AI will help write laws in Kazakhstan

Kazakhstan has begun using artificial intelligence in the process of drafting new laws. The announcement was made by Dmitry Mun, Vice Minister of Artificial Intelligence and Digitalization.

Together with the Ministry of National Economy, the Ministry of Justice, and the Agency for Strategic Planning, the ministry is developing AI agents trained on Kazakhstan’s legislation. These agents help ensure that new legal norms do not contradict existing laws.

“We already have a prototype of an agent trained on our legislation and re-engineering experience. It helps verify that new norms do not conflict with current laws,” said Mun.

How it works

The development of these systems is based on the domestic language models KazLLM and AlemLLM, as well as the open platform KazLaw, available through OpenAI.

According to the Vice Minister, the best performance so far comes from an AI model built on foreign data.

“Today, GPT shows stronger results mainly because it has access to vast amounts of data and develops through hundreds of millions of dollars in investment. It incorporates international practice, which improves the quality of its responses. But when we talk about AI agents, it’s a longer process that can take up to a year. They need to be fine-tuned to perform well in specific, narrow tasks. We take a large language model and build alongside it our own knowledge base of Kazakhstan’s laws, updates, effects, and re-engineering documents — and as a result, such an agent becomes more accurate and useful than a universal model,” Mun explained.

What’s next

The Ministry of Artificial Intelligence and Digitalization is already working with other government bodies to create more than 15 specialized AI agents. In the future, they will not only identify legal contradictions but also analyze how legislative changes might affect the economy and public administration.

These technologies will become part of Kazakhstan’s new digital governance infrastructure, where artificial intelligence supports decision-making — without replacing people.