AI with Ukrainian DNA: Inside Ukraine's push to create its own language model
Ukraine is creating its own AI infrastructure (illustrative photo: Getty Images)
Ukraine plans to become one of the top three global leaders in the development and implementation of artificial intelligence. The infrastructure for this goal is already being built today, says Danylo Tsvok, Chief AI Officer of the Ministry of Digital Transformation, in an interview with RBC-Ukraine.
Key points
- Ambitious goal. Ukraine aims to rank among the top three countries in AI development by 2030.
- AI sovereignty. Ukrainian specialists are creating their own AI Factory and a national large language model.
- Ukrainian context. A domestic AI infrastructure will help protect data and operate with accurate information.
- Accessibility. At first, the Ukrainian LLM will serve as the foundation for government chatbots and AI assistants, but later, the model will be open-source.
How Ukraine is building AI infrastructure
The expert said there is currently an important mission to enter the world’s top three countries in AI development and implementation by 2030.
“Now we're moving from a digital state to an agentic state, where AI assistants proactively meet the needs of citizens and the government,” Tsvok added.
He noted that “real progress is impossible without AI sovereignty.”
For this purpose, Ukraine is building its own infrastructure:
- AI Factory (an artificial intelligence factory — the country’s first state infrastructure for AI solutions);
- a national LLM (large language model).
“This guarantees data security and the development of Ukraine as a technological leader,” Tsvok stressed.
What LLM is
“What is a Ukrainian large language model? We're not creating a new ChatGPT or Gemini, or any other product,” the representative of the Ministry of Digital Transformation said.
He explained that every such product has a core, a brain center.
“This core is a large language model. It is a large neural network trained on massive volumes of information,” Tsvok said.
Why Ukraine needs its own language model
“Why does Ukraine need a national LLM? In general, foreign models do not have a deep understanding of the Ukrainian context.”
The head of the AI direction at the ministry explained that specialists want to create a model that will be pro-Ukrainian and understand the Ukrainian language, culture, history, terminology, and national specifics.
“In essence, this means laying a brick of our DNA into the field of artificial intelligence. The Ministry of Digital Transformation is currently doing this in partnership with the Kyivstar company,” Tsvok said.
The second reason, he noted, is that people today live in conditions of information warfare.
“It is especially important that models provide answers on political and historical topics without bias against Ukraine,” the expert emphasized.
For example, there should be no doubt about Crimea, about the causes of the war in eastern Ukraine, or about what the so-called LNR or DNR are.
“We must understand that our enemy is also doing its work. It has its own approaches to poisoning LLM available on the market. So, we need to have our own primary source of truth,” Tsvok said.
Finally, the third reason is that the Ukrainian LLM will be 2.5–3 times cheaper than foreign counterparts.
“By having our own infrastructure and our own national large language model running on state infrastructure, we fully ensure AI sovereignty. This guarantees that data does not leave the country in any way,” the official stressed.
AI sovereignty is the ability of a country to independently develop, launch, and control artificial intelligence systems without critical dependence on foreign companies, clouds, or models.
When the first Ukrainian LLM may appear
Tsvok said the Ukrainian language model is currently in the active phase of development.
“There is a technological part handled by our partner company, Kyivstar. They are responsible for developing and training the large language model. We, as the state, are responsible for collecting and aggregating the data used to further train the model,” he explained.
He noted that the most important stage right now is data collection.
Government institutions, universities, research institutions, businesses, and media organizations are involved in this process. Efforts are also focused on a testing system (benchmarks) that measures how well the model knows Ukrainian history, language, and whether it follows ethical standards and intellectual property rights.
Developers have also created critical architectural components (for example, a tokenizer) necessary to build a large language model, as well as a pipeline — a sequence of engineering steps for training the model on large datasets.
“The Ukrainian LLM will serve as the foundation for government chatbots and AI assistants. First, we will integrate the model into Diia.AI, and later into an AI tutor in Mriia,” Tsvok said.
After testing, he added, the model will become open-source and available to everyone. This will allow businesses, scientists, and civil society organizations to create their own products.
“We plan to release the first version of the model by the end of spring 2026,” the head of the AI direction at the Ministry of Digital Transformation concluded.
We also explored whether AI will take your job and how the labor market may change in the era of artificial intelligence.
