Uzbekistan adopts ethics code for artificial intelligence

2026-03-17

Order No. 284 of the Minister of Digital Technologies, registered on 14 March 2026, establishes mandatory ethics rules for AI development, deployment and use in Uzbekistan. The code enters into force three months after official publication and introduces eight core principles covering human oversight, transparency, non-discrimination and data protection.

What has changed and why it matters

Order No. 284, registered on 14 March 2026, establishes Uzbekistan's first comprehensive AI ethics framework, implementing Presidential Decree No. 189 (October 2025) and the AI Development Strategy to 2030. The code applies to all individuals and legal entities developing, deploying or using AI systems in Uzbekistan.

Eight core principles

• Legality. Compliance with all applicable legislation
• Protection of interests. AI must not harm individuals, society, the state or the environment
• Explainability. AI decisions must be understandable, with logic, criteria and data sources disclosed
• Responsibility and human oversight. All AI decisions must remain under human control
• Fairness and non-discrimination. Prohibition of discrimination based on personal characteristics
• Openness and transparency. AI systems must be accessible and understandable
• Data protection. Protection of state secrets, personal data and restricted information
• Reliability and security. Stable operation with minimal errors and secure data storage

Prohibition on AI-only decisions

Legally significant decisions affecting human rights cannot rely solely on AI conclusions. This extends to healthcare decisions including diagnosis, treatment and medical image analysis.

Key obligations

Developers and deployers must: ensure algorithm transparency, prevent bias, protect data, disclose limitations and risks, inform users outputs are advisory only, maintain human oversight and publish guidelines.

Users must: comply with usage rules, protect data, respect intellectual property and avoid causing harm.

What this means for your business

The three-month implementation window provides time to assess compliance gaps. Companies using AI for automated decision-making, analytics, profiling or HR should review explainability, transparency and non-discrimination standards. The human oversight requirement may require updates to documentation and internal workflows.