Brussels, BELGIUM – The second edition of the European AI Roundtable successfully concluded critical discussions about the drafting of the Code of Practice for general-purpose AI (GPAI), a vital step in the implementation of the EU’s landmark Artificial Intelligence (AI) Act.
As developers of GPAI – such as large language models that underpin many popular AI applications – prepare to comply with the Act by August 2025, they have to trust on this Code for the necessary clarity and guidance that the high-level political deal never provided.
The AI Roundtable – facilitated by the Computer & Communications Industry Association (CCIA Europe) – gathered experts from leading AI firms (extending well beyond CCIA’s membership), academia, governments, civil society, and EU institutions to debate the challenges and opportunities in operationalising critical aspects of Europe’s new AI rules through a Code of Practice.
Yann Padova, legal expert and co-author of a study on getting the Code right, identified four key risks that could undermine the process if not addressed swiftly.
Padova first noted that reconciling the views of almost 1,000 stakeholders will be a major challenge, and that the expertise of GPAI developers – being the primary subjects of the rules – must take priority to ensure the final Code is workable.
However, with GPAI providers only representing 5% of all participants admitted by the EU AI Office to the drafting process, this already risks contradicting the AI Act. Indeed, the Act specifies that GPAI providers should “participate in the drawing-up” of the Codes, while other stakeholders only “may support” the process according to the letter of the law.
The drafting process also risks being distracted by discussions outside the Act’s scope, such as copyright law and internal risk management. Padova emphasised that off-topic discussions should be avoided at all costs in order not to dilute or delay the process.
Likewise, Padova raised concerns about possible compliance requirements, such as third-party audits, that were mentioned in recent AI Office consultations despite never being required by the AI Act. While administrative support might be helpful, the legal expert also cautioned against outsourcing any of the AI Office’s core responsibilities to external consultants.
Yesterday’s edition of the European AI Roundtable marked a pivotal moment in the drafting process, as the Roundtable not only aims to ensure that the GPAI Code of Practice operationalises key parts of the AI Act, but also fosters a thriving AI ecosystem in Europe.
The next gathering of the European AI Roundtable is scheduled to take place before the end of the year, and will focus on the interplay between AI, privacy, and data protection laws.
The following can be attributed to CCIA Europe’s Senior Policy Manager, Boniface de Champris:
“We’re on a tight deadline. There are just seven months left to deliver a general-purpose AI Code of Practice that provides the legal certainty needed for developers to continue innovating in Europe. Despite the many challenges identified by participants, the AI Roundtable showed strong resolve to keep the drafting process on track.”
The following can be attributed to Senior Vice President and Head of CCIA Europe, Daniel Friedlaender:
“Getting this Code of Practice right really is a make-or-break moment for AI innovation in Europe. Without clear guidelines that will stand the test of time, Europe’s home-grown AI pioneers might soon become discouraged, while other AI developers will hesitate to enter the EU market. European consumers and businesses would risk missing out on cutting-edge AI technologies.”
