Audiology has no FDA-cleared autonomous AI diagnostic tool, and no CPT code for one either

A patient mentions the new Galaxy Buds hearing test she saw advertised, five minutes, no appointment needed, and asks whether your practice offers the same thing with AI. A vendor rep at a trade show tells you their platform diagnoses sensorineural hearing loss straight from a tablet screening. Both claims borrow the language of a cleared diagnostic device. Neither one describes what optometry already has in CPT 92229: an FDA-cleared AI system that issues a screening result on its own, backed by a Medicare billing code built specifically for that use. Audiology, as of this writing, has no equivalent to either half of that pairing, and knowing why matters more than any single vendor's marketing copy.

What "autonomous" means, and the one specialty that already has it

An autonomous AI diagnostic tool issues its own screening result without a clinician independently interpreting the raw data first. Optometry already has a cleared, billable example of that: CPT code 92229 pays a national Medicare rate of $46.76 in 2026 for point-of-care autonomous AI retinal analysis, and three systems, LumineticsCore, EyeArt, and AEYE-DS, hold FDA clearance to issue that result for diabetic retinopathy screening specifically (AMA CPT code set and CMS Physician Fee Schedule, 2026). An assistive tool does something narrower: it scores, flags, or measures a data point, and a clinician still makes the diagnostic call every time. Nearly everything currently marketed as AI in hearing care falls into the second category, and the gap between the two is exactly what a vendor demo is least likely to spell out.

So what this means for you: before any AI diagnostic claim registers as a real capability, ask which category it belongs to. The two are regulated, billed, and trusted on completely different terms, and only one of them currently exists in audiology's own billing code set.

Audiology's twelve new 2026 CPT codes don't cover diagnosis

Audiology got its own overhaul in the same CPT 2026 codebook that created 92229's cousins elsewhere in medicine. Twelve new codes, 92628 through 92642, replaced the old 92590 to 92595 hearing aid service codes effective January 1, 2026, covering candidacy determination, hearing aid selection, fitting, post-fitting follow-up, verification, and assistive-device fitting, all billed on a time basis (American Academy of Audiology and American Speech-Language-Hearing Association coding guidance, published September 2025). Every one of them describes a hearing aid service, not a diagnostic test. In the same 2026 release, the American Medical Association added dedicated Category I codes for augmented intelligence services elsewhere in medicine, including software that analyzes CT angiography for coronary plaque and software that classifies burn wound healing from multispectral imaging (American Medical Association, CPT 2026 code set announcement, 2025). Audiology's new codes sit in that same book. None of them are AI codes.

So what this means for you: if a vendor tells you their hearing-loss AI bills like 92229, ask them to name the code. Audiology's 2026 update did not create one, and nothing else currently fills that gap.

No FDA-cleared autonomous hearing-loss AI exists, as of this writing

A direct search for an FDA-cleared autonomous AI system for diagnosing hearing loss, the audiology equivalent of LumineticsCore or EyeArt, turns up nothing as of September 2026. That is a genuine gap, not a search limitation: the FDA's own device clearance record spans cardiology, radiology, ophthalmology, and dermatology, but carries nothing in audiology with an autonomous diagnostic clearance the way diabetic retinopathy screening has in eye care. If a specific product exists that this check missed, the fix is the same either way: ask for the clearance number and check it against the FDA's own database rather than taking a sales claim at face value.

So what this means for you: the burden of proof sits with the vendor, not with your skepticism. A confident claim on a slide deck is not a substitute for a clearance number you can verify yourself in minutes.

Three things are real. None of them are autonomous diagnosis.

What exists today in hearing care AI splits into three categories, and confusing them is where most vendor pitches get their power. The first is signal-processing AI built into hearing aids themselves. Starkey's Edge AI uses a neural processing unit to classify soundscapes and reduce noise in real time, and Phonak's Sphere Infinio and ReSound's Vivia are, per one 2026 industry review, the only two products currently running real-time AI processing rather than pre-set algorithms (Soundly, "Best AI Hearing Aids of 2026," 2026, an industry review compilation rather than an independent clinical source). None of this diagnoses anything. It processes sound for someone who already has a hearing aid.

The second category is administrative AI, already covered on this site: Sycle's AI suite summarizes and drafts clinical notes from dictation, with an audio-to-note transcription feature planned for the second quarter of 2026 (Sycle company product announcements, 2026). That is workflow software, not a diagnostic tool, and it sits alongside the recall and scheduling automation this site profiled in our piece on audiology's real AI opportunity.

The third is consumer self-testing. Samsung's Galaxy Buds Hearing Test received FDA clearance in August 2026 as part of a self-fitting, over-the-counter hearing aid pathway for adults with perceived mild-to-moderate hearing loss, registered as Software as a Medical Device and built on a five-minute, self-administered pure-tone test (Samsung Newsroom, August 2026). Samsung's own materials describe the results as comparable in accuracy to a professional audiologist's evaluation, a claim worth flagging as the manufacturer's own characterization rather than an independently confirmed finding. More importantly for your practice, it is cleared under the OTC hearing aid pathway for consumer use at home, not as a clinical diagnostic tool an audiology practice can adopt, bill for, or rely on in place of its own exam.

So what this means for you: when a patient or a vendor describes any of these three as AI diagnosis, the accurate answer is that it's real technology doing something else: sound processing, note-taking, or a consumer screening test, not a cleared autonomous diagnostic decision.

The research pipeline is active. Nothing in it has reached your practice yet.

The underlying science is moving faster than the regulatory record suggests. A 2024 study in Scientific Reports trained a model to classify pure-tone audiometry results into four categories, normal, conductive, mixed, and sensorineural hearing loss, with 99.33% accuracy on data the model had not seen during training ("Automated hearing loss type classification based on pure tone audiometry data," Scientific Reports, 2024). A clinical trial registered on ClinicalTrials.gov, studying AI-assisted diagnosis, triage, and assessment of hearing loss and tinnitus, lists an estimated start date of May 30, 2026, and had not begun recruiting as of the most recent check this week (NCT07586098, ClinicalTrials.gov, checked September 2026). Both are genuine signs of where the field is heading. Neither is a cleared product, a CPT code, or something you can put in front of a patient today.

So what this means for you: a compelling research paper or a registered trial is evidence the category is coming, not evidence it has arrived. "In development" and "FDA cleared" are two different claims, and only one of them should change how you evaluate a purchase.

The call

If you're being pitched software that claims to diagnose hearing loss autonomously, ask three questions before the demo goes any further: what is the specific FDA clearance or De Novo number, is the clearance for autonomous use or assistive decision support, and what CPT code, if any, bills for it. As of this writing, the honest answer to that third question in audiology is none, because the billing infrastructure optometry built around CPT 92229 does not yet exist here. That doesn't make every AI tool in hearing care worthless. The signal-processing, documentation, and consumer-screening categories above are all real and improving. It does mean that anything sold to you as autonomous diagnostic AI in audiology is, at minimum, ahead of the FDA and ahead of Medicare, and should be evaluated on that basis rather than the language around it.

For a scored, independent read on which AI claims in your own vendor pipeline are backed by a real clearance and which are ahead of one, the AI Opportunity and Growth Assessment is built to separate the two before you sign anything. Start with a free 20-minute discovery call.

See also: our look at the one FDA-cleared autonomous AI test optometry actually has, and why audiology's real AI advantage already spans 6,000 clinics in the admin layer, not the exam room.

Related: 92229: optometry's only autonomous AI code · Audiology AI already spans 6,000 clinics · 6 codes became 12: audiology's real AI opportunity