Telemedicine AI and the Scope-of-Practice Question Emergency Medicine Has Been Avoiding

By Chester "Chet" Shermer, MD, FACEP | Published June 21, 2026 | 10 min read

The advanced practice provider on the telemedicine platform at 0300 is taking a call from a small-town emergency department where the on-shift clinician needs guidance on a patient with new neurologic findings. The platform’s AI clinical decision support layer has already surfaced a differential, weighted the probabilities, suggested the imaging workup, and flagged the medication-interaction risk on the patient’s existing regimen. The APP reviews the AI output, talks to the bedside clinician, recommends the disposition, and signs off. The encounter takes eleven minutes. The patient is admitted to the receiving hospital seventy-five minutes later with a confirmed posterior-circulation stroke. The outcome is good. The institutional question is whether the encounter that produced the good outcome was practiced within the APP’s scope, the supervising physician’s scope, the platform’s scope, or some emergent scope that the AI layer created and that nobody at the institution has formally defined.

That question is the scope-of-practice question that emergency medicine has been avoiding since telemedicine AI started absorbing the routine cognitive workload of remote consultation in 2024. The technology has been moving faster than the institutional governance. In 2026, the regulatory environment finally started catching up, and the institutions that have not been paying attention are about to find out that the governance question is going to be answered by somebody. The choice is whether the institution writes the answer or inherits it.

What Changed in 2026

Three regulatory developments this spring reset the conversation.

The first is the Federation of State Medical Boards’ June 5, 2026 communication on the new AI and cybersecurity executive order, which formalizes the state medical boards’ role in physician oversight of AI-enabled clinical tools. FSMB has historically been the coordinating body for state-level licensure and discipline, and its 2026 posture treats AI-enabled clinical decision support as a category of practice that state boards have an obligation to address through licensure, credentialing, and disciplinary frameworks. The implication for telemedicine is direct. The physician who supervises an AI-augmented remote consultation across state lines is now operating under a regulatory regime in which the state board on each end of the consultation has a defined interest in how the AI layer affected the clinical decision.

The second is the American Medical Association’s June 10, 2026 policy statement on AI transparency and oversight, which establishes the AMA’s institutional position that AI-enabled clinical tools require physician oversight, patient transparency, and outcome accountability as conditions of safe deployment. The policy is not regulation. It is, however, the position the state medical boards will reference when they evaluate complaints, the position the malpractice carriers will reference when they price coverage, and the position the health systems will reference when they write their internal governance frameworks. The AMA’s framework is the operational template the institution has to translate into local policy.

The third is the JD Supra May 26, 2026 analysis of state AI regulation, which documents the growing patchwork of state-level AI healthcare regulation that telemedicine platforms now have to navigate. California, Texas, New York, and a dozen other states have introduced or passed legislation in 2026 that imposes specific disclosure, oversight, or scope-of-practice constraints on AI-enabled clinical tools delivered across state lines. The telemedicine platform that operates in twenty states is now operating under twenty regulatory regimes, and the scope-of-practice constraints in those regimes are not consistent. The institution that uses the platform inherits the regulatory complexity whether it understands the complexity or not.

Why This Is Specifically a Scope-of-Practice Question

The conventional framing treats telemedicine AI as a clinical decision support tool that the licensed clinician evaluates and either accepts or overrides. That framing is correct as far as it goes, but it misses the institutional reality. The AI layer in a modern telemedicine platform is doing meaningful cognitive work — differential generation, evidence weighting, dose calculation, interaction checking, imaging-protocol suggestion, disposition recommendation — that historically required physician training to perform. When the AI does that work and the APP signs off, the APP is operating with a cognitive scaffold the APP’s training did not include. When the supervising physician reviews the APP’s encounter through the platform’s audit trail, the physician is reviewing a decision the AI shaped and the APP executed.

The de facto scope of the encounter is not the scope of the APP’s license. It is not the scope of the supervising physician’s license either. It is some emergent scope that the AI layer has created and that the institutional governance has not formally defined. The licensure frameworks the state boards inherited from the pre-AI era do not have a category for this. The 2026 FSMB posture is the first formal acknowledgment that the boards intend to develop one.

The Medium piece I wrote earlier this year on prehospital decision support made the same argument for the EMS scope-of-practice context: the AI layer is quietly expanding the operational capability of the prehospital provider in ways the protocol and the medical-director oversight structure have not formally addressed. The telemedicine version of that argument is harder because the institutional accountability is more dispersed. The platform vendor, the consulting physician, the bedside clinician, the receiving facility, and the patient’s home-state regulator are all in the chain. The scope-of-practice question is the question of who is responsible for what part of the encounter that the AI shaped.

What the Institution Has to Build

Four pieces.

First, an inventory. The institution that uses telemedicine AI has to know what AI is in the platform, what version is currently deployed, what categories of clinical decision the AI participates in, and what evidence supports the AI’s performance for each category. The vendor’s marketing material is not the inventory. The inventory is the institutional record that the medical executive committee, the credentialing committee, and the malpractice carrier can reference. Without the inventory, the scope-of-practice conversation cannot start.

Second, a scope-and-oversight policy. The institution has to define which categories of decision the AI is permitted to participate in, which credential tier of clinician is required for which category, what the supervising-physician review obligation looks like, and how the audit trail captures the AI’s contribution to the decision. The policy is the institutional answer to the question the state board will eventually ask. The institution that has the policy has a defensible position. The institution that does not has a discoverable gap.

Third, training. The clinician who uses the AI-augmented platform needs explicit training on the AI’s strengths, weaknesses, override patterns, and failure modes. That training has to be specific to the platform, specific to the clinician’s role, and refreshed when the platform updates. The training infrastructure is the institutional commitment that converts the policy from paper into practice. For the simulation infrastructure that supports clinician-level competency with AI-augmented telemedicine scenarios, see emsmedsim.globalmedopscommand.com.

Fourth, the audit loop. The institution has to review the cases the AI shaped, identify the patterns of agreement and disagreement between the AI and the clinician, document the override decisions, and feed the findings back into the scope-and-oversight policy. The audit loop is what tells the institution whether the policy is producing the outcomes the policy intended. Without the loop, the policy is a static document that does not respond to the operational reality. With the loop, the policy is a living governance artifact.

Dr. Chet’s Take

The scope-of-practice question has been the unanswered question at the center of telemedicine AI since the technology started absorbing meaningful clinical cognition. The 2026 regulatory environment is now making the question impossible to defer. The state medical boards, the AMA, the state legislatures, and the malpractice carriers are all positioning to answer it. The institutions that have built the inventory, the policy, the training, and the audit loop will be ready to defend their answer. The institutions that have not will be answering questions they cannot answer well, in front of regulators and litigators who are operating from frameworks the institution did not help build.

The physician leadership task is to write the institutional answer now, while the regulatory environment is still in formation. The answer does not have to be perfect. It has to be defensible, documented, and adaptive. The five-year window in which the emergency medicine specialty can shape the scope-of-practice framework for telemedicine AI is open in 2026, and the leadership that engages the conversation now will be the leadership that determines what the framework looks like in 2031. The leadership that defers will inherit a framework written by people who do not understand what emergency physicians actually do.

— Chester "Chet" Shermer, MD, FACEP | Emergency Medicine, 25+ Years Clinical Experience

AI Won’t Wait. Neither Should You.

If you direct an emergency medicine group, lead a telemedicine network, or serve on a medical executive committee that has not formalized its position on AI-augmented remote consultation, the regulatory clock has started. Consider enrolling in my course: AI in Emergency Medicine: Becoming AI Bulletproof. The course walks through the inventory, scope-and-oversight policy, training architecture, and audit-loop structure that the 2026 regulatory environment now requires of any institution deploying AI-augmented telemedicine at scale.

If you’re an emergency physician (or any clinician treating patients daily) trying to understand how AI will actually impact your clinical practice — not just the hype — I put together a free practical guide. You can download it here: AI in EM Survival Guide.

Related Reading

• Read more from Dr. Shermer on Medium: medium.com/@chet.shermer
• Course catalog: courses.globalmedopscommand.com/store
• GMOC main site: globalmedopscommand.com

Sources

1. Federation of State Medical Boards, “FSMB Communication on AI/Cybersecurity Executive Order,” June 5, 2026, https://www.fsmb.org/siteassets/advocacy/news/june-5-2026.pdf
2. American Medical Association, “With AI Increasingly Part of Care, Transparency and Quality Are Musts,” June 10, 2026, https://www.ama-assn.org/practice-management/digital-health/ai-increasingly-part-care-transparency-and-quality-are-musts
3. JD Supra, “States Continue Efforts to Regulate AI in Healthcare,” May 26, 2026, https://www.jdsupra.com/legalnews/states-continue-efforts-to-regulate-ai-3082341/
4. Center for Connected Health Policy, “State Telehealth Policies for Online Prescribing,” June 15, 2026, https://www.cchpca.org/topic/online-prescribing/
5. College of Nurses of Ontario, “New Guideline Supports Nurses in Providing Virtual Care,” June 16, 2026, https://www.cno.org/news/new-guideline-supports-nurses-in-providing-virtual-care
6. Shermer, C., “Prehospital Decision Support: What EMS Crews Need and What AI Companies Are Selling,” Medium, https://medium.com/@chet.shermer/prehospital-decision-support-what-ems-crews-need-and-what-ai-companies-are-selling-6331f493e376