HomeEverythingEducation
Equities & FundsCrypto & Digital AssetsAI & TechnologyBusiness & CorporateUS Politics & PolicyGeopolitics & Global RiskMacro, Rates & FXCommodities & EnergyEuropean Politics & MarketsAsia-PacificReal Estate & Property
← All Stories

Ukraine to prioritize self-hosted AI models amid control concerns

Created at 7 Jul · 3:53 PM1 source↑ Market-relevant
IN SHORT

Ukraine will favor AI systems that can be run on its own servers, a senior official said, to avoid dependence on remote models that providers can restrict. This approach prioritizes AI sovereignty and security for government, business, and military applications.

✉Newsletter

PiQ Daily

Pick your topics. Get only what matters, on your cadence.

Who's Involved

Roman Kyslyi
Chief AI Officer at Ukraine's Ministry of Digital Transformation
Ukraine
nation prioritizing AI sovereignty and self-hosted models
Anthropic
AI provider whose models were subject to U.S. government access restrictions
OpenAI
AI provider whose main models are considered remote-only
Google
provider of Gemini and Gemma AI models
Kyivstar
partner in developing Ukraine's new AI model
Mistral
AI model provider considered for Ukraine's use
Ukraine to prioritize self-hosted AI models amid control concerns

↳ Why This Matters

Ukraine's strategic decision to prioritize self-hosted AI models highlights growing global concerns about data sovereignty and the potential for geopolitical influence over critical digital infrastructure, especially in wartime. This policy could influence how other nations approach AI adoption and development.

Key facts

  • Ukraine will prioritize AI systems that can be run on its own servers (on-premise).
  • This policy aims to maintain control over digital tools for government, business, and military use.
  • The decision was influenced by U.S. government actions restricting access to AI models.
  • Ukraine is developing its own AI model based on Google's Gemma open-source variant.
  • The new model is slated for release in autumn and will be used across sectors.

Ukraine intends to prioritize artificial intelligence systems that can be operated on its own servers, a move aimed at ensuring the country's digital tools for government, business, and military purposes are not subject to external control or potential shutdowns. This policy favors self-hosted, or "on-premise," models over solutions that remain under the provider's control, such as the main offerings from Anthropic and OpenAI.

Roman Kyslyi, Chief AI Officer at Ukraine's Ministry of Digital Transformation, stated that this approach was reinforced after the U.S. government ordered Anthropic to restrict access to its powerful AI models. He emphasized that "AI sovereignty isn't just a defensive talking point, it's a necessity." Kyslyi clarified that the origin of the model is not the primary concern; rather, it is the ability to run it on Ukrainian infrastructure without restrictions.

Currently, Ukraine's AI assistant within the Diia government app utilizes Google's Gemini, accessed via servers in the European Union. Kyslyi described this as an "interim" solution, noting that personal data is stripped before queries are sent due to a lack of control over the model. Ukraine is actively developing its own AI model in collaboration with Kyivstar, based on Google's open-source variant, Gemma. This model is expected to be released in the autumn and is intended for deployment across government services, private enterprises, and the military. Kyslyi mentioned that Gemma and Mistral models were evaluated and found to perform comparably to remote-only alternatives on various tests.

Frequently asked questions

Ukraine's decisive criterion is whether the AI model can be run on its own on-premise infrastructure, ensuring it is under Ukrainian control.

The shift is driven by a desire to avoid dependence on systems that providers can restrict or switch off, a concern heightened by recent government actions against AI providers.

Ukraine currently uses Google's Gemini for its Diia app and is developing its own model based on Google's Gemma, with open-source options like Mistral also being considered.

What Happens Next

01Ukraine plans to release its self-developed AI model based on Google's Gemma in autumn.
02Ukraine will continue to evaluate AI models based on their deployability on domestic infrastructure.

Get the newsletter.

Pick the topics you actually care about. We'll email when there's news worth your time, on the cadence you choose. Cancel any time from your account.

Cadence

How It Developed

Ukraine will prioritize AI systems that can be run on its own servers.
This approach aims to avoid dependence on remote systems controlled by providers.
The policy was reinforced by a U.S. government order to Anthropic to cut access to powerful models.
Ukraine's Chief AI Officer stated AI sovereignty is a necessity.
Ukraine is developing its own model with Kyivstar based on Google's Gemma.
The new model is intended for government services, private enterprises, and the military.
Gemma and Mistral models were compared to remote-only alternatives and performed comparably.

Sources

T1
Ukraine to pick AI models operated without provider control, official saysReuters

Related Stories

China Considers Curbing Overseas Access to Advanced AI Models, DeepSeek Develops Own Chip
7 Jul · 10:18 AM
US Cyber Agency Uses Anthropic's Mythos AI to Audit Government Code
6 Jul · 8:12 PM
Darktrace backs Anthropic's AI model pause on security grounds
7 Jul · 11:25 AM
Meituan Open-Sources 1.6T AI Model Trained on Chinese Chips
6 Jul · 8:25 PM
China's DeepSeek plans custom chips amid US export controls
7 Jul · 4:21 PM