Federal health officials are weighing how to regulate generative AI medical devices that can perform tasks traditionally reserved for doctors, and some legal experts are questioning whether the FDA has the authority to do so. The agency's "competency" approach to AI regulation has sparked concerns about how it will classify and oversee devices that diagnose, treat, or manage patient care without direct physician input.
The debate centers on a fundamental tension: existing medical device regulations were written for software that assists clinicians, not systems that can independently interpret scans, recommend treatments, or monitor patients between visits. As generative AI models become more capable, the line between tool and practitioner is blurring faster than the regulatory framework can adapt.
The competency question
The FDA has signaled that it plans to evaluate AI medical devices based on their demonstrated competency rather than their underlying technology. That approach sounds reasonable in theory, but legal experts say it raises difficult questions about what standard of proof the agency will require and whether it has the statutory authority to make those judgments.
At issue is whether the FDA can regulate an AI system that performs clinical functions as a "device" under current law, or whether such systems might fall outside the agency's jurisdiction entirely. If an AI system is effectively practicing medicine, some argue, it may be the responsibility of state medical boards rather than the FDA.
The stakes are considerable. Generative AI products are already being marketed for radiology, pathology, and clinical documentation. Healthcare systems are adopting these tools to address staffing shortages and reduce physician burnout. A regulatory framework that is either too slow or too permissive could shape the market for years.
What the FDA has proposed
The agency has not yet published final rules for generative AI medical devices, but its public statements suggest a performance-based approach. Under this model, the FDA would evaluate what an AI system can actually do - its accuracy, reliability, and safety in real-world conditions - rather than trying to fit it into predefined device categories.
Legal experts following the rulemaking process say the approach has merit but also creates uncertainty. Manufacturers would need clarity on testing requirements, validation standards, and post-market surveillance obligations before they can confidently bring products to market. AI for Regulatory Affairs Specialists is becoming a critical discipline as companies navigate these unresolved questions.
The broader question of agency authority remains unresolved. Some legal scholars argue that Congress needs to pass new legislation explicitly granting the FDA jurisdiction over autonomous AI systems that perform clinical work. Others contend the agency already has sufficient authority under the Federal Food, Drug, and Cosmetic Act and simply needs to exercise it.
Industry response and uncertainty
Medical device manufacturers have largely welcomed the FDA's willingness to engage with generative AI, but they are pressing for clearer guidelines. The cost of developing and validating these systems is substantial, and regulatory uncertainty makes it harder to justify those investments.
Clinicians have their own concerns. Many are wary of AI systems that make recommendations they cannot easily verify, particularly in high-stakes specialties like oncology and cardiology. The competency framework, if it requires demonstrable performance in clinical settings, could help address those concerns - but only if the standards are transparent and consistently applied.
Why this matters for healthcare professionals
For clinicians, hospital administrators, and healthcare executives, the FDA's regulatory approach will determine which AI tools reach their institutions and how quickly. A competency-based framework could accelerate approval of systems that clearly improve patient outcomes, but it could also leave gaps where evidence is thin.
Healthcare professionals should pay attention to how the FDA defines competency and what evidence it requires. Those standards will shape the AI tools available in clinical practice, the training needed to use them safely, and the liability questions that follow when AI systems are involved in patient care. Understanding the regulatory landscape is no longer an administrative concern - it is becoming part of clinical competence itself. For professionals seeking to stay ahead of these changes, AI for Healthcare training can provide practical grounding in how these systems work and where they fit in clinical workflows.
Your membership also unlocks:
