Anna Jiwon Park has been covering the politics at The Korea Times since the summer of 2024, when she joined the press pool for the Office of the President in Korea. Prior to that, she spent about five years reporting extensively on financial markets, regulatory authorities and the financial industry. She joined The Korea Times in 2019 after spending eight years as a broadcast journalist at Arirang TV, Korea’s leading global broadcaster, covering politics, defense and culture.
Gov't to phase out HWP file format for AI compatibility
The icon image of a HWP format file / Courtesy of Hancom
By Anna J. Park
Korea is moving to phase out the Hangul Word Processor (HWP) file format across key government document systems, as public institutions seek to modernize workflows and improve compatibility with artificial intelligence (AI) tools.
The Presidential Council on National Artificial Intelligence Strategy (NAIS), together with the Ministry of the Interior and Safety and the Ministry of Culture, Sports and Tourism, plans to begin restricting HWP file attachments across key public communication channels starting May. These include the government’s internal document platform as well as official email services.
Beginning May 18, the government will expand adoption of HWPX — an open-format version of HWP designed for greater interoperability — across the On-Nara System, the document drafting and distribution platform used by public officials, extending its application to local governments. On-Mail, an internal communication tool for civil servants, will follow suit by October. Both systems are overseen by the interior ministry.
The Integrated Government Mail service, which operates under the korea.kr domain and is managed by the culture ministry, will implement a full restriction on HWP attachments in October following a grace period beginning in early May. The service functions as an official external email channel for civil servants communicating with the public.
The primary driver behind this policy shift is the structural limitation of HWP files. Developed in the late 1980s by Hancom, Hangul became the dominant word processor in Korea, particularly in government institutions, due to its strong Korean-language support and localization advantages at a time when global alternatives like Microsoft Word were less accessible. Over time, however, the format has remained largely proprietary and closed, limiting interoperability and complicating integration with newer technologies.
gettyimagesbank
As a closed-format system, HWP makes it difficult for AI technologies to access, analyze and learn from embedded data. While recent advancements, such as support for reading HWP files in tools like ChatGPT, have improved usability to some extent, fundamental constraints persist due to its closed architecture.
HWPX, by contrast, offers a transparent structure far more compatible with AI-based data processing. The interior ministry is also working with Hancom to enable automatic conversion of existing HWP documents into HWPX during editing or saving, a step intended to make the transition largely seamless for most users.
The policy shift does not spell the end of the Hangul itself, which remains widely used in Korea. Recent versions of the software already support HWPX, meaning users can continue working in the word processor as before. The change targets the underlying file format, not the application.
NAIS Vice Chairman Im Moon-young emphasized that the initiative marks the beginning of broader data innovation in the public sector.
"Starting with this measure, we will push forward small but significant and fast-paced changes for public-sector data innovation in the AI era through close inter-ministerial cooperation," Im said.
The policy direction reflects priorities set by President Lee Jae Myung, who flagged the limitations of the HWP-centered document system for AI during a government briefing in December 2025.
"Government documents are valuable data assets, but their structure often makes them difficult for machines to read," Lee said at the time, noting the importance of improving the AI usability of public data.