasing visit volume expectations, meaning your doctor sees more patients per day, not fewer. The after-hours time that was supposed to be reclaimed gets absorbed into longer clinic days.
There is also a professional autonomy problem. Clinicians must attest that billed codes are accurate. When an AI suggests more intensive codes than the clinician believes are appropriate, the doctor is caught between the algorithm and their own judgment. In ObGyn, where visit complexity varies widely (a straightforward OB check vs. counseling about preeclampsia risk), AI-generated codes that default to maximum intensity put clinicians in a difficult position.
The trust issue is real. Patients who come in for a routine prenatal visit and later receive a bill reflecting a complex medical encounter will notice. Surprise billing in obstetrics is already a significant source of patient frustration. AI-driven upcoding could make it worse.
The Bottom Line
AI scribes were built to solve a real problem: doctors spending more time on computers than on patients. That problem is especially acute in ObGyn, where the volume of prenatal visits, the emotional weight of the encounters, and the complexity of documentation all collide. Anything that gives your doctor more time to actually talk to you is worth pursuing.
But once these tools are optimized for billing rather than documentation, the incentives flip. The patient becomes a revenue opportunity. The visit becomes a coding exercise. The technology designed to reduce burnout may instead create new administrative burdens as clinicians spend time reviewing and correcting AI-generated billing recommendations.
Three policy actions proposed: a national registry of AI use in healthcare organizations, stronger Medicare claims auditing with clinician-in-the-loop requirements, and automatic downward reimbursement adjustments for services susceptible to AI-driven upcoding.
For patients: if your bill from an ObGyn visit looks more complex than the visit felt, ask questions. You have the right to understand every code on your bill and to challenge codes that do not reflect the care you received.
Reference
Nong P, Neprash HT. Unintended Consequences of Using Ambient Artificial Intelligence Scribes for Billing. JAMA Health Forum. 2026;7(1):e255771. doi:10.1001/jamahealthforum.2025.5771