New Working Paper: Open-Source AI made in the EU: Why it is a Good Idea

I'm excited to share our recent paper that examines the potential and strategic advantages of developing open-source (OS) foundation models within the European Union (EU). The work is a collaborative effort with my colleagues Luciano Floridi, Carlotta Buttaboni, Emmie Hine, Claudio Novelli, Tyler Schroder, and Grant Shanklin. Here's an overview of our findings and recommendations:

Abstract

Our paper explores the feasibility and benefits of developing open-source foundation models within the EU. Given recent global developments in AI, particularly in China, the EU's stringent regulatory framework and commitment to ethical AI development uniquely position it to produce trustworthy open-source foundation models. These models could offer a competitive, ethically aligned alternative to proprietary and foreign models. We provide concrete recommendations to leverage this opportunity through initiatives like AI Factories and EuroHPC infrastructure.

Key Points

1. Feasibility and Strategic Advantages

The EU's strong regulatory framework, including the GDPR and forthcoming AI Act, can ensure that open-source foundation models developed in the EU are both trustworthy and competitive. Developing these models within the EU can reduce dependency on foreign technologies, enhancing economic competitiveness and security.

2. A Third Way in AI Development

EU-developed models could provide a "third way" between US commercial models and Chinese state-influenced alternatives, supporting European SMEs and appealing to global actors seeking alternatives outside the US-China technology duopoly.

3. Policy Recommendations

• Establish an EU-wide OS Foundation Model Governance Framework: Develop clear regulatory guidelines tailored explicitly to open-source foundation models, ensuring transparency and accountability.
• Create a European OS Foundation Model Certification and Benchmarking System: Implement standardized benchmarks and certification processes to assess models for security, reliability, ethical compliance, and performance.
• Expand Dedicated Funding Programs: Provide targeted financial support for research projects, startups, and consortia focused on open-source foundation model development through frameworks like Horizon Europe and the Digital Europe Programme.
• Scale up AI Computing Infrastructure: Accelerate the expansion of high-performance computing facilities to support the training and fine-tuning of open-source foundation models.
• Enhance Cross-Border and Cross-Sector Collaboration: Support EU-wide consortia composed of academia, industry, and public entities to jointly develop and deploy open-source foundation models.
• Launch Public Awareness and Education Initiatives: Educate citizens, businesses, and policymakers about the benefits, risks, and responsible use of open-source foundation models.

Our collective efforts aim to guide the EU to a position as a leader in ethical AI development and leverage open-source models to foster innovation and economic growth across sectors.

Our work not only highlights the strategic importance of developing trustworthy AI within the EU but also provides actionable recommendations to achieve this goal. By promoting open-source AI, we can ensure that the development of AI technologies aligns with ethical principles and serves the broader public interest.

Reference: Floridi, Luciano and Buttaboni, Carlotta and Hine, Emmie and Novelli, Claudio and Schroder, Tyler and Shanklin, Grant, Open-Source AI made in the EU: Why it is a Good Idea (March 24, 2025). Available at SSRN or http://dx.doi.org/10.2139/ssrn.5191372