Key Developments

The European Commission has published the second draft of its Code of Practice on Marking and Labelling of AI-generated content, marking a significant milestone in operationalizing the EU AI Act’s transparency requirements. This refined version incorporates feedback from hundreds of stakeholders including industry, academia, and civil society, gathered through surveys, meetings, and workshops held in January 2026.

The updated code introduces a streamlined two-layered marking approach featuring secured metadata and watermarking, optional fingerprinting and logging capabilities, and standardized protocols for detection and verification. Feedback collection continues until March 30, 2026, with final publication expected by June.

Industry Context

This development comes as Ireland prepares to take the EU presidency in the second half of 2026, with AI competitiveness as a key focus. The timing is strategic - Ireland will host a dedicated International AI and Digital Summit during its presidency, positioning the country at the center of European AI governance discussions.

Meanwhile, high-level European dialogue continues to address AI’s broader implications. On March 9, the Council presidency, European Central Bank, and Commission met with social partners to discuss “Artificial Intelligence and the Future of the EU labour market,” highlighting the Act’s intersection with economic and social policy.

Practical Implications

For AI system deployers, Section 2 of the revised code provides more flexibility in labelling deepfakes and text publications concerning matters of public interest. The practice-oriented approach reduces compliance burden while maintaining robust transparency standards.

The August 2, 2026 effective date for transparency rules gives organizations limited time to implement marking systems. Companies should begin evaluating technical requirements now, particularly the two-layered marking approach that combines multiple identification methods.

Open Questions

Several implementation details remain unclear. The code’s flexibility provisions may create inconsistent industry practices without clearer technical specifications. Additionally, the relationship between optional fingerprinting requirements and mandatory watermarking needs clarification for cross-border AI services.

The broader Digital Omnibus proposal, introduced in November 2025, adds another layer of complexity by simplifying overlapping rules across AI, data access, and cybersecurity - particularly relevant for financial sector applications.