On 5 March 2026, the European Commission published a second draft Code of Practice on marking and labelling AI-generated content. The Code is intended to support compliance with Article 50 transparency obligations under the EU AI Act.
The Commission reiterates that the transparency rules will become applicable from 2 August 2026. The draft Code therefore acts as a practical implementation reference during a tight window ahead of that date.
The draft differentiates between provider-side measures (machine-readable marking of AI-generated content) and deployer-side measures (disclosures when publishing deepfakes or AI-generated text on matters of public interest). It promotes an approach that combines visible disclosures with technical marking, while recognising that no single technique will fit every use case.
Feedback on the second draft is open until 30 March 2026, and the Commission indicates the Code is expected to be finalised by early June 2026.
Acompli perspective: If you ship a user-facing system built on a third-party model API, you may still carry provider and/or deployer duties under Article 50. Map systems now and decide how marking and disclosure will be implemented end-to-end. Maintaining compliance documentation and meeting regulatory transparency obligations starts with understanding what you process and why.