AI Scribe Oversight Is Shrinking as Rural Adoption Accelerates: RHTP Funds, OpenEvidence, and What CAH Leaders Need Now

KFF Health News published a detailed investigation this week into a quiet but consequential policy shift: the Trump administration and HHS Secretary Kennedy are rolling back the federal standards that have governed electronic health record usability and clinical decision support transparency. The piece centers on ambient scribe software, where clinical documentation AI is now embedded across major health systems with no federal vetting process in place.

Clinician reaction is divided. Many physicians report genuine time savings. A JAMA study cited in the piece found that the highest-use clinicians saved more than half an hour of documentation work daily after one year. But researchers and patient safety advocates note that there is no standardized safeguard to catch errors in AI-generated notes before they become part of a patient's medical record. A researcher at MedStar Health said directly that there is currently no federal safeguard in place to vet scribe software. For rural CAHs operating with fewer administrative checks than large health systems, the governance gap that deregulation widens deserves immediate attention. A facility without an internal AI review process is a facility relying entirely on vendor accuracy.

NBC News published a major feature this week on OpenEvidence, an AI-powered clinical search platform that has become the most widely adopted physician-facing AI tool in the country, now used by roughly two-thirds of U.S. doctors. The tool allows clinicians to query medical literature in real time using natural language, and returns synthesized clinical answers faster than legacy reference tools like UpToDate. Sanford Health's CMO, who oversees more than 2,500 providers across the country's largest rural healthcare system, described OpenEvidence as "remarkably easy to adopt" and noted that most physicians across their network are familiar with it.

The rural angle is significant. A physician at an Indian Health Service medical center in rural South Dakota described using OpenEvidence to quickly determine whether a CT scan was needed for a suspected spinal fracture, getting an evidence-backed answer without a specialist on site. That is a credible description of AI as a force multiplier in a resource-constrained rural environment. OpenEvidence has surged to a $12 billion valuation in little over a year and counts Sequoia, Google Ventures, and Andreessen Horowitz among its backers. UpToDate has since launched its own AI product in response, now active at roughly 2,000 hospitals and health systems.

Arkansas opened applications on May 11 for $209 million in Rural Health Transformation Program funding, with an initial $55.6 million directed to the state's THRIVE initiative: Telehealth, Health-Monitoring and Response Innovation for Vital Expansion. The program targets telehealth access, emergency medical response infrastructure, and remote patient monitoring for chronic disease management. Applications are due June 12, with funding distributed on a rolling basis. Governor Sanders indicated that the remaining balance of the $209 million annual allocation will be awarded through additional rounds over the summer and fall.

Arkansas is a high-need state by most rural health measures: 45 percent of the state's 1.3 million residents live in rural areas, with median household incomes just over $50,000 and elevated rates of chronic disease concentrated in the Ozarks, Southwest Arkansas, and the Delta. Eligible applicants include rural hospitals, FQHCs, EMS providers, rural health clinics, pharmacists, universities, and community organizations. For CAH and FQHC leaders in Arkansas, the June 12 deadline is near enough to warrant immediate review of the Notice of Funding Opportunity. The state expects its total five-year RHTP allocation to exceed $1 billion.

The AHA's Center for Health Innovation published a market scan this week profiling four health systems that have moved AI from pilot to embedded clinical workflow, with two examples directly relevant to rural and community hospital leaders. Advocate Health, which initially piloted AI imaging models at 22 sites in Wisconsin and North Carolina in late 2024, announced in mid-2025 that it would expand the technology across its clinical imaging workflows to improve detection of pulmonary embolisms and intracranial hemorrhages. Wake Forest University's enterprise radiology chair stated that the results support responsibly deployed imaging AI as a best practice regardless of facility size or geography. Emory University Hospital Midtown in Atlanta has also deployed a virtual nursing initiative using AI and LIDAR technology for patient monitoring and fall prevention, with a model now being evaluated for replication across additional sites.

For CAH and rural hospital leaders, the imaging AI data points are the most transferable near-term signal. Radiology AI for condition detection does not require an enterprise IT team to operate, and the Advocate results suggest measurable clinical impact at community-scale facilities. The market scan also noted a Microsoft and AHA webinar from May 7 focused specifically on building AI-ready rural hospital workforces, which is now available on demand through the AHA.

Organ Mountain News published a first-person account this week from Dr. Peter Jewell, a family medicine physician at Artesia General Hospital in southeastern New Mexico, describing his experience adopting Microsoft Dragon Copilot in a small community hospital setting. Jewell sees 20 to 25 patients per day and previously spent several hours after each shift completing clinical documentation. He describes himself as someone who is not comfortable with technology, and his account of the transition is practical rather than promotional: the tool records and transcribes patient conversations and updates the EHR, eliminating a significant share of after-hours charting for a physician who did not expect to find AI adoption straightforward.

Artesia General serves patients traveling long distances from across southeastern New Mexico, including from Lovington, Ruidoso, Alamogordo, and Cloudcroft. The story is valuable not because it describes a large-scale deployment with enterprise IT support, but because it does not. It describes a single physician at a small community hospital adopting a clinical AI tool and finding it workable. That specificity is more useful to most CAH and rural health clinic leaders than aggregate data from major health systems. The original reporting was produced by Searchlight New Mexico and republished by Organ Mountain News.