Kazakhstan. AI / automated decisions. Law

AI LawLaw of the Republic of Kazakhstan No. 230-VIII 'On Artificial Intelligence'

In force

Partly verified

Medium confidence. Official adilet text URL not captured.

§1Summary

Synopsis

Risk-based AI framework requiring owners to classify systems as minimum, medium or high risk, manage risk over the lifecycle and keep documentation. It bans manipulative techniques, social scoring, non-consensual emotion recognition and discriminatory biometric categorisation, and requires labelling of synthetic content and disclosure of AI interactions.

Key themes

risk classification
prohibited practices
transparency
deepfake labelling
user rights over AI decisions
copyright and TDM opt-out

Unique features

One of the first comprehensive AI statutes in Central Asia; machine-readable opt-out for training on copyrighted works.

SaaS relevance

SaaS offering AI features to Kazakh users must label synthetic outputs, disclose AI interaction and honour users' rights to explanations and correction.

§2Obligations

Breach notification: None
DPO: No
Cross-border transfers: None specific
Maximum penalty: Administrative fines in MCI units (amounts not confirmed)

§3Dates

Key dates

Adopted 17 November 2025
In force 18 January 2026
Signed 2025-11-17; in force 2026-01-18 with accompanying Administrative Offences Code amendments.

§4Text and authority

Text

Authoritative text adilet.zan.kz

Authority

Ministry of Artificial Intelligence and Digital Development

§5Where it applies

Domestic law of: Kazakhstan

§6Sources and verification

Sources consulted

Verification

Medium confidence.

Partly verified

Medium confidence. Official adilet text URL not captured.