Last week, the law was returned to the Mazhilis for revision after senators proposed changes regarding mandatory liability insurance for damage caused by AI systems. Previously, the law only considered voluntary insurance. Now, AI owners must not only mitigate damage but also take preventive measures to avoid potential harm.
Kazakhstan is not the first country to adopt a law on AI. The European Union has introduced comprehensive legislation concerning high-risk systems, including liability and penalties, which will come into effect gradually in 2025 and 2026. South Korea has also adopted a framework law on AI. Other countries, such as China, the USA, the UK, and the UAE, have their own regulations governing this area, while laws in Japan, Brazil, and India are in the development stage.
Previously, another law was adopted in Kazakhstan regulating the use of AI, which prioritizes human rights, transparency, and national values. This document prohibits the leakage of confidential data into neural networks, emotion analysis, and introduces mandatory labeling of deepfakes.
Note In the Kazakh parliament, draft laws are first adopted in the Mazhilis, then sent to the Senate. There, the law can be approved and sent to the president for signing or returned to the Mazhilis for revision, as happened with the AI draft law.
