Artificial intelligence conformity assessment in Ukraine – CEPS

Ukraine’s goal of joining the EU shapes how it plans to regulate artificial intelligence (AI). The EU AI Act makes conformity assessment a key part of AI regulation. This process checks whether AI systems meet the required rules and standards. As part of the EU’s body of laws and rules, it is also a condition linked to Ukraine’s EU membership process and the Ukraine Facility. The Digital Omnibus on AI has delayed the rules for high-risk AI until December 2027 and August 2028. This gives Ukraine an extra 12 to 16 months to prepare its AI rules to match those of the EU.

This report looks at what Ukraine needs to do to make its AI conformity assessment system compatible with the EU’s system, and how far it has come. Under the EU system, most rules for high-risk AI are checked through internal controls carried out by the companies that provide these systems, rather than through certification by an outside organisation. This makes shared standards developed by CEN-CENELEC JTC 21 especially important. Ukraine is already following these standards with the aim of putting them directly into practice. Its plan to prepare voluntarily before introducing legally binding rules is sensible. Its existing systems for technical regulation, accreditation and standardisation also provide a useful foundation for making these changes. However, the main gaps are in institutions and expertise. Ukraine has no independent data-protection authority and no appointed body to oversee AI. The creation of a Single Digital Regulator may be delayed, and there are few experts with the skills needed to audit AI systems. The war and the unclear boundary between civilian and dual-use technology also create special challenges for checking whether AI systems meet the rules.

The report recommends using 2026–27 to make sure the rules can be enforced before passing new laws. It proposes several steps: completing data-protection reform, bringing standards more closely into line with EU standards, building skills in AI auditing, preparing a backup option for AI supervision, checking how ready different sectors are, and organising the work to align with EU rules as a partnership for putting them into practice. This partnership should involve EU bodies, especially as new EU laws and policy plans are expected.