Over the course of 2026, Korea’s Ministry of Food and Drug Safety (MFDS) and Ministry of Health and Welfare (MOHW) introduced a series of measures that affect nearly every stage of a medical device’s lifecycle in Korea. Most of these measures serve three policy objectives. First, the government wants new devices to reach Korean patients faster. Second, it is building a defined regulatory framework for digital and AI-based devices, which now make up a growing share of applications. Third, it aims to reduce the administrative burden on manufacturers after approval while keeping safety oversight in place.
Shortening the Path to Market
Korea’s effort to speed up approvals centers on the Medical Product Approval and Review Innovation Plan, which the MFDS released in May 2026. The plan sets a target of reducing average review times for innovative medical devices, new drugs, and biosimilars, from roughly 420 days to 240 days, beginning in July 2026. To meet this target, the MFDS plans to grow its bio-health review staff from 369 to 564 reviewers. It is also adopting a parallel review model, in which separate teams assess different parts of an application at the same time.
The plan also places more weight on preparation before submission. The MFDS is expanding pre-application consultations and providing applicants with detailed checklists covering good clinical practice (GCP), good manufacturing practice (GMP), and risk management plans. Because a 240-day timeline leaves less room for repeated deficiency requests, manufacturers that use these consultations and submit complete dossiers are more likely to benefit from the shorter review period.
Building a Framework for Digital and AI Devices
Shorter review times depend on clear evidence requirements, and for digital and AI-based devices, those requirements had remained unclear in Korea. Much of the MFDS’s 2026 activity has therefore focused on defining what these products must submit and how the agency will evaluate them.
The central change came on July 28, 2026, when the MFDS revised its rules for digital medical devices to clarify when clinical trial data is required. Manufacturers had reported for some time that the existing exemption standards were vague. After reviewing about 1,000 past approval cases, the MFDS now ties the evidence requirement to a product’s function. Software that builds treatment plans from medical images can be approved with performance verification data alone. AI products that diagnose disease or interpret test results must still provide clinical data. The revision also extends the real-world use waiver for advanced AI products, including generative and multimodal AI devices, from one year to as long as three years.
The MFDS has also issued rules for specific technologies. On June 30, 2026, it published approval guidelines for devices built on large language models (LLMs), such as clinical decision support tools, medical record summarizers, and image interpretation aids. These guidelines build on the agency’s January 2025 rules for generative AI medical devices. They provide examples of LLM-based devices, instructions for preparing license applications, and methods for verifying clinical and functional performance. In particular, they require manufacturers to address risks specific to LLMs, including errors that arise over extended conversations, omitted or distorted information, and misreading of context. As a result, LLM developers should design validation studies that test for these failure modes directly.
In September, the MFDS also added three guidance measures under the Digital Medical Products Act, which governs digital products as a separate category. Under guidance released on September 16, manufacturers can have key components, such as an AI algorithm or biosensor, evaluated in advance and then reference those cleared components in a later application for the full system. The MFDS expects this component-level review to lower development costs and shorten time to market. The agency also updated its clinical trial protocol guidance for digital products across seven indications, including depression, attention deficit hyperactivity disorder (ADHD), and mild cognitive impairment. Finally, a GMP Q&A explains that companies holding medical device GMP certification must still complete a separate audit for digital products, although a streamlined process may apply in some cases.
Easing Requirements After Approval
After a device is approved and enters the market, manufacturers must manage design and manufacturing changes, maintain GMP compliance, and meet surveillance obligations. The MFDS addressed each of these areas in 2026.
In March, the MFDS published a draft revision to the Enforcement Rule of the Medical Devices Act that introduces a “negative list” system for device changes. Under this system, manufacturers need prior MFDS approval only for changes that could significantly affect safety or performance, such as changes to intended use, mechanism of action, materials, or manufacturing site. Manufacturers can manage all other changes internally under the change management procedures they submitted with their original application. The same draft cuts the GMP verification period for conditionally approved devices from 20 days to 10 days. At the same time, it reinforces oversight by clarifying that the MFDS may recall devices that fail to meet either standard specifications or approved testing specifications during post-market inspections. It does this by allowing regional authorities to publish investigation results when foreign substances or other issues are found in devices.
The MFDS also clarified its post-market surveillance rules for rare disease devices. Korea requires surveillance of rare and newly developed devices for four to seven years after they go on sale. Rare disease devices with prior overseas use may qualify for an exemption, but until recently the MFDS decided these cases individually, without defined criteria. A September 2026 revision now defines what counts as qualifying overseas use and lists acceptable evidence, including overseas clinical data. For foreign manufacturers that already market rare disease devices in other countries, this gives a defined route to avoid duplicating surveillance in Korea.
Supporting Adoption After Approval
Although the MFDS has cleared a growing number of AI medical devices, hospitals have been slow to use them because many lack the clinical evidence and reimbursement frameworks needed for routine care. On April 10, 2026, the MOHW launched a funding program to address this. The program supports clinical validation, real-world evidence generation, health economic studies, and reimbursement planning. To qualify, companies must form consortia with hospital partners, which will support multi-center clinical studies and real-world data collection. The program is part of AX-Sprint, a 754 billion won (about USD 510 million) government initiative to expand AI use across the Korean economy. In this way, the funding program complements the MFDS’s regulatory changes for AI devices by helping approved products build the evidence that hospitals and payers require.
Conclusion
Korea’s 2026 reforms form a comprehensive set of policies across the device lifecycle. The MFDS has shortened review timelines, defined evidence requirements for digital and AI devices by function and technology, and reduced prior approval requirements for post-market changes. At the same time, the MOHW has funded the evidence generation that AI devices need for hospital adoption.
Written by: Ames Gross – President and Founder, Pacific Bridge Medical (PBM)
Mr. Gross founded PBM in 1988 and has helped hundreds of medical companies with regulatory and business development issues in Asia. He is recognized nationally and internationally as a leader in the Asian medical markets. Mr. Gross has a BA degree, Phi Beta Kappa, from the University of Pennsylvania and an MBA from Columbia University.