Published on March 20, 2026
Japanese Government Selects Domestic Language Model for State AI Initiative
KDDI and ELYZA have been selected as language model providers for the Japanese Digital Agency's government AI program.
Regulation 3 – 4 minutes min readIn Japan, government AI initiatives typically rely on foreign models – American or, less frequently, European ones. Therefore, the news that Japan's Digital Agency (デジタル庁) has selected a domestic language model for testing as part of its Government AI program is noteworthy.
Telecommunications giant KDDI, in partnership with ELYZA – one of Japan's leading language model developers – has been selected as a provider in the “domestically produced large language models” (国内大規模言語モデル) category. Simply put, their technology will be piloted within Japan's government AI infrastructure.
What is Government AI and Why is it Needed
Government AI is a state program through which Japan's Digital Agency is exploring the application of generative AI in government operations. This is not about implementing AI for AI's sake, but about addressing specific tasks: drafting documents, processing citizen inquiries, searching internal regulatory databases, and other administrative processes that currently demand significant human resources.
A key detail is the emphasis on data security and sovereignty. Government agencies handle sensitive information, and transferring it to foreign cloud services is undesirable, to say the least. This is precisely why the Digital Agency is purposefully seeking solutions based on Japanese models that can be deployed in compliance with national security and data storage requirements.
Why KDDI and ELYZA Were Chosen
ELYZA is a company that specializes in developing language models deeply adapted for the Japanese language. This isn't just about translating or fine-tuning Western models: the Japanese language, with its logographic writing system, complex grammar, and cultural nuances, demands serious dedicated work, and ELYZA has built up considerable expertise in this field.
KDDI, in turn, is one of the country's largest telecommunications operators with a sophisticated cloud and network infrastructure. The combination of a powerful language model and a reliable operational foundation is exactly what is needed for a stable and secure AI deployment in the government sector.
Simply put, ELYZA is responsible for the “brains,” while KDDI ensures that this brain operates in a stable, scalable, and secure manner.
What This Means for Japan's AI Market
Selection for the government program is more than just a vote of confidence. It sends a signal to the entire industry: Japanese language models are mature enough to compete with foreign counterparts, even in the most demanding environment – public administration.
For ELYZA, participating in Government AI provides an opportunity to refine its model on real-world administrative tasks and scenarios, which are fundamentally different from commercial ones. Government texts are a distinct genre: formal, saturated with legal jargon, and requiring precise phrasing. Working in this context will likely make the model significantly more useful in other professional areas too.
Moreover, such precedents influence how other agencies and large organizations view domestic solutions. If the government trusts a Japanese model, that serves as a powerful endorsement for the private sector too.
For Now, It's Only a Pilot
It should be clarified that this is currently a trial phase (試用), not a full-scale implementation. The Digital Agency is testing several solutions, and the final configuration of the government AI system has yet to be determined. The specifics of how it will interact with various agencies and which processes will benefit most from the model are all to be figured out during the pilot.
Nevertheless, simply being included on the list is an achievement in itself. For Japan's AI market, this marks another step toward domestic solutions becoming a full-fledged alternative to foreign platforms, not just in the commercial sector but in government as well.
You May Also Like
Events are only part of the bigger picture. These materials help you see more broadly: the context, the consequences, and the ideas behind the news.
SK Telecom has successfully completed the first phase of a national project to create a sovereign AI – a language model with 519 billion parameters.
AI: Events
Infrastructure
The Barcelona Zettascale Lab has completed validation of the Cinco Ranch TC1 processor. This is a crucial step toward building Europe's own infrastructure for training artificial intelligence models.
Crusoe has announced the opening of a manufacturing facility where prefabricated modular units will be assembled for the rapid deployment of computing infrastructure for AI tasks.
How This Text Was Created
This material is not a direct retelling of the original publication. First, the news item itself was selected as an event important for understanding AI development. Then a processing framework was set: what needs clarification, what context to add, and where to place emphasis. This allowed us to turn a single announcement or update into a coherent and meaningful analysis.
We openly show which models were used at different stages of processing. Each performed its own role — analyzing the source, rewriting, fact-checking, and visual interpretation. This approach maintains transparency and clearly demonstrates how technologies participated in creating the material.
The neural network studies the original material and generates a coherent text
Correction of errors, inaccuracies, and ambiguous phrasing
Generating a textual prompt for the visual model
Generating an image based on the prepared prompt
Don’t miss a single experiment!
Subscribe to our Telegram channel —
we regularly post announcements of new books, articles, and interviews.
