Fourteen major organizations sign joint letter warning the EU on the Second Draft of the General Purpose AI Code of Conduct

• EU Risks Creating a ‘Bureaucratic Monster’: Fourteen Major Organizations Warn Against AI Code Draft

Leading organizations from the AI, startup, and SME sectors, collectively representing dozens of member companies, associations, and industry groups, have signed a joint open letter raising serious concerns over the Second Draft of the General Purpose AI (GPAI) Code of Conduct. The GPAI Code serves as a key framework for general-purpose AI model providers to demonstrate compliance with the AI Act throughout the entire lifecycle of their models. The signatories warn that, in its current form, the Code could stifle AI innovation and weaken Europe’s global competitiveness. Organizations and industry stakeholders are encouraged to join and add their support by signing the letter.

The statement, led by the AI Chamber, warns that the current draft imposes excessive and impractical obligations. These include overly complex regulations, unfeasible compliance measures, controversial copyright provisions, excessive disclosure requirements that threaten public security and trade secrets, and additional burdens on AI model providers that go beyond the AI Act. 

Previously, the AI Chamber – a newly established organization dedicated to advancing AI innovation in the CEE region – submitted an official statement on behalf of its members to the EU AI Office, outlining critical recommendations for revisions to the second draft of the GPAI Code of Conduct. With the final version expected to be completed by April 2025, and the AI Act set to take effect in August 2025, time for necessary improvements is running out. 

“The business impact of these requirements cannot be understated. By creating competitive disadvantages for EU companies and imposing particularly burdensome requirements on startups and SMEs, the Code risks discouraging AI development in Europe altogether. The time to act is now.” warns Tomasz Snażyk, CEO of AI Chamber.

EU bureaucracy vs. global competitiveness

The AI Chamber highlights that the second draft of the Code is excessively detailed, creating additional obligations that go beyond what is required by the AI Act. The regulatory burden threatens to stifle innovation, particularly for small and medium-sized enterprises (SMEs) and startups, which may find compliance costs prohibitively high.

As the EU struggles to maintain its position in the global AI race – particularly against the U.S. and China – excessive regulation risks driving AI development out of Europe. The AI Chamber emphasizes that the GPAI Code of Conduct should facilitate compliance with the AI Act, ensuring that Europe remains an attractive hub for AI development rather than burdening companies with red tape.

“Instead of fostering a dynamic and competitive AI ecosystem, the EU risks turning the Code into a ‘bureaucratic monster.’ While some improvements have been made compared to earlier drafts, the current version still fails to achieve its primary goal. From the perspective of European businesses, the added value of the Code is minimal, while its financial and administrative burdens could suffocate innovation” says Snażyk.

Excessive regulations are killing innovation

The draft Code introduces overly detailed Key Performance Indicators (KPIs) that are vague and difficult to implement. Documentation requirements are extensive but lack clear justification, and demands for detailed disclosure of AI model parameters and training data could compromise trade secrets and intellectual property rights, while also posing risks to public security.

Although transparency is a key goal, the level of disclosure required – including training, testing, and validation datasets, computational resources, and energy consumption reporting – raises concerns over feasibility and competitiveness.

GPAI Code of Conduct vs. AI Act and Copyright Law

The AI Chamber warns that the draft Code introduces compliance obligations that are not foreseen in the AI Act, particularly regarding third-party dataset copyright requirements. The mandate that internal copyright policies must be “aligned with the Code’s commitments” is problematic, as the Code should not interfere with independent legal assessments of EU copyright law.

Additionally, requiring AI model developers to take “reasonable steps” to evaluate third-party dataset compliance oversteps existing EU legal frameworks. The broad and impractical requirement to apply policies at every stage of model development and monitor third-party rights creates legal uncertainty and unnecessary trade restrictions.

AI model size does not define risk

The AI Chamber argues that AI model size or computational power is not a reliable indicator of actual risk. While the Chamber supports exemptions for SMEs – as they often lack the resources to fully comply – the current approach may create a regulatory loophole where AI models from smaller firms operate under lower safety standards. To truly ensure AI safety, the AI Act should focus on the real-world impact and risk potential of AI models, rather than imposing rigid size-based restrictions.

A Code built on undefined standards

The Code of Conduct is based on future AI standards that have yet to be defined by the EU AI Office, creating uncertainty about actual compliance requirements. This ambiguity hinders planning and investment in AI model development, ultimately slowing European innovation.

“The guidelines remain overly burdensome and prescriptive, imposing additional, unnecessary obligations on AI system providers. This makes the Code impractical. New AI startups, facing the choice of whether to build their own models, may simply decide it’s not worth the effort. As a result, Europe risks missing out on the potential of AI” emphasizes Tomasz Snażyk.

Official Letter Sent to European Commission

On 14 February 2025, an official letter was sent to Ms. Henna Maria Virkkunen, Executive Vice President of the European Commission, regarding concerns over the Second Draft of the General Purpose AI Code of Conduct.

The letter, sent on behalf of AI Chamber CEE and co-signed by multiple industry organizations, raises critical issues related to regulatory overreach, excessive compliance burdens, and the potential negative impact on AI innovation in Europe. Additionally, the AI Chamber has requested a formal meeting with Ms. Virkkunen to discuss the concerns and collaborate on more effective regulatory solutions.

Signatories of the Open Letter

Co-signatories:

1. Tomasz Snażyk – CEO – AI Chamber

2. Csongor Bias – MD – Startup Hungary

3. Theodor Panayotov – CIO – DNA Bulgaria

4. Europuls – Center of European Expertise

5. Michał Kanownik – CEO – Związek Cyfrowa Polska

6. Michal Kardoš – Executive Director – Slovak Alliance for Innovation Economy (SAPIE)

7. Olga Kyryliuk (on behalf of SEEDIG Executive Committee) – Chair – South Eastern European Dialogue on Internet Governance (SEEDIG)

8. Przemysław Kuśmierek – CEO – Migam

9. Maciej Witucki – President – Polish Confederation Lewiatan

10. Corina Vasile – Executive Director – Employers’ Association of the Software
    and Services Industry (ANIS Romania)

11. Jakub Bińkowski – Członek Zarządu – ZPP

12. Diana Popp, co-founder, Vice-President, Smart Everything Everywhere

13. Milena Jabůrková – Vice-President – Confederation of Industry of the Czech Republic

14. Lukáš Kačena – Director – Czech National AI platform (Prg.Ai)

Read the Full Open Letter

To learn more about the concerns raised by leading AI, startup, and industry organizations regarding the Second Draft of the General Purpose AI Code of Conduct, you can read the full open letter addressed to the EU AI Office.

📄 Open Letter on the Second Draft Code of Practice

This letter outlines the key regulatory challenges, implementation concerns, and potential impact on innovation in Europe, calling for urgent revisions to ensure a balanced and effective AI framework.