This analysis synthesizes sources published the week ending September 14, 2026. Editorial analysis by the PhysEmp Editorial Team.
A wave of legislative activity across state capitols and Congress is moving to codify what many physicians have long assumed but never had guaranteed: clinical decisions must remain under human physician control. The Doctors Not AI Act, introduced with bipartisan support, would prohibit AI systems from making final determinations on diagnosis, treatment, or coverage without physician oversight. Simultaneously, state lawmakers in Wisconsin, Florida, and California are holding hearings on regulatory frameworks that would require human sign-off on AI-generated clinical recommendations. For physicians facing an uncertain labor market, these developments represent a potential structural floor under their professional authority—and, by extension, their bargaining power. This emerging regulatory landscape directly intersects with Physician Compensation & Demand dynamics and could reshape contract negotiations and hiring competition for years.
The Regulatory Mechanism Taking Shape
The federal Doctors Not AI Act, backed by Representative Buddy Carter and gaining traction in both chambers, would establish that AI tools in healthcare cannot independently authorize treatments, issue diagnoses, or determine insurance coverage. Medicare reimbursement discussions have begun exploring how to pay for AI-assisted care, but federal officials are also signaling that payment models will require documented physician involvement. The New York Times reports that FDA officials are working with CMS to ensure any AI billing codes maintain physician accountability as a prerequisite for reimbursement.
At the state level, Wisconsin’s bipartisan study committee is drafting recommendations that would require healthcare facilities to disclose AI use and mandate physician review of AI outputs before clinical action. Florida’s legislature is considering bills that would classify certain autonomous AI diagnostic functions as practicing medicine without a license—effectively making physician oversight a legal requirement rather than an institutional choice.
When legislation defines AI as a tool that requires physician authorization rather than a replacement for physician judgment, it turns a substitution threat into a technology that can increase demand for credentialed oversight.
Employment Protections Through Regulatory Design
The compensation fallout from these bills is more than symbolic. If AI systems cannot legally operate without physician sign-off, health systems cannot simply cut physician headcount by deploying autonomous software—they can only shift what physicians spend their time doing. That creates a structural floor under physician demand that pure market forces would not guarantee.
The AMA’s CEO, quoted in STAT, emphasized that AI should “augment, not replace” physician decision-making. The association’s lobbying suggests they know this outcome will likely require regulation rather than voluntary restraint. Health systems under margin pressure have incentives to push autonomy as far as law allows; tighter rules narrow that option and protect physician roles.
For specialties most exposed to AI diagnostics—radiology, pathology, and primary care triage—these rules might matter more than any single contract clause. A radiologist’s bargaining power in 2028 compensation talks will depend heavily on whether their interpretive authority is legally mandated or simply preferred by their current employer.
The Medicare Reimbursement Question
Federal officials weighing Medicare payment for AI-assisted care face a clear tension: how to pay for efficiency gains while keeping physician accountability in the loop. Becker’s Hospital Review reports that CMS is exploring billing modifiers that would require documented physician review for AI-flagged findings to qualify for reimbursement. That would make physician involvement a financial necessity for health systems seeking to monetize AI investments.
The practical result could be that AI adds reimbursable work for physicians rather than substituting for it. A primary care doctor who reviews an AI-generated risk assessment may bill for that cognitive work; an autonomous AI system generating the same assessment cannot. If payment policy goes this way, it will push systems to build more AI-assisted encounters that still require—and can bill for—physician oversight.
Health systems investing heavily in AI diagnostic tools may find their ROI calculations depend less on reducing physician FTEs and more on increasing the volume of AI-assisted encounters that still require—and can bill for—physician oversight.
Contract Implications Already Emerging
Physicians signing contracts in 2026 and 2027 should expect AI governance to be a legitimate negotiation point. California providers interviewed by Capitol Weekly said they worried about being held responsible for AI-generated recommendations they didn’t have time to review. That liability can become bargaining material: if physicians are accountable for AI outputs, they can demand protected review time, limits on AI-assigned caseloads, or extra pay for oversight duties.
Recruiters and executives face the inverse problem. Job descriptions that assume AI will reduce clinician workload may look legally shaky if oversight rules require additional review time. Productivity models that bank on AI-driven headcount cuts may need rewrites if those “savings” come with new physician hours attached. Systems that factor regulatory constraints into 2027 hiring plans will avoid expensive mid-contract fixes.
State-by-state variation matters, too. Physicians considering moves to states with stronger AI oversight might find more structural protections than in places with looser rules. That regulatory arbitrage could shape where specialties cluster and widen existing geographic pay gaps.
The Unresolved Tension
One core question remains unsettled: what does meaningful “physician oversight” look like in practice? A radiologist who skims 500 AI-flagged images per shift provides very different oversight than one who carefully reviews 50. If laws require sign-off without setting review standards, systems may comply in form while still squeezing productivity.
Over the next year we will see whether physician groups can translate legislative momentum into operational rules—minimum review times, maximum AI-assisted caseloads, or documentation standards that actually limit workload. If those specifics never arrive, the regulatory floor will be thin and compliance can be mostly paper. Contracts signed now should try to lock in AI workload protections.
Expect fights about what counts as “review.” Expect long addenda and awkward audits. Expect an overnight hospital radiologist, eyes burning, clicking through hundreds of flagged scans because regulations say she signed off.
Sources
Wisconsin lawmakers weigh AI regulation in healthcare – WMTV (NBC15)
Lawmakers study AI and its growth in the healthcare sector to help craft the regulatory framework – CBS58
Florida targets AI doctors as local physicians urge cautious adoption – WEAR-TV
Buddy Carter backs bill limiting AI use in healthcare decisions – WSAV
Congressman Buddy Carter backs Doctors Not AI Act – WSAV
AI in medicine won’t replace doctors AMA CEO says – STAT
Federal officials weigh how Medicare could pay for AI doctors – Becker’s Hospital Review
How AI Is Changing Health Care from Diagnosis to Discharge – The New York Times
Growing use of AI tools has some medical providers on edge – Capitol Weekly
State lawmakers discuss AI in Wisconsin healthcare – WI Proud