Scientific Bulletin of Mukachevo State University. Series “Economics”

Vol. 13, No. 2, 2026 25.05.2026 open access Open access

The economic implications of shadow AI in Georgia’s public sector

Irakli Manvelidze

DOI https://doi.org/10.52566/msu-econ2.2026.56 Pages 56 –67 Views 38 Views

Abstract

The purpose of this study was to systematically investigate the scale, drivers, and economic implications of Shadow AI in Georgia’s public sector, and to propose evidence-informed policy interventions for enhancing AI readiness, workforce productivity, and fiscal risk mitigation. Although over 70% of public servants globally use AI tools, systematic empirical research on unauthorised “Shadow AI” in Georgia is lacking. To address this gap, a convergent mixed-methods design was employed, including a survey of 322 public servants, 15 semi-structured interviews, a three-round Delphi panel with 12 experts, a Ministry of Internal Affairs case study, and systematic document analysis. Quantitative survey data, qualitative interviews, and Delphi panel inputs were integrated to achieve methodological triangulation. Findings showed that 52.8% of public servants use AI tools (17.4% daily), while 42.9% engage in Shadow AI. Institutional readiness is limited: only 14.3% have a formal AI policy, 61.5% received no AI training, and 54% enter sensitive data into AI platforms. Thematic analysis identified five patterns: Silent Innovation, Double Standard, Training Desert, Data Fear, and Lack of Trust. Logistic regression shows that perceived productivity gains (OR = 2.34, p < .01) and absence of formal policy (OR = 3.12, p < .001) are significantly associated with Shadow AI adoption. These findings suggested coexisting productivity gains and unmitigated institutional risks, including inefficiencies from unmanaged AI use, uneven human capital development, and fiscal exposure from data breaches or algorithmic errors. The study suggested a contextual adaptation of the Unified Theory of Acceptance and Use of Technology (UTAUT), providing evidence that in transition economies perceived usefulness may outweigh absent facilitating conditions. For policymakers and public administrators, the research proposes four actionable interventions: mandatory AI literacy programs, a regulatory sandbox for controlled experimentation, AI Champions networks for peer support, and revised performance evaluation systems. These recommendations are designed to enhance economic efficiency, mitigate fiscal risks, and promote responsible governance in Georgia and comparable post-Soviet transition economies

Keywords

digital transformation; generative AI; institutional readiness; data privacy; UTAUT model; administrative efficiency; transition economies

References

  1. Aryfiyanto, H., & Alamsyah, A. (2024). Public service with generative AI: Exploring features and applications. In 2024 7th international conference of computer and informatics engineering (IC2IE) (pp. 1-7). Bali: Institute of Electrical and Electronics Engineers. doi: 10.1109/IC2IE63342.2024.10747963.
  2. Barcelona Provincial Council. (2025). Public personnel positively value the use of AI in their day-to-day work. Retrieved from https://www.aoc.cat.
  3. Braun, V., & Clarke, V. (2006). Using thematic analysis in psychology. Qualitative Research in Psychology, 3(2), 77-101. doi: 10.1191/1478088706qp063oa.
  4. Bryman, A. (2016). Social research methods (5th ed.). Oxford: Oxford University Press.
  5. Cantens, T. (2025). How will the state think with ChatGPT? The challenges of generative artificial intelligence for public administrations. AI & Society, 40, 133-144. doi: 10.1007/s00146-023-01840-9.
  6. Creswell, J.W. (2014). Research design: Qualitative, quantitative, and mixed methods approaches (4th ed.). Thousand Oaks, CA: SAGE Publications.
  7. Davis, F.D. (1989). Perceived usefulness, perceived ease of use, and user acceptance of information technology. MIS Quarterly, 13(3), 319-340. doi: 10.2307/249008.
  8. Dunleavy, P., Margetts, H., Bastow, S., & Tinkler, J. (2006). Digital era governance: IT corporations, the state, and e-government. Oxford: Oxford University Press. doi: 10.1093/acprof:oso/9780199296194.001.0001.
  9. European Commission. (2025). Georgia 2025 report (SWD(2025) 757 final). Retrieved from https://www.eeas.europa.eu.
  10. Farrell, R. (2025). The silent risk in government AI: Shadow models are already at work. Retrieved from https://www.govloop.com.
  11. Field, A. (2018). Discovering statistics using IBM SPSS statistics (5th ed.). London: SAGE Publications.
  12. Geostat. (2025). Average monthly earnings of employees by economic activity. Retrieved from https://www.geostat.ge.
  13. Government of Georgia. (2024). Digital governance strategy of Georgia 2025-2030. Retrieved from https://matsne.gov.ge.
  14. Hall, I. (2025). GenAI on the rise: UK public sector’s AI appetite and attitudes revealed in survey. Retrieved from https://www.globalgovernmentforum.com.
  15. IDFI. (2021). Use of artificial intelligence systems in Georgia: Legislation and practice. Retrieved from https://idfi.ge.
  16. IDFI. (2024a). EU AI Act. Retrieved from https://idfi.ge.
  17. IDFI. (2024b). Council of Europe convention on AI. Retrieved from https://idfi.ge.
  18. IDFI. (2024c). Global index on responsible artificial intelligence and Georgia’s outcomes. Retrieved from https://idfi.ge.
  19. Kopper, A., & Westner, M. (2016). Towards a taxonomy for shadow IT. In Proceedings of the 22nd Americas conference on information systems (AMCIS 2016) (pp. 1-10). San-Diego: Association for Information Systems.
  20. Lāma, G., & Lastovska, A. (2025). AI competence and sentiment: A mixed-methods study of attitudes and open-ended reflections. Frontiers in Artificial Intelligence, 8, article number 1658791. doi: 10.3389/frai.2025.1658791.
  21. Microsoft. (2025). AI diffusion report 2025-H2. Retrieved from https://www.microsoft.com.
  22. Mission Square Research Institute. (2025). AI adoption in state and local government: What drives success and employee confidence? Retrieved from https://icma.org.
  23. OECD. (2026). Building an AI-ready public workforce. Paris: OECD Publishing. doi: 10.1787/b89244c7-en.
  24. Oxford Insights. (2025). Government AI readiness index 2025. Retrieved from https://oxfordinsights.com.
  25. Public Defender of Georgia. (2025). AI and human rights. Retrieved from https://www.ombudsman.ge.
  26. Rogers, E.M. (2003). Diffusion of innovations (5th ed.). New York: Free Press.
  27. Sampson, C. (2025). Ukraine AI nation initiative. Retrieved from https://www.kyivpost.com/post/63986.
  28. Silic, M., & Back, A. (2014). Shadow IT – A view from behind the curtain. Computers & Security, 45, 274-283. doi: 10.1016/j.cose.2014.06.007.
  29. Silic, M., Silic, D., & Kind-Trüller, K. (2025). From Shadow IT to Shadow AI: Threats, risks and opportunities for organizations. Strategic Change, 34(5), 291-304. doi: 10.1002/jsc.2682.
  30. State of Delaware News. (2025). Delaware launches AI sandbox initiative. Retrieved from https://news.delaware.gov.
  31. Tangi, L., Rodriguez Müller, A.P., & Combetto, M. (2026). A silent partner: The shadow presence of generative artificial intelligence in public administrations. In S. Hofmann et al. (Eds.) Lecture notes in computer science (Vol. 15978). Cham: Springer. doi: 10.1007/978-3-032-02515-9_5.
  32. Tskhvariashvili, E. (2025). Artificial intelligence and public governance: Innovative approaches in health policy in the context of gender equality and social insurance. Vectors of Social Sciences, 10. doi: 10.51895/VSS.
  33. UNESCO. (2026). Key insights: Georgia – artificial intelligence governance. Retrieved from https://www.unesco.org/ethics-ai/en.
  34. Venkatesh, V., Morris, M.G., Davis, G.B., & Davis, F.D. (2003). User acceptance of IT. MIS Quarterly, 27(3), 425-478. doi: 10.2307/30036540.
  35. Voog, M. (2025). Kallas at the Tallinn Digital Summit: Estonia’s true leap is to learn to think with artificial intelligence, not instead of it. Retrieved from https://hm.ee.
  36. Waters-Lynch, J., Allen, D.W.E., Potts, J., & Berg, C. (2024). Managing generative AI in firms: The theory of shadow user innovation. SSRN. doi: 10.2139/ssrn.4754950.
  37. World Medical Association. (1964). Declaration of Helsinki. Retrieved from https://www.wma.net.

Suggested citation

Manvelidze, I. (2026). The economic implications of shadow AI in Georgia’s public sector. Scientific Bulletin of Mukachevo State University. Series “Economics”, 13(2), 56-67. https://doi.org/10.52566/msu-econ2.2026.56