Europe’s AI-content draft separated technical marks from public labels

The European Commission published its first AI-content transparency code draft on December 17, 2025. The proposal addressed two different jobs: providers would mark generated or manipulated material in machine-readable form, while professional deployers would disclose deepfakes and certain AI text concerning the public interest. The code was a voluntary compliance aid for underlying AI Act obligations. Feedback was requested until January 23, 2026; that consultation has ended. ^[1]

The draft proposed combining marking methods because it judged no single active technique sufficient. It left technical questions open for short text and software code. Public disclosure raised a different problem: icons should be visible at first exposure and accessible to people with disabilities. Machine-detectable information and a visible label were complementary tools. These proposed commitments remained open to revision, rather than proving any implementation effective. ^[2]

On June 10, 2026, the Commission published the final code, still describing participation as voluntary. It presented transparency as a way to help people recognize generated or altered material and reduce deception. That follow-through means the December draft should be read as the starting proposal, rather than the final instructions. ^[3]

For historical comparisons, check the document’s version and whether its subject is technical marking or public disclosure. ^{[1] [3]}

About this item: Retrospective report. No first-hand reporting or independent testing is claimed. Corrections policy.