Listen instead
When a medical software developer updates an algorithm to improve diagnostic accuracy, the change can trigger a bureaucratic cascade. For years, the Food and Drug Administration evaluated these algorithmic updates using a regulatory framework originally built for physical hardware, like pacemakers and artificial joints. That structural mismatch is now prompting a reorganization at the highest levels of the agency. The U.S. Department of Health and Human Services has appointed Jared Seehafer, a medtech entrepreneur and former compliance software executive, as the first Deputy Commissioner for Technology and Artificial Intelligence at the FDA. The appointment represents a deliberate shift in how the federal government intends to govern clinical algorithms, moving from reactive review panels to a centralized, proactive strategy.
A Software Founder in Federal Leadership Seehafer enters the newly created role with direct experience in the commercial software regulatory pipeline. He is the co-founder and former chief executive officer of Enzyme, a digital regulatory platform designed to streamline compliance for health technology developers. His transition to public service began in 2025 when he joined the FDA as a senior advisor. In that advisory capacity, Seehafer helped author the agency's generative AI discussion paper, which outlined the initial parameters for managing advanced models. By placing a software founder in a permanent executive position, the agency is integrating practical development experience directly into its leadership team.
This background is critical because software development operates on timelines measured in weeks, while traditional medical device approvals often take months or years. Seehafer's understanding of the commercial pressures and technical hurdles faced by startups may help the agency build review processes that are thorough without being dilatory.
Unifying a Fragmented Regulatory Landscape For the past decade, medical artificial intelligence regulation was managed across a fragmented series of device review panels. This decentralized structure meant that algorithmic tools were evaluated under frameworks that did not account for the unique properties of digital systems. As generative models emerged, their capability to adapt and process diverse datasets quickly outpaced traditional statutory definitions. The creation of the Deputy Commissioner role signals that algorithmic oversight is now central infrastructure rather than a task delegated to side committees.
Seehafer is tasked with leading FDA-wide strategy across software, artificial intelligence, and digital health. His office will coordinate policy across four major centers: the Center for Drug Evaluation and Research, the Center for Biologics Evaluation and Research, the Center for Devices and Radiological Health, and the Center for Tobacco Products. This cross-departmental coordination aims to ensure that software standards remain consistent, whether an algorithm is used to design a biologic, monitor a drug trial, or operate a medical device.
Historically, these centers operated with a high degree of autonomy, leading to inconsistent requirements for developers whose products crossed traditional clinical boundaries. For example, a software tool that analyzes patient data to predict drug efficacy might face different validation standards depending on whether it was submitted as a companion diagnostic or part of a clinical trial protocol. A centralized authority under Seehafer is designed to resolve these discrepancies, providing a single point of policy coordination that should make the regulatory pathway more predictable for industry participants.
Managing the Scale of Modern Algorithmic Medicine The volume of software seeking regulatory clearance has increased steadily. The FDA's public registry now includes more than 1,000 authorized artificial intelligence and machine learning medical devices, a number that continues to grow as developers integrate machine learning into clinical workflows. Managing this volume while preparing for future technologies requires new methods of evaluation. The agency is already testing premarket models through its generative AI pilot initiatives, which explore how algorithms behave under clinical conditions before they receive formal clearance. This testing represents a move toward continuous monitoring, reflecting the reality that modern software does not remain static once deployed.
Under Seehafer, the agency will need to balance the rapid iteration cycles of software developers with the safety standards required for clinical validation. This balance is particularly challenging because machine learning models can exhibit drift, where their performance degrades when applied to patient populations different from those used during training. To address this, the FDA is exploring lifecycle management frameworks that require developers to monitor performance in real-world clinical settings post-clearance. This shift from a one-time premarket gatekeeper to an ongoing evaluator represents a fundamental change in the agency's operational philosophy.
The Evolution of Software as a Medical Device The regulatory concept of Software as a Medical Device, often abbreviated as SaMD, has existed for several years, but its practical application has evolved. Early software applications in medicine were largely administrative or served as simple calculators. Today, algorithms are used to interpret complex imaging data, identify potential cardiac anomalies from wearable sensor feeds, and recommend specific oncology treatment paths. These high-stakes clinical applications require a level of scrutiny that matches their potential impact on patient outcomes.
The challenge for the FDA is that traditional clinical trials are designed for static interventions, such as a pill or a physical implant, which do not change after they are manufactured. Software, by contrast, is designed to be updated continuously to patch security vulnerabilities, improve user interfaces, and refine underlying models. Under the current statutory framework, significant changes to a cleared device often require a new regulatory submission, a process that can discourage developers from making beneficial updates. Seehafer's office will need to work within existing laws to find administrative solutions, such as pre-specifying change control plans, which allow developers to outline future software modifications and the validation methods they will use before the product is cleared.
Source: Enzyme/U.S. Department of Health and Human Services (HHS)
