Freund, R., Anisic, Z. (2026): Digital Sovereignty and Open-Source AI: The European Way for Innovative SMEs, in Anisic, Z., Forza, C. (Eds): Proceedings of the 12th International Conference on Customization and Personalization, September 17-18, 2026, Balatonfüred, Hungary, pp. 86-95 · 17. Sep. 2026
Abstract: Digital sovereignty is increasingly critical for innovative small and medium-sized enterprises (SMEs) in Europe, particularly as dependence on proprietary cloud AI systems raises concerns around data control, transparency, and cost predictability. This paper examines how European SMEs can design resilient, AI-driven innovation processes while preserving full control over their data and infrastructure. Drawing on a conceptual framework supported by proof-of-concept implementations, we analyse Europe’s rights-based approach to AI governance under the EU AI Act and contrast it with the market-oriented model of the United States and the state-controlled paradigm of China. We then demonstrate, step by step, how a coherent digital sovereignty stack, comprising Nextcloud as a sovereign collaboration platform, LocalAI for on-premise model deployment, European open-source large language models (including Teuken-7B and EuroLLM), open data sources, and open-source AI agents coordinated via the Model Context Protocol (MCP) can be integrated to support all four stages of the SME innovation process. Challenges including hidden implementation costs, AI hallucinations, vendor lock-in, and the need for Explainable AI (XAI) are critically discussed. The paper concludes with a scalable, layered framework enabling European innovative SMEs to achieve AI-driven innovation with full digital sovereignty.
