Abeja Joins NEDO Project to Develop Privacy-Focused AI Learning Methods

TOKYO, Aug 04 (Pulse News Wire) – Abeja,inc. (5574.T) announced its participation in a project led by CODAS, a non-profit organization, aimed at developing efficient AI learning methods that consider data privacy.

The initiative, funded by NEDO, focuses on establishing a “privacy-first” approach to AI that allows companies to securely utilize industrial data without security risks. ABEJA will take charge of constructing industry-specific large language models (LLMs) within the project. The company plans to integrate advanced confidentiality computing techniques to ensure secure processing while meeting stringent security requirements. Initial applications will target finance and insurance sectors, expanding to manufacturing and other key industries subsequently.

The project runs from 2026 to 2028 and aims to create an ecosystem where domestic enterprises can safely adopt privacy-focused AI technologies through open-source development. ABEJA’s involvement leverages its expertise in enterprise AI implementation and recent advancements such as NVIDIA's Nemotron 3 model integrated into its ABEJA Platform. ABEJA continues to support AI adoption across various industries through its platform, aiming to drive transformative changes in industrial structures. The company believes this collaboration could contribute positively to its long-term performance despite having a minor impact on the current fiscal period.