Ministry of Science and ICT Requests Two Years of Data
National Court Administration: “Reviewing Scope of Provision”

The Ministry of Science and ICT is in discussions with the Supreme Court, having requested the provision of a large-scale judgment dataset, to build the government’s independent AI foundation model (the so-called “national AI”). The Ministry has asked the National Court Administration for approximately 1 million judgments issued before August 2025, which are subject to disclosure through the Judicial Information Disclosure Portal, to be used as training data for AI. The National Court Administration is reportedly reviewing the scope of disclosure with caution, citing reasons such as personal data protection and the need for a legal basis.


The Artificial Intelligence Data Policy Division of the Ministry of Science and ICT and the Judicial Informatization Office of the National Court Administration recently held a working-level meeting regarding the provision of judgment data. The materials requested by the Ministry include anonymized trial and Supreme Court judgments from the past two years, as of August 2025, totaling around 1 million cases. This dataset is intended as core training data for the government’s “national AI” project, which aims to develop an independent AI foundation model with a deep understanding of the Korean language and domestic social and cultural context by 2030. It has also been reported that, in pursuit of securing “high-quality data,” the Ministry of Science and ICT has requested relevant data held by other government agencies beyond the courts.


The National Court Administration maintains that this is a matter requiring careful consideration. Providing the past two years’ undisclosed judgments in bulk must be reviewed thoroughly for legal grounds and personal information protection. The court is said to be continuing discussions with the Ministry, focusing first on establishing the scope and criteria for data provision.


The Civil Procedure Act and Criminal Procedure Act require strict personal data protections when disclosing judgments. Originals of judgments included in litigation records cannot be disclosed or accessed without a specific legal basis. As of now, there is no clear standard for how the Ministry of Science and ICT would store, delete, or manage data used for AI training after the fact.


Within the judiciary, some hold the view that providing judgments should be approached with caution. A chief judge at a district court said, “There must be a prior statutory provision to confirm the parties’ intentions before using judgments for AI training,” adding, “It is necessary to confirm the parties’ intentions at the conclusion of pleadings or at the time of the ruling, according to regulation.” The judge continued, “No matter how thoroughly data is anonymized, it is difficult to completely rule out the possibility of re-identification, and if such data is provided to domestic AI but restricted from overseas big tech companies, potential issues of fairness or trade may arise.” The judge also noted, “It is highly questionable whether the public would consent if they knew the intimate details revealed in their judgments were being used for AI training.”


The National Court Administration plans to comprehensively review the legal grounds and scope of data provision, with the Planning Division taking the lead and involving related departments such as the Judicial Support Office and Judicial Informatization Office.



Jisoo Kim, Law Times Reporter


※This article is based on content supplied by Law Times.

This content was produced with the assistance of AI translation services.

© The Asia Business Daily. All rights reserved. Unauthorized AI training and use prohibited.

Today’s Briefing