Starting in January 2026, Kazakhstan will enforce the law “On Artificial Intelligence”, signed by the President last year. One of the key provisions of the document is the mandatory labeling of content created or modified using artificial intelligence systems.
The law defines the rights and obligations of owners and operators of AI systems and introduces requirements for informing users about so-called synthetic outputs.
What Is Considered AI Content
Synthetic outputs include images, videos, audio, texts, or their combinations that are created or modified using artificial intelligence and imitate the appearance, voice, behavior of people, or events that did not actually occur in reality.
The publication and distribution of such content is permitted only if it is clearly and visibly labeled. The absence of labeling will be treated as a violation.
Who the Requirement Applies To
Labeling is mandatory for individuals and legal entities, including companies, media outlets, bloggers, and other organizations, if they:
• are owners or operators of AI systems;
• distribute the results produced by such systems.
At the same time, responsibility for the absence of labeling will rest with the owners of AI systems, not with individual users who use neural networks to create content.
Should Ordinary Users Be Concerned
If a user creates images or texts using popular neural networks, the risk of liability is minimal. According to the relevant authorities, many major AI platforms have already implemented automatic content labeling. For example, in the Gemini neural network, labeling is applied automatically — images contain a logo or visual indicator showing that the content was generated using AI.
It is expected that automatic labeling will eventually become the standard for all AI services.
What the Labeling Should Look Like
The law does not establish a strict or unified format for labeling. The main requirement is that it must be clear and understandable to users.
Labeling may take the form of:
• a visual mark such as “Created with AI” on an image or video;
• a text notice placed at the beginning of or before the material;
• an audio message played when media content is reproduced;
• any other format that clearly indicates the content was created using AI.
Special attention will be paid to deepfakes and materials involving images or voices of real people.
What Penalties Are Provided
Administrative liability is introduced for the absence of AI labeling:
• for individuals — 15 MCI;
• for small businesses and non-profit organizations — 20 MCI;
• for medium-sized businesses — 30 MCI;
• for large businesses — 100 MCI.
In the case of a repeated violation within one year, fines increase, and in certain cases the operation of an AI system may be suspended or fully prohibited.
When the Law Takes Effect
The law will come into force 60 calendar days after its official publication, on January 16, 2026.
The requirement for AI content labeling is introduced as part of Kazakhstan’s broader policy on digitalization and regulation of artificial intelligence.