Artificial intelligence is reshaping modern healthcare, bringing both optimism and apprehension. As its influence expands from diagnostics to administrative functions, the American Medical Association (AMA) has taken an active role in shaping policy to ensure that AI enhances—rather than compromises—patient care. With physicians increasingly integrating AI into clinical practice, concerns have mounted over liability, transparency and patient safety. Federal and state responses have emerged unevenly, revealing a fragmented regulatory landscape. The AMA and allied experts now advocate for robust governance frameworks that centre the physician’s role, protect patients and hold technology to account.
Balancing Innovation and Oversight
Policymakers at both federal and state levels face the challenge of encouraging innovation without compromising safety. While federal activity has lagged, the past two years have seen a surge of state-level initiatives. More than 250 AI-related health bills were introduced across 34 states in 2025 alone, targeting transparency, discrimination, payer practices and clinical usage. States like Colorado, California and Utah have emerged as early leaders. Colorado’s broad AI law addresses algorithmic discrimination and sets developer-to-user transparency standards, though its implementation has faced criticism and delay efforts. California’s focus includes mandating disclosures to patients when generative AI is used without clinical oversight. These legislative experiments serve as policy laboratories, with potential to shape national approaches.
Must Read: The Evolution of AI in Healthcare: Navigating Innovation and Regulation
However, the pace of technological advancement far outstrips the capacity of most lawmakers, complicating efforts to introduce meaningful safeguards. Concerns remain about the overreach of deregulation efforts. A proposed ten-year federal moratorium on state-level AI regulation, if enacted, would significantly curtail local governance in a space still in flux. Experts argue that such a freeze would hand disproportionate power to technology developers, leaving health systems and physicians without the guardrails necessary to responsibly deploy AI.
Physician-Centred Policy Priorities
Physicians face growing uncertainty over their responsibilities when adopting AI tools, particularly concerning liability and ethical practice. The AMA prefers the term “augmented intelligence” to reflect AI’s supportive—not replacement—role in clinical decision-making. Despite this framing, many physicians are wary. Concerns over legal exposure are widespread, particularly where AI use may deviate from or become perceived as the standard of care. While AI is not yet uniformly considered standard practice, the lack of legal clarity makes physicians potentially liable for decisions made using flawed or opaque tools.
Transparency is a central demand. Physicians require detailed information about AI tools—how they are trained, their performance across different populations and their clinical limitations. Without these disclosures, meaningful consent and responsible use become nearly impossible. Equally, patients must be informed when they are interacting with AI, particularly in non-clinical scenarios like chatbots or administrative communications. Ethical practice demands openness, especially when trust and safety are at stake.
Meanwhile, AI’s use by health insurers adds another layer of complexity. Physicians report increasing rates of care denials driven by automated systems. In response, several states are crafting legislation to ensure that medical necessity decisions involve qualified human reviewers and that AI-generated denials are not based solely on population data. There is also growing interest in requiring public reporting of AI use in claims and prior authorisation processes. These initiatives aim to ensure that efficiency gains do not come at the cost of fairness or patient access.
Governance and the Path Ahead
Institutional governance is emerging as a critical priority. In the absence of comprehensive federal standards and amidst varied state approaches, health systems and payers are being encouraged to establish their own internal guardrails. This includes ensuring meaningful human oversight of AI recommendations, maintaining accountability structures and validating systems for bias and clinical efficacy. However, this decentralised model risks inconsistency and unequal protection across organisations and jurisdictions.
Federal agencies have taken some steps. The Office of the National Coordinator for Health IT issued rules mandating algorithmic transparency in certified electronic health records. The FDA released draft guidance for AI-enabled medical device approval, including labelling, validation and cybersecurity requirements. Yet many tools fall outside FDA purview. Post-market surveillance—critical to identifying issues such as algorithmic drift—is not uniformly enforced. Simultaneously, the shift in federal administration suggests a tilt toward deregulation. Proposals to roll back executive orders and remove consumer protections raise concerns about the long-term commitment to patient-centred safeguards.
Data privacy remains another unresolved issue. AI requires vast data to function effectively, but patient records are uniquely sensitive. Existing privacy laws like HIPAA may prove insufficient against the ambitions of big tech firms seeking unfettered access to healthcare data. Calls for stronger federal privacy legislation are growing louder, not only to limit misuse but to uphold public trust in health systems increasingly mediated by algorithms.
The future of AI in healthcare hinges on regulation that prioritises patient safety, clinician responsibility and equitable access. While the technology promises to enhance care delivery and ease administrative burdens, its deployment must be governed with rigour and foresight. Physicians need clear guidance, robust transparency and protection from undue liability. Patients need confidence that AI serves their interests, not merely those of insurers or developers. The AMA’s push for augmented intelligence rightly emphasises the continued centrality of human judgement.
Achieving a balanced, effective policy framework will require collaboration across sectors, grounded in clinical realities and ethical imperatives. Without it, the promise of AI risks becoming another source of fragmentation in an already complex healthcare system.
Source: American Medical Association
Image Credit: iStock
