Dmytro YEFREMOV, Ph.D. in Economics, the Board Member of the Ukrainian Association of Sinologists

Acknowledgements

This policy comment is based on ideas and perspectives expressed during the roundtable discussion by Kaiser Kuo (moderator, Sinica Podcast), Dmytro Yefremov (Board Member, Ukrainian Association of Sinologists), Karman Lucero (Associate Research Scholar and Senior Fellow, Paul Tsai China Center, Yale University), Guan Wang (Chairman, Learnable.ai, China)

Executive summary

• China and the U.S. offer contrasting AI governance models with lessons for Ukraine. China emphasizes scale and integration (e.g., smart cities, AI hotlines), while the U.S. prioritizes regulatory oversight and citizen accountability. Ukraine’s bottom-up strategy—led by the Ministry of Digital Transformation—focuses on adapting international best practices for wartime governance and EU integration.
• AI can deliver quick wins in education, bureaucracy, and healthcare, but requires ethical safeguards. Ukraine should apply AI to standardized testing, citizen service platforms, and remote diagnostics, while ensuring transparency, human oversight, and protection against algorithmic bias.
• Geopolitical and regulatory risks must shape AI procurement. While Chinese AI tools are cost-effective, concerns over data security, ideological influence, EU compatibility require independent audits, supplier diversification, and legal alignment with EU standards to maintain sovereignty and public trust.

1. Public Sector AI Implementation Insights from China and the U.S. and Their Relevance for Ukraine.

The Issue. Artificial intelligence is transforming public administration worldwide, with China and the United States leading many of the most visible applications. Both countries have implemented AI to improve efficiency, streamline government services, and enhance transparency, though they do so in different institutional and regulatory contexts. In China, large-scale smart city programs and citizen service platforms have demonstrated remarkable success, while in the United States, AI has been integrated into social services and legislative analysis, though with mixed results. For Ukraine, currently focused on both wartime governance and post-war reconstruction, understanding the key achievements and lessons from these two models is critical. The challenge is to identify best practices that can be adapted to Ukraine’s institutional context, while considering regulatory alignment with the European Union and managing geopolitical risks tied to technology procurement.

Analysis. Ukraine’s government sees AI as a strategic tool to modernize the state, lower costs, and provide more citizen-oriented services despite resource constraints, as Dmytro Yefremov emphasized. The Ministry of Digital Transformation is spearheading these efforts with a deliberately bottom-up approach, encouraging dialogue between businesses, civil society, and government. This participatory strategy allows experimentation before strict regulatory frameworks are imposed. Ukraine has issued AI White Paper, which envisions the country as a “digital nation,” with ambitions to adopt innovative solutions faster than many peers. Importantly, Ukraine’s government approach is to selectively learn from global leaders like China and the United States, especially in areas such as smart platforms, predictive analytics, and automated administrative procedures. Ukraine’s aim is not merely to copy but to adapt proven solutions to its unique environment.

Wang Guan, chairman of Learnable.ai, provided concrete examples of how China’s AI deployments have improved public administration. The first is the 12345-citizen hotline, a widely used call-in service for reporting urban problems and requesting assistance. Previously, call centers required significant manpower and often duplicated efforts. By deploying AI to triage and route calls intelligently—and answer straightforward queries automatically—Chinese municipalities dramatically reduced costs and improved response times. This has clear relevance for Ukraine, where similar citizen-service platforms could enhance responsiveness without overburdening limited administrative staff, especially in post-war recovery.

Second example concerns AI-assisted grading of China’s high-stakes zhongkao and gaokao exams, where AI now handles billions of test items with greater speed and accuracy than humans. While education contexts differ, the broader lesson is that AI excels in highly repetitive, high-volume administrative tasks. Ukraine could apply similar principles to processes like digital service delivery, tax filings, or benefits administration. Both examples demonstrate how AI, when properly tested and audited, can build public trust by delivering tangible improvements in fairness, speed, and cost-efficiency.

Notably, Wang stressed that trust did not come automatically. Initially, Chinese education officials were skeptical of AI grading until rigorous pilot programs demonstrated accuracy and fairness superior to humans. This iterative, evidence-based approach offers a valuable lesson for Ukraine: AI adoption should proceed gradually, with transparency and opportunities for independent verification to foster public confidence.

From a governance perspective, Karman Lucero underscored that institutional context determines the feasibility and desirability of AI applications. The U.S. and China differ fundamentally in regulatory systems: China is centralized and hierarchical, while the U.S. is federal with a strong role for courts and decentralized innovation at the state level. Lucero argued that Ukraine must tailor its regulatory approach to its own political and legal institutions, as copying wholesale from either China or the U.S. would be ineffective.

This serves as an important cautionary point: AI can enhance state capacity when implemented thoughtfully, but it can also undermine it if mismanaged. Lucero cited the example of the Trump administration’s Department of Government Efficiency, where replacing experienced staff with AI-savvy coders failed to deliver promised cost savings and reduced agency effectiveness. The lesson is that AI should augment institutional knowledge, not replace it. Ukraine’s Ministry of Digital Transformation appears aware of this, as it declared a collaborative code-of-conduct approach with business and civil society prior to codifying regulations.

Attention was also drawn to public–private partnerships as a key feature of both Chinese and American AI ecosystems. Government agencies often lack the capacity to develop AI independently and rely on private companies for technology and expertise. This blurs traditional distinctions between public and private actors, raising regulatory challenges in areas like accountability, data governance, and privacy. For Ukraine, this implies the need to establish clear procurement guidelines and oversight mechanisms for AI vendors, whether domestic or foreign.

China’s successes, such as Hangzhou’s “City Brain,” show the potential of integrating AI with urban infrastructure to optimize traffic flows and resource use. However, as Lucero noted, not all Chinese AI deployments are equally effective; some local governments adopt AI superficially without the necessary infrastructure or expertise, leading to inefficiency. This suggests Ukraine should prioritize areas where it has sufficient data quality, institutional capacity, and clear performance metrics.

On the U.S. side, AI deployments tend to focus on enhancing citizen access and analysis, for example through AI-assisted benefits processing or legislative research. While results have been uneven, the U.S. offers lessons in ethical oversight, due process, and transparency. The U.S. courts and advocacy groups play an important role in ensuring AI use does not violate rights, an important consideration for Ukraine given its EU integration trajectory. Ukraine will ultimately need to balance flexibility – using lighter-touch regulation now to encourage innovation, with future compliance with the EU’s AI Act.

Geopolitical factors cannot be ignored. As Kaiser Kuo alluded, U.S.-China rivalry increasingly shapes global AI supply chains. Ukraine will need to navigate carefully when sourcing technology or expertise from China, ensuring it aligns with national security and EU obligations. Selective adaptation, rather than wholesale adoption, is thus critical.

Finally, cultural factors influence public acceptance of AI. It was observed that many Chinese citizens are open to AI because their material well-being has improved alongside technological adoption over decades. In Western contexts, there is often greater skepticism, especially regarding automation replacing human workers. Ukraine should anticipate both enthusiasm and resistance, and manage expectations through transparent communication about AI’s role as a tool to improve services, not as a panacea.

2. AI in Public Sector Reform: Institutional, Regulatory, and Ethical Pathways for Ukraine

The Issue. The adoption of artificial intelligence in public administration has significant potential to improve the effectiveness and transparency of governance. For Ukraine, a country focused on digital transformation while simultaneously managing wartime and post-war recovery needs, AI solutions could provide critical support in modernizing public services and strengthening state capacity. China and the United States offer different models of AI use: China emphasizes scale and integration across sectors, while the U.S. approach highlights regulatory safeguards and citizen accountability. Ukraine is exploring how to adapt lessons from both, particularly from China, while aligning with future European Union regulations. The main challenge is balancing rapid technological adoption with ethical, legal, and institutional safeguards that foster public trust.

Analysis. Ukraine’s government already treats AI as a strategic priority. As Dmytro Yefremov explained, the Ministry of Digital Transformation has adopted a bottom-up approach: engaging businesses and expert communities to experiment with AI applications before codifying strict regulatory frameworks. Ukraine’s experience with military applications of AI, such as autonomous drone targeting under electronic warfare conditions, illustrates both the promise and the ethical challenges of AI. These lessons must inform civilian AI use, especially with attention to European human rights standards. This dual-use experience creates opportunities for civilian spillovers, such as applying battlefield-tested algorithms to crisis management and cybersecurity in public administration.

Three immediate “quick win” opportunities for AI adoption in Ukraine’s public sector could be identified. First, education: Ukraine is already digitalizing university entrance exams. Drawing on Chinese practices highlighted by Wang Guan, such as AI-assisted grading for Zhongkao and Gaokao exams, AI could improve fairness, speed, and cost-effectiveness in Ukraine’s admissions processes. Automating grading of standardized exams and using AI for proctoring could reduce administrative burdens and increase transparency.

Second, bureaucracy reduction: AI tools can substitute for routine bureaucratic tasks at the ground level. Automating simple administrative workflows can reduce corruption and increase efficiency. China’s 12345-citizen hotline, described by Wang, demonstrates how AI can triage citizen complaints and route them effectively, reducing the workload of human staff and improving response times. A similar citizen-service platform in Ukraine could offer both immediate efficiency gains and increased public trust.

Third, healthcare: AI-enabled diagnostic tools and telemedicine solutions have potential in war-affected or remote regions. While Ukraine’s use of AI in healthcare is nascent, experience from other countries shows strong potential for applications such as medical image analysis, logistics optimization for medical supplies, and predictive analytics for epidemic control.

However, opportunities come with significant challenges, particularly when adapting Chinese AI practices. Wang emphasized that cross-border AI deployment faces classic localization hurdles familiar from earlier internet and mobile technology eras: language, legal compliance, and data governance. Even if AI platforms can help solve language barriers through automated translation, applications must be localized culturally and contextually. Regulations on data privacy and use vary by jurisdiction, and Ukraine’s future EU membership requires alignment with General Data Protection Regulation and the EU’s upcoming AI Act. Companies like Learnable.ai invest heavily in legal expertise and local partnerships to navigate these issues.

Institutional culture is another challenge. Wang highlighted that success depends on understanding the needs and expectations of end users in the target country. AI solutions cannot simply be transplanted; they require co-design with local institutions and communities. For Ukraine, this implies the need to foster domestic AI capacity and ensure foreign solutions are integrated thoughtfully, avoiding over-reliance on external vendors.

Legal and ethical safeguards are crucial. Lucero warned that governments eager to cut costs with AI sometimes misapply the technology, leading to backlash. The example from Michigan, where an AI system wrongly flagged thousands of unemployment benefit applicants as fraudulent without meaningful recourse, underscores the importance of due process. Such failures not only harm citizens but also erode trust and result in costly litigation. The lesson for Ukraine is that AI should augment human decision-making, not replace it entirely. Smart people using AI to enhance their work perform better than AI used as a blunt substitute for experienced personnel.

Lucero also stressed that AI in governance is a techno-social system. Technical safety measures like red teaming—stress-testing systems to identify vulnerabilities—are necessary but insufficient. Governments must maintain clear communication with citizens about how AI decisions are made and ensure appeal mechanisms are available. The relationship between government and citizens does not fundamentally change with AI; what changes is the toolset. If AI obscures accountability or denies people the ability to question decisions, it will undermine rather than enhance public trust. Thus, transparency and explanation mechanisms must be built into any AI-enabled public service.

Another challenge relates to public-private partnerships, which are indispensable since government agencies often lack the capacity to develop AI solutions independently. This blurs the line between public and private actors, complicating regulatory oversight. Ukraine must develop procurement standards and accountability frameworks to manage risks, especially when engaging Chinese AI companies amid geopolitical sensitivities. Aligning with EU regulatory norms while selectively adopting Chinese technical solutions could allow Ukraine to balance innovation with security and compliance.

Finally, ethical concerns loom large. Ukraine’s experience with AI-guided military drones raises serious questions about human oversight and unintended harm. While this is a military context, similar ethical concerns exist in civilian administration: algorithmic bias, discrimination, and potential misuse of personal data. Drawing on EU human rights standards and global best practices is essential to prevent harm.

3. Evaluating Chinese AI Technologies in the Context of Ukraine’s European Integration

The Issue. As artificial intelligence becomes integral to governance, countries must carefully assess not only its technical benefits but also its geopolitical implications. Chinese AI solutions are often cost-effective, scalable, and technologically advanced, making them attractive for governments with limited resources. However, concerns persist about data sovereignty, technological dependency, and potential political leverage by the Chinese state. Ukraine, pursuing EU integration while rebuilding after war, faces heightened sensitivity regarding Chinese technology, given Beijing’s perceived alignment with Russia. This context raises fundamental questions: what strategic and geopolitical risks should be considered when evaluating Chinese AI systems, and how can they be mitigated while leveraging technological opportunities?

Analysis. A key strategic driver for Ukraine is its European integration agenda. Yefremov emphasized that Ukraine will have to follow the European Union’s regulatory frameworks on AI, including strict data protection, cybersecurity, and ethical standards. EU partners regularly raise questions about data storage, model training, and the potential for foreign interference. Many Chinese AI systems legally allow government access to data stored in China. For Ukraine or any EU-oriented country, this creates a risk that sensitive personal or strategic data might be exposed to a foreign government. Training AI models on domestic datasets could indirectly leak strategic information. Even if unintentional, such risks must be part of any procurement assessment.

Another concern is ideological and informational integrity. There is evidence that some Chinese platforms embed state-driven narratives or filtering aligned with official positions. If such models were deployed in sensitive contexts — for example, education or public communications — they could subtly influence public discourse or introduce biases that align poorly with democratic norms. This risk is magnified in Ukraine’s case, where public opinion of China is already negative due to Beijing’s perceived support for Russia. Introducing Chinese AI without proper safeguards could undermine trust in government services and AI tools themselves.

At the same time, cost and performance cannot be ignored. As Kaiser Kuo pointed out, there is a familiar dilemma reminiscent of the Huawei network debates: Chinese solutions often deliver high performance at low cost, attractive for countries with limited budgets. Wang highlighted that many fears about Chinese AI are based on misperceptions. From his perspective as an AI developer, core technologies (transformer models) are globally similar, and open-source ecosystems allow countries to customize and audit models. Wang stressed that AI “doesn’t have a nationality” — what matters most is the quality of service, user experience, and affordability. Governments should test multiple solutions and judge them by technical merit and transparency, not solely by country of origin.

However, as Lucero cautioned, openness should be tempered with security awareness. In an increasingly adversarial geopolitical environment, surveillance and espionage tools — hardware or software — are becoming more sophisticated. Even if core AI technology is similar worldwide, the governance environment matters. Chinese firms operate under a legal system that grants authorities broad access to company-held data. Thus, foreign governments must conduct due diligence not only on the code but also on the legal and institutional context of providers. The best way to build public trust is transparency: explain how algorithms make decisions, provide recourse mechanisms for citizens, and maintain accountability for outcomes. If a government adopts foreign AI, there must be independent auditing and clear legal frameworks governing data and model use.

There are also strategic considerations around technological dependency. Geopolitical trends will likely result in a bifurcated AI landscape: inexpensive Chinese solutions dominating much of the Global South, while American and EU firms dominate in Europe. For Ukraine, over-reliance on any one supplier — Chinese, American, or otherwise — risks lock-in and reduces future policy flexibility. Diversification of suppliers, domestic capacity-building, and alignment with EU standards can mitigate these risks.

Another important dimension is public communication and trust. Citizens must be confident that AI in governance serves them, not foreign interests. This involves not only technical safeguards but also active engagement with the public about how AI works and why certain suppliers are chosen. Negative perceptions of China in Ukraine heighten the importance of transparency. Public consultations and pilot programs with rigorous oversight could help build confidence before large-scale deployment.

Wang’s comments also point to a practical pathway: treat AI like other foundational technologies (electricity, internet) and focus on user needs. Many AI components are open-source; governments can fine-tune models locally, mitigating risks of foreign control while benefiting from global innovation. Open-source solutions also allow third-party audits, reducing the “black box” problem. However, as Lucero stressed, even with open models, security vetting and institutional safeguards remain necessary.

Finally, there is the broader geopolitical signaling dimension. The choice of AI suppliers sends signals about a country’s alliances and values. For Ukraine, whose future lies in the EU, adopting Chinese AI for core governance functions without strong safeguards could create friction with European partners. Conversely, building domestic AI capacity and collaborating with EU and trusted foreign partners signals strategic alignment while still allowing selective engagement with Chinese firms for non-sensitive applications where risks are low and benefits clear.

Policy Recommendations

• Countries evaluating Chinese AI for governance should adopt a diversified strategy that combines rigorous security and data safeguards, independent auditing, and public transparency, while leveraging cost-effective foreign solutions only in low-risk areas and simultaneously investing in domestic AI capacity to avoid strategic dependency.
• Ukraine could implement a phased, evidence-based AI strategy that draws selectively on international best practices, focusing first on practical applications in areas like education, public services, and healthcare.
• Ukraine should prioritize the creation of robust legal and regulatory safeguards, establish effective public–private partnerships, and align all AI initiatives with EU standards to strengthen transparency, build citizen trust, and ensure the long-term resilience of its public institutions.

This policy comment is the result of the first open discussion initiated by the Center for Slavic, Eurasian, and East European Studies (CSEEES) at the University of North Carolina at Chapel Hill, held on the Ukrainian Platform for Contemporary China. The event was co-organized in partnership with the Sinica Podcast, the National Institute for Strategic Studies (Ukraine), the Ukrainian Association of Sinologists, and the A. Krymskyi Institute of Oriental Studies at the National Academy of Sciences of Ukraine.