South Korean courts and prosecution offices are developing a proprietary artificial-intelligence model for legal research and case-law search [1].
This initiative aims to eliminate reliance on commercial AI services, which pose significant security risks. By building a private system, the government intends to prevent the accidental ingestion of suspect personal data or confidential investigation details into external databases [1].
According to reporting by journalist Lee Joon-yeop, a pilot AI system has been in trial operation since February 2024 [1]. This trial serves as the foundation for a more comprehensive digital transformation of the legal system. The prosecution has also begun similar work to integrate AI into its reporting and research workflows [1].
The full-scale project carries an investment of 161 billion won [1]. This funding supports the creation of a closed-loop environment where legal professionals can search precedents, and generate reports without compromising state secrets or privacy [1].
Officials said the project is expected to reach completion around 2028 [1]. The transition is designed to improve efficiency in legal research while ensuring that the judiciary maintains absolute control over its data, a critical requirement for maintaining the integrity of the legal process [1].
The current shift reflects a broader trend of government agencies moving away from public large language models toward "sovereign AI" to ensure data sovereignty [1].
“South Korean courts and prosecution offices are developing a proprietary artificial-intelligence model”
This move signals a strategic shift toward 'sovereign AI' within the South Korean government. By investing heavily in a closed system, the state is acknowledging that the efficiency gains of commercial AI are outweighed by the security risks of data leaks in the judicial process. This sets a precedent for other high-security government sectors to develop isolated AI infrastructures rather than utilizing third-party cloud services.
