As health care systems across the United States grapple with rapid technological advancements and evolving patient needs, a wave of new regulations, professional standards, and federal initiatives is attempting to redefine accountability. From state-level restrictions on artificial intelligence in clinical settings to new Department of Health and Human Services (HHS) programs targeting vaccine injury research, policymakers, medical boards, and academic institutions are confronting critical questions about human oversight, transparency, and public trust. These developments arrive against a backdrop of shifting public sentiment, as recent polling indicates that while patients remain open to modern innovations like AI, they continue to place a high value on human judgment and rigorous verification.
State Laws and Professional Guidance Address Human Oversight of AI in Health Care
The integration of artificial intelligence into medicine has generated widespread excitement alongside profound regulatory challenges. Across the country, state lawmakers and professional organizations are actively developing new approaches to ensure responsible AI use, heavily emphasizing human oversight and accountability to protect patients.
According to a comprehensive report released by the Transparency Coalition, an AI-policy advocacy organization, state legislatures passed 115 AI-related measures this year across 30 states. By mid-October, 108 of these measures had been signed into law, while seven awaited gubernatorial approval. The report highlighted human oversight of health care AI and mandatory AI disclosure requirements as central priorities for state lawmakers.
For instance, eight states enacted laws specifically designed to limit health insurers’ reliance entirely on automated systems or AI when denying or reducing health insurance coverage, including critical prior authorization decisions. Furthermore, at least five states enacted legislation restricting the deployment of AI in mental health care, generally mandating that therapy must be administered directly by a licensed human professional. In California, a newly enacted law requires health facilities to guarantee that clinicians retain independent judgment when utilizing clinical AI tools. Most of these mental health regulations are already in effect, while the laws in Rhode Island and California are scheduled to take effect in January 2027.
These state-level efforts intersect with a broader, ongoing national debate over federal policy. The Trump administration has previously urged Congress to establish a unified national framework for AI regulation that would preempt existing state laws deemed unduly burdensome. However, a broad preemption appears less likely following the results of the 2026 midterm elections, which secured Democratic control of Congress starting in January. Many congressional Democrats have historically opposed overriding state-level AI safeguards without comprehensive federal rules in place. Concurrently, incoming congressional Democrats have indicated plans to introduce fresh legislation requiring experts to evaluate AI safety and clarifying corporate liability for harmful outputs. KFF continues to monitor how both state and federal officials navigate this complex regulatory terrain.
In tandem with legislative action, medical boards and professional organizations are issuing guidance for clinical practice. Unlike binding state statutes, these recommendations are voluntary. In October, the American Medical Association (AMA) released a governance framework for agentic AI, outlining critical questions for systems that assume clinical functions. These questions examine the underlying evidence supporting AI use, appropriate oversight mechanisms, and the allocation of liability and risk. While the framework does not mandate specific compliance for physicians or developers, the AMA intends for it to steer future work on licensing and regulation.
Similarly, the Federation of State Medical Boards (FSMB) convened a workgroup in Washington to draft recommendations for state medical boards overseeing clinical AI tools performing tasks like diagnosis and prescribing with minimal or no direct physician supervision. FSMB leadership has consistently argued that AI should be governed through existing medical accountability structures, keeping physicians accountable for technology use. Draft recommendations are expected soon for public feedback.
Academic publishers are also establishing stricter rules. In October, the health policy journal Health Affairs updated its publishing policies regarding AI accessibility, research ethics, and human oversight. The policy prohibits authors from listing AI as a co-author or substituting it for human judgment, while requiring researchers to review AI-assisted work, maintain full content responsibility, and transparently disclose all AI usage. The journal additionally discourages using AI for reference management, citing documented tendencies of automated systems to generate inaccurate or nonexistent citations.
These widespread standards reflect a shared societal concern: maintaining human judgment and accountability. Whether clinicians are evaluating software tools, publishers are vetting scientific manuscripts, or policymakers are drafting legislation, the demand for transparency remains paramount, heavily mirroring public polling preferences for human-led healthcare decisions.
Low Trust in Federal Health Agencies May Shape Perceptions of Vaccine Injury Initiatives
While artificial intelligence reshapes clinical workflows, federal health agencies are concurrently tackling public health trust through new initiatives focused on vaccine safety. Last month, the Department of Health and Human Services announced a series of actions designed to expand research and reporting related to health problems following vaccination.
The sweeping initiative includes establishing a specialized vaccine injury clinic at the National Institutes of Health (NIH) Clinical Center and proposing a specialized Medicare billing code to evaluate and treat suspected vaccine adverse effects. HHS described the billing code as a mechanism to reimburse physicians for reporting observations to the Vaccine Adverse Event Reporting System (VAERS). Additional plans involve modernizing VAERS and requiring electronic health records to systematically capture adverse events.
Federal health officials have framed these measures as a direct response to patients who feel their post-vaccination health concerns have been overlooked or dismissed by the medical establishment. During the announcement, officials emphasized that the government must actively listen to these patients. Studies on patient-provider communication indicate that dismissive judgments can significantly erode trust and empathy over time, suggesting that thorough clinical investigations could help repair relationships with alienated patients. Nevertheless, health experts note that the temporal proximity of symptoms does not automatically establish direct causation.
Monitoring vaccine safety remains a continuous federal priority. While serious adverse events remain rare, agencies track potential safety signals through multiple channels, including pre-approval clinical trials and post-market surveillance programs like VAERS. Administered jointly by the Food and Drug Administration and the Centers for Disease Control and Prevention, VAERS serves as a passive surveillance system where anyone can submit unverified reports of health issues following vaccination, which do not independently prove causality.
Despite the clinical intent behind the HHS initiatives, the rollout has sparked internal and external debate. The board overseeing the NIH Clinical Center unanimously approved a statement raising concerns about how findings from the new vaccine injury clinic will be perceived amid low public confidence in federal health agencies. The board warned that research outcomes might be dismissed by the public regardless of their scientific validity. Furthermore, board members expressed concern that the clinic’s opening ceremony—which featured discussions with individuals reporting vaccine injuries—might create the premature impression that official conclusions had already been reached before scientific inquiry began.
Prominent medical organizations have also voiced skepticism. The American Academy of Neurology and the Infectious Diseases Society of America (IDSA) criticized the proposed Medicare billing code. IDSA informed the Centers for Medicare & Medicaid Services that the code lacks clinical necessity and could inadvertently lead patients and the public to overestimate the actual frequency of vaccine injuries. Furthermore, a distinction has emerged regarding the scope of the new center: while HHS framed the initiative narrowly around vaccine injuries, NIH leaders described a broader investigative focus encompassing long-term post-infection illnesses.
Ultimately, these initiatives highlight the delicate balance public health institutions must strike. Given prevailing public skepticism, the success of new vaccine safety research in restoring trust will depend heavily on clear, transparent communication regarding methods, limitations, and scientific conclusions.
Health and Wellness Influencers Frequently Promote Products, While Citing Sources Is Rare
Beyond clinical settings and federal agencies, everyday health information encountered on social media continues to heavily influence public perceptions. A recent analysis by the Pew Research Center examining over half a million posts across Instagram, TikTok, and YouTube found that more than one-third of health and wellness influencer posts incorporate sponsorship or encourage product use.
The study analyzed 535,512 posts from 6,334 influencers with at least 100,000 followers who regularly share health content. Among posts directly discussing health topics, nearly half featured a product or service promotion. Overall, 92% of evaluated influencers included promotional content at least occasionally, with commercial messaging proving especially prevalent in posts concerning nonprescription medications, dietary supplements, men’s health, and physical appearance.
Crucially, the analysis revealed that relatively few posts—only 9%—referenced external sources such as scientific experts, government agencies, or peer-reviewed research when making health claims. Citations were slightly more common only when influencers discussed health policy or warned of specific health harms.
This proliferation of commercial health content on social media presents significant evaluation challenges for the public. In KFF’s tracking poll on health information and trust, six in ten adults expressed confidence in their ability to distinguish true health information from false claims on social media, while roughly four in ten reported feeling little to no confidence. Although the Pew analysis did not measure the factual accuracy of influencer claims, the high frequency of commercial promotions coupled with a general lack of verifiable citations underscores the ongoing difficulty consumers face when navigating digital health landscapes.
