The FDA is seeking public input on how to regulate generative AI-enabled medical devices, highlighting both their potential benefits for patient care and the unique risks they present.
The Food and Drug Administration (FDA) is currently soliciting stakeholder feedback on the regulation of medical devices that incorporate generative artificial intelligence (AI). This initiative, announced on August 19, 2026, follows the publication of a discussion paper by the FDA’s Center for Devices and Radiological Health (CDRH), which outlines preliminary considerations for developing a regulatory framework for these emerging technologies.
Generative AI systems, which are capable of creating text, images, or audio by mimicking patterns in data, hold significant promise in enhancing patient care. However, the FDA acknowledges that such systems may introduce unique risks that differ from those associated with traditional AI and software applications. As of now, the CDRH has authorized over 1,000 AI-enabled medical devices, yet only a limited number utilize generative AI, prompting the agency’s call for public input before establishing a comprehensive regulatory approach.
Understanding the Risks of Generative AI
The FDA’s discussion paper emphasizes that generative AI devices pose distinct regulatory challenges. The technology can produce outputs that may appear authentic but can also lead to “hallucinations”—instances where the AI generates misleading or inaccurate information. Furthermore, these devices typically accept open-ended inputs and can perform a variety of tasks, resulting in variable outputs that complicate the assessment of their intended use and real-world performance. This complexity necessitates a regulatory framework that recognizes the nuanced differences between generative AI devices and their traditional counterparts.
In its paper, the CDRH proposes a risk framework that categorizes generative AI-enabled devices based on the type of activities they perform. The framework identifies three primary risk categories: informational, non-directive action, and fully autonomous actions. Software that provides risk scores for predicting cardiovascular events, for example, falls within the lowest risk category, while devices that autonomously prescribe treatment or direct users to take specific actions represent higher risk levels. The paper also notes that the context in which a device is used may influence its risk categorization, adding another layer of complexity to the regulatory process.
Regulatory Approaches: Premarket Evaluation and Postmarket Monitoring
As the CDRH considers how best to evaluate generative AI devices prior to market entry, it advocates for a competency-based approach. This methodology would involve benchmarking devices against established standards and testing them in clinical settings to evaluate their safety and effectiveness in real-world applications. The goal is to ensure that the final versions of these devices perform reliably and meet safety protocols before they are made available to clinicians and patients.
In addition, the discussion paper highlights the importance of postmarket monitoring. While the primary responsibility for this oversight would lie with device manufacturers, the FDA suggests that healthcare professionals, institutions, payers, and other authorities may play a role in ensuring ongoing performance and safety of generative AI devices after their deployment. This collaborative approach to postmarket monitoring reflects a growing recognition of the need for shared responsibility in overseeing the safety and efficacy of innovative medical technologies.
The CDRH is also contemplating whether to accept a greater degree of premarket uncertainty regarding the benefits and risks associated with generative AI-enabled devices. This could involve a regulatory shift towards increased reliance on postmarket surveillance to address any emerging issues that arise once the devices are in use. Such an approach seeks to balance the imperative of promoting innovation with the need to protect public health and safety in an era of rapidly evolving medical technologies.
Engaging Stakeholders for Comprehensive Feedback
The FDA is accepting comments on the discussion paper until October 19, 2026, as part of its broader effort to refine its regulatory framework for generative AI in medical devices. This outreach is a continuation of the agency’s commitment to incorporating stakeholder input into its decision-making processes. In recent years, the CDRH has engaged with various stakeholders, including forming a Digital Health Advisory Committee in 2024 to discuss total product lifecycle considerations for generative AI-enabled devices. Subsequent meetings have also focused on the application of generative AI in digital mental health devices.
As generative AI technology continues to advance, the FDA’s proactive stance in seeking feedback and developing a regulatory framework will be crucial in ensuring that these innovations can be safely and effectively integrated into healthcare. Ongoing dialogue between the FDA, industry stakeholders, and the public is essential for shaping the future of generative AI in medical devices and ensuring that the potential benefits can be realized without compromising patient safety.
No Comment! Be the first one.