Dentacor: Lack of Access Is Not Consent to Reduced Dentist Supervision
Evidence-Based Responsibility Reconstruction℠ Case Study
Dentacor is a mobile dental-care provider founded in 2021. Its public website describes a conventional clinical model: hygienists collect images and information, and a Dentacor dentist creates the patient’s diagnosis and treatment plan.
Utah authorized a different model.
Under a 12-month regulatory mitigation agreement, Dentacor’s dental hygienists may diagnose specified conditions with the concurrence of an AI radiograph tool instead of working under the general supervision of a dentist. If the hygienist and AI agree, treatment can proceed without dentist review. If they disagree, the case is escalated.
The pilot is directed toward people experiencing homelessness, poverty, addiction, and limited access to Medicaid dentists. Patients must consent after being told that no dentist is supervising the procedure and that a hygienist does not have the same training or professional scope as a dentist.
The AI system replacing that supervision is never identified.
The public agreement does not name its developer, vendor, product, model, version, validation study, or party responsible for retaining its technical records.
This is the responsibility problem: Utah used an access crisis to authorize reduced dentist supervision for patients with the fewest alternatives, while leaving the technology actor outside the public record.
The AI Has No Public Identity
The Agreement Names a Function, Not a System
The agreement repeatedly refers to an “AI-assisted radiograph diagnostic tool.” It does not identify:
- The AI developer or vendor
- The product, model, or version
- Its regulatory status
- Its diagnostic thresholds
- Its validation evidence
- Who controls software updates
- Who retains the technical logs
- How responsibility is divided between Dentacor and the technology provider
Dentacor’s proposal claims that AI radiograph tools have surpassed diagnostic benchmarks associated with junior dentists. It does not identify the benchmark, cite the supporting evidence, or establish that the system used in this pilot achieved that performance.
A general claim about dental AI cannot validate an unidentified product.
The Silence Is Not an Industry Default
Dental AI companies routinely identify their systems and regulatory status.
Videa, for example, publicly identifies its dental AI products and FDA clearances and says its technology analyzes more than 500 million X-rays annually. FDA records identify the manufacturer, product, intended uses, supported image types, patient populations, and operating limitations. They also describe the system as an aid to trained professionals rather than a replacement for professional diagnosis. Videa’s FDA-clearance announcement, FDA 510(k) record
The Dentacor record provides none of that information.
Because the tool is unnamed, the public cannot determine whether it was validated or cleared for the role Utah authorized—or whether that role exceeds its intended use.
Dentacor Is Visible; the Technology Provider Is Not
Dentacor’s website presents a mobile dental-care company. Its leadership page identifies dental, provider-services, operations, and information-technology leadership. It does not identify an AI product, AI development team, clinical AI validation function, or outside technology partner.
Utah may have reviewed private information about Dentacor’s technical capability. The public responsibility record does not show it.
A healthcare provider can license a third-party AI product. But when that system helps remove a dentist from a clinical decision, its developer becomes part of the responsibility chain.
Governance surrounds the deployment. It does not visibly reach the technology.
Lack of Access Becomes the Reason to Reduce Supervision
The Same Need Justifies and Supplies the Test
Dentacor describes untreated dental disease among vulnerable Utah residents as a moral and clinical emergency. It identifies homelessness, chronic pain, addiction, poverty, and the shrinking number of dentists accepting Medicaid.
In his public announcement, Dentacor CEO Nate Wilson presented the pilot as using AI to restore access and dignity rather than optimize profits.
The need is real. The conflict lies in what that need is being used to justify.
The pilot does not add AI on top of ordinary dentist supervision. It permits AI concurrence to replace that supervision.
A patient with reliable access to a dentist can reject the arrangement and obtain care elsewhere. A patient experiencing pain, homelessness, or limited Medicaid access may effectively be choosing between AI-assisted care without dentist supervision and no treatment.
The patient’s lack of alternatives becomes:
- The problem Dentacor claims to solve.
- The reason Utah relaxes the supervision requirement.
- The condition making participation more likely.
- The source of evidence used to evaluate the model.
Lack of access is converted into consent to reduced supervision, while the patient bears the clinical consequences.
The Evidence Is Intended to Support Expansion
Dentacor estimates that the model could double the number of patients served without increasing staff. Its proposal describes the pilot as a possible national blueprint for shelters, rural communities, tribal communities, and other care deserts.
The patients receive access. Dentacor receives greater capacity and evidence supporting expansion.
Before that model can be treated as a blueprint, the record must establish which AI produced the results, whether the system changed during the pilot, how often dentists reviewed its decisions, and whether agreement between the AI and hygienists was independently verified.
Dual Verification Does Not Prove Independent Judgment
The Control Detects Disagreement, Not Shared Error
The agreement does not say whether the hygienist records a diagnosis before seeing the AI output or whether the AI can influence the hygienist’s conclusion.
It also does not establish whether:
- The AI and hygienist can share the same image-quality failure
- Confidence scores are preserved
- Agreement is later reviewed by a dentist
- False agreement is measured
- The system distinguishes uncertainty from a negative finding
If the AI and hygienist disagree, the safeguard works as described: a dentist enters the case.
If they agree incorrectly, no escalation occurs.
Both could misread the same radiograph, miss the same condition, or reach agreement because the AI influenced the human reviewer. Counting agreement would not distinguish independent confirmation from shared error or automation bias.
“Dual verification” names the safeguard. It does not prove that two independent judgments occurred.
Dentacor Controls the Initial Evidence
Compliance Exists Around the Pilot
Dentacor must use licensed hygienists, obtain patient consent, train participating staff, maintain privacy and cybersecurity protections, monitor the pilot, and report monthly to Utah.
The reports cover patient volume, demographics, AI efficacy, escalations, complaints, system errors, and adverse outcomes. Third-party researchers approved by Utah may review them, but independent review is not required.
Dentacor therefore operates the pilot, collects the evidence, evaluates the AI’s performance, and supplies the initial account to Utah.
Utah Has Announced Success Without Publishing the Evidence
In its 2025 Year in Review, the Utah Department of Commerce said early Dentacor data indicated that the care was of exceptionally high quality.
The public statement does not provide the underlying data, accuracy measure, disagreement rate, dentist-review rate, false-negative rate, adverse outcomes, or identity of an independent evaluator.
A favorable conclusion has entered the public record without the evidence needed to reconstruct how it was reached.
The agreement also says Utah may request records after a complaint or an action reasonably likely to constitute an incident “as described in Schedule A.” Schedule A does not appear to define an incident.
If a patient does not know that a diagnosis was wrong, no complaint may be filed. If no one recognizes the event as an incident, Dentacor’s internal record may remain the only account.
The unnamed AI provider may separately control the model version, confidence score, inference record, update history, and evidence of similar failures. The agreement does not establish whether Dentacor can compel that provider to preserve or produce those records.
If a Patient Is Harmed, Responsibility Is Already Split
The Responsibility Chain
Utah
Utah authorized the reduced-supervision model, approved the testing structure, and can terminate the mitigation. The agreement says the pilot is not a state endorsement.
Dentacor
Dentacor selected or integrated the AI, employs the hygienists, operates the pilot, monitors performance, and creates the reports.
The agreement explicitly preserves the patient’s right to seek available remedies from Dentacor if harm results from the AI tool. Dentacor must also protect Utah’s Office of Artificial Intelligence Policy and Division of Professional Licensing from claims and losses arising from the pilot.
The agreement therefore places the visible financial responsibility on Dentacor and away from Utah while leaving the unnamed AI provider outside that allocation.
The Hygienist
The hygienist performs the clinical assessment, confirms the diagnosis, and provides the treatment. The hygienist is named, licensed, and professionally visible.
The AI Provider
The AI provider may have developed, validated, updated, and logged the system whose concurrence allowed treatment to proceed. That organization does not appear in the public agreement.
The Patient
The patient experiences the clinical outcome without having selected the system, approved the supervision change, or evaluated the safeguards.
Calling a later failure an AI error, hygienist error, poor radiograph, or missed escalation would not establish who chose the system, validated it, controlled its updates, monitored shared errors, or accepted the remaining risk.
What the Public Record Establishes vs. What Remains Unresolved
| What the Public Record Establishes | What Remains Unresolved |
|---|---|
| Authorization: Utah authorized Dentacor’s pilot and reduced the dentist-supervision requirement for specified diagnoses and procedures. | Technology identity: The AI developer, product, model, and version are not named. |
| AI role: AI concurrence can allow diagnosis and treatment to proceed without dentist review. | Regulatory status: The record does not show whether the tool was cleared or validated for the role Utah authorized. |
| Human role: A licensed hygienist performs the assessment, confirms the diagnosis, and provides the treatment. | Independent judgment: The record does not show whether the hygienist reaches a conclusion before seeing the AI output. |
| Escalation: Disagreement or uncertainty sends the case to a licensed dentist. | Shared error: The record does not show how agreement between the hygienist and AI is checked for accuracy. |
| Consent: Patients must consent after receiving disclosures about the absence of dentist supervision and the hygienist’s training gap. | Meaningful choice: The record does not establish what alternatives are realistically available to patients unable to obtain conventional dental care. |
| Operation: Dentacor operates, monitors, and maintains the pilot. | Technical control: The record does not identify who controls model updates, thresholds, validation, or technical logs. |
| Reporting: Dentacor submits monthly reports covering efficacy, escalations, complaints, errors, and adverse outcomes. | Independent evidence: Independent review is not required, and the underlying performance data have not been published. |
| Financial responsibility: Patients retain available remedies against Dentacor, and Dentacor protects Utah from claims arising from the pilot. | Technology-provider responsibility: The unnamed AI provider does not appear in the public allocation of responsibility. |
| Public conclusion: Utah has described the pilot’s early care as exceptionally high quality. | Supporting proof: The public record does not provide the measures or evidence supporting that conclusion. |
| Intended expansion: Dentacor presents the model as a possible blueprint for other underserved communities. | Risk acceptance: The record does not identify who determined that the remaining risk to patients was acceptable. |
The institutional authorization is traceable.
The technology responsibility is not.
Responsibility Reconstruction Finding
Utah authorized reduced dentist supervision.
Dentacor designed and operates the workflow.
A licensed hygienist supplies the visible human judgment.
An unnamed AI provides the concurrence required for treatment to proceed.
A dentist enters only when the hygienist and AI disagree or identify uncertainty.
Dentacor creates the initial performance record. Utah receives that account and has already described the early care as exceptionally high quality.
Patients with limited alternatives receive the treatment and bear the consequences.
Dentacor’s mission may be sincere, and the access crisis is real. Neither fact completes the responsibility record.
A moral purpose does not identify the AI.
A consent form does not validate the system.
A disagreement process does not detect shared error.
A monthly report does not create independent evidence.
Utah authorized an unnamed AI to stand in for dentist supervision for patients least able to obtain a dentist. The patients are named as the beneficiaries, but the technology actor remains missing from the responsibility record.
Lack of access is not consent to reduced supervision.
If a patient is harmed, naming the event an AI error, a hygienist error, or an unexpected outcome will not establish who authorized, validated, controlled, monitored, and permitted the conditions that produced it.
Naming the error is not owning responsibility.