Kazakhstan introduces mandatory audits for high-risk AI
Kazakhstan has approved rules for compiling lists of trusted high-risk artificial intelligence systems. The corresponding order was signed by Deputy Prime Minister – Minister of Artificial Intelligence and Digital Development Zhaslan Madiyev, Qazinform News Agency reports.
According to the document, sectoral government agencies will compile these lists to enhance trust and promote best practices in the use of artificial intelligence within their respective fields, with the lists to be published on an ongoing basis on official government websites.
The lists will be compiled based on applications submitted by owners of high-risk AI systems. To be included, applicants must provide a formal request, documents confirming intellectual property rights (such as a registration certificate or author’s agreement), along with a positive audit conclusion.
The relevant government authority will review the submitted documents within 10 working days, assessing the completeness of information about the system — including its purpose, functionality, and conditions of use — as well as the availability of the required legal documents and audit report.
If the system meets all requirements, it will be added to the list, with its details published on the official website of the relevant government body within five working days, available for viewing and download.
If any inconsistencies are identified, the applicant will be notified. Once these are addressed, the documents may be resubmitted, with the review period for resubmission set at up to five working days.
Updated versions of the lists will be published on government websites as they are revised.
Earlier, Qazinform reported that top Kazakhstani executives will undergo AI training with professors from Stanford and OpenAI Academy.
