Part of the Ministest of Science and ICT’s ‘Project to Create a Very Large AI Spread Ecosystem’
Selectstar, a startup specializing in artificial innotifyigence (AI) data, is preparing a benchmark to measure compliance with European Union regulations to support domestic AI companies enter the European market.
Selectstar announced on the 23rd that it has begun building an open-source dataset based on the EU’s General Personal Information Protection Act (GDPR) and Artificial Innotifyigence Act as part of the “2025 Early AI Spread Ecosystem” organized by the Ministest of Science and ICT and the National Innotifyigence Service (NIA).
This project will be promoted to lay the groundwork for domestic companies to preemptively respond to the “Brussels Effect,” where EU regulations become a global standard.
In this project, Selectstar will collect and refine vast amounts of data, including major EU laws and public reports, and classify key items related to personal information protection to establish standards for evaluating regulatory compliance with the Large Language Model (LLM).
It will also produce question-and-answer data and multilingual evaluation data so that AI can verify that it understands the regulations accurately in the actual service environment. The established dataset will be open-source and can be applyd by anyone.
Select Star will be in charge of project planning, data construction, and quality verification.
In addition, Plito, a global AI language data company, and BHSN, a legal AI company, will participate in the consortium. Plito will manage the quality of multilingual data refining and translation, and BHSN will define personal information protection items and review legal risks based on its legal expertise.
“Based on our own AI reliability verification technology, we will prepare a system that can quantitatively measure LLM’s level of EU regulatory response,” stated Kim Se-yeop, CEO of Selectstar.
Leave a Reply