EU Publishes Final Code of Practice as AI Act Implementation Framework Takes Shape

Key Developments

The European Commission has published the final Code of Practice on marking and labelling of AI-generated content, marking a critical milestone in the practical implementation of the EU AI Act. This voluntary guidance tool is designed to help AI providers and deployers comply with the Act’s transparency obligations, which come into force in August 2026.

The Code addresses a fundamental challenge facing the AI industry across Europe: how to properly disclose AI-generated material to end users. It covers marking AI-generated content and disclosing the artificial nature of images, audio (including deepfakes), and text—areas where ambiguity has created compliance uncertainty for European and international companies operating in the bloc.

Simultaneously, the Commission opened a call for applications to select members of the RAISE High-Level Academic Advisory Board, which will bring together 10 to 15 leading academics in AI science. The deadline for applications is September 4, 2026.

Industry Context

These governance and transparency moves arrive as the EU’s broader AI Act implementation enters a critical phase. The landmark regulation—the world’s first comprehensive horizontal AI framework—entered force on August 1, 2024, but its most consequential provisions governing high-risk AI systems were originally set to apply from August 2, 2026.

However, the implementation timeline proved challenging. In response, EU institutions reached a provisional political agreement on May 7, 2026, on Digital Omnibus amendments that would stagger compliance deadlines. These amendments, awaiting formal adoption before August 2, 2026, would defer high-risk obligations to December 2, 2027, for stand-alone systems and August 2, 2028, for those embedded in regulated products.

The Code of Practice now provides the immediate practical bridge to that August 2026 transparency deadline, giving providers concrete guidance on what compliant disclosure looks like in practice.

Practical Implications

For AI builders and deployers across the EU and beyond, the published Code offers actionable clarity on a previously ambiguous requirement. Voluntary adoption now positions companies ahead of formal enforcement, reducing implementation friction later.

For Irish tech companies and European AI startups, the Code’s emphasis on practical, sector-appropriate marking standards matters significantly. Ireland hosts major AI development operations and research centres; clarity on transparency compliance directly affects how those operations structure their AI systems.

The RAISE Board itself signals the Commission’s intention to embed independent academic expertise directly into AI Act enforcement and standard-setting—a governance approach that could shape how European AI regulation evolves beyond 2026.

Open Questions

Key uncertainties remain: How strictly will the Code be enforced post-August 2026, and will voluntary compliance prove sufficient, or will the Commission move toward more prescriptive requirements? The formal adoption of the Digital Omnibus amendments remains pending; their final language could still shift enforcement timelines for high-risk systems. And it’s unclear how the Code will interact with emerging Member State-level implementation frameworks, particularly in jurisdictions like Ireland with significant AI sector presence.

Source: European Commission