EU Launches Consultation on AI Act Rules for GPAI Providers
The European Union has initiated a consultation on rules for general-purpose AI models (GPAIs) under the AI Act, targeting providers like Anthropic, Google, Microsoft, and OpenAI. The AI Act, adopted earlier this year and effective from August 1, 2024, includes phased compliance deadlines, with Codes of Practice set to apply by April 2025. The consultation aims to guide developers on legal compliance to ensure “trustworthy” GPAIs.
The Commission is seeking input from GPAI providers operating in the EU, businesses, civil society representatives, rights holders, and academic experts. The consultation, divided into three sections, covers transparency, copyright provisions, risk taxonomy, assessment, and mitigation for GPAIs with systemic risk, and monitoring Codes of Practice. Responses will shape the template for summarizing model training content.
Eligible stakeholders are invited to participate in drafting the Code via virtual meetings in four working groups. The iterative drafting process will be transparent, with meeting minutes available to all participants. The AI Office is also calling for expressions of interest for key steering roles, including meeting chairs and vice chairs, from independent experts.
Concerns have been raised about potential exclusion of civil society organizations from the drafting process. The Commission assures broad participation, inviting submissions from academia, industry representatives, civil society organizations, rights holders, and public authorities. The deadline for consultation submissions is September 10, 2024, and for expressions of interest in drafting, it is August 25, 2024.
