From Guidance to Statute
On June 12, 2026, Ukraine's parliamentary Committee on Digital Transformation convened the first session of a new working group tasked with drafting the country's first standalone AI law. More than 100 people took part — lawmakers, ministry officials, civil society representatives, and Ukrainian and international AI experts — under committee chair Mykhailo Kryachko (golos.com.ua). Then-Acting Minister of Digital Transformation Oleksandr Bornyakov presented the ministry's roadmap, describing a law meant to "synchronize our field with EU rules" so Ukrainian AI developers can "confidently enter international markets" (thedigital.gov.ua).
This is the second half of a plan Ukraine's Ministry of Digital Transformation laid out two years ago. Its June 2024 White Paper on AI regulation explicitly sequenced the approach: "first, we prepare companies for the requirements, and then adopt a law analogous to the European AI Act," in Bornyakov's own words at the time (kmu.gov.ua). The working group's launch marks the pivot from stage one — voluntary codes and a regulatory sandbox — to stage two: binding legislation with the EU AI Act as its template.
The Case for Locking In Now
The strongest argument for moving to hard law is not abstract. Ukraine is a EU candidate country, and legislative alignment with the acquis — including AI rules — is part of what accession requires. A Ukrainian AI law that mirrors the EU AI Act's risk tiers gives domestic developers a single compliance target instead of two, and lets Ukrainian AI products cross into the EU's market without a second regulatory gauntlet. For a wartime economy that needs its tech sector generating export revenue, not re-litigating conformity assessments in Brussels, that's a genuine efficiency gain. Citizens also get a real backstop: voluntary codes bind no one, and a state increasingly running welfare, tax, and defense-adjacent logistics through AI-assisted systems has a legitimate interest in enforceable rules on high-risk deployments, not just goodwill.
What the Sandbox Actually Delivered
The case for caution starts with the fact that the soft-law phase Ukraine is now stepping away from was working. In its first year, Mintsyfry's regulatory sandbox drew 90 applications from companies seeking to test AI products against the ministry's guidance, and 15 solutions passed a full audit conducted by 23 outside experts — no binding statute required (dev.ua). Fourteen companies, including Grammarly, MacPaw, SoftServe, and Uklon, formed Ukraine's first industry body for ethical AI use on their own initiative. And on the Cabinet of Ministers' own accounting, Ukraine climbed 14 places in Oxford Insights' Government AI Readiness Index 2025, reaching 40th of 195 countries assessed (kmu.gov.ua). None of that required a Ukrainian AI Act. It required a sandbox, a white paper, and companies with enough legal certainty to build.
Pegged to a Deadline Brussels Isn't Keeping
The timing complicates the synchronization argument further. Ukraine's working group is drafting toward a target that the EU itself is currently loosening. The AI Act's high-risk system obligations were due to bind on August 2, 2026 — days from this article's publication. But under the "Digital Omnibus" agreement reached provisionally on May 6, 2026, the European Parliament and Council pushed that deadline for standalone high-risk systems to December 2, 2027, and the equivalent deadline for high-risk systems embedded in regulated products to August 2, 2028 — a delay of roughly 16 months, pending formal adoption and Official Journal publication expected "in the coming weeks" (Gibson Dunn). Brussels concluded, in effect, that its own compliance regime for high-risk AI was not ready to bind on schedule. Kyiv is nonetheless building a statute to mirror it in real time — a reasonable bet on where the acquis eventually lands, but a live illustration of the risk in hard-coding domestic law to a moving foreign target rather than to Ukraine's own assessment of what its AI sector needs regulated first.
The Sequencing Question That Matters
None of this makes an eventual Ukrainian AI law wrong. Accession conditionality is real, and a harmonized framework has genuine export value for an IT sector Ukraine badly needs to keep growing through the war. But the working group would do well to import the EU's caution along with its risk categories: phase in high-risk obligations only once compliance infrastructure — conformity assessment bodies, a functioning market-surveillance authority, guidance clear enough for a 20-person startup to follow — actually exists, rather than binding by statutory deadline and hoping capacity catches up. The sandbox proved Ukrainian regulators can build trust with industry without heavy machinery. The law that follows should preserve that, not trade it for a compliance calendar set in Brussels and already being rewritten there.