A coalition of civil society organizations, including CDT Europe, has raised concerns over the latest draft of the European Union's Code of Practice for General Purpose AI. In a letter to Executive Vice-President Virkkunen, the group criticizes the draft, stating that important protections for fundamental rights have been reduced to voluntary suggestions. The coalition argues that these changes could weaken the intended framework of the AI Act and reduce accountability for upstream model providers.
According to the coalition, the third draft significantly weakens the Code’s approach to systemic risks by shifting responsibility away from model developers and making important risk categories optional. They claim this contradicts the AI Act’s principles and differs from the global consensus on AI safety. Given the potential impact on privacy and democratic freedoms, the letter calls for immediate changes to ensure that the Code aligns with the AI Act’s goal of safeguarding fundamental rights and ensuring strong protections throughout the AI lifecycle.