Voluntary AI involvement label: D
AI-generated
machine-made, lightly edited · Predominantly AI-generated, limited human changes
AI produced most of the piece. Human involvement went beyond prompting — a pass of edits, some corrections — but stopped short of substantive transformation.
Typical examples
- AI-drafted newsletters sent after a quick pass
- Bulk product copy with spot fixes
- Generated reports with light edits
Human review at this level
A person skimmed or lightly edited it; no thorough reworking of the substance.
Article 50 context
AI Act · bindingThe grade alone triggers nothing. Check the situation instead: the deepfake and public-interest text duties turn on what the content depicts and how it is published, not on how much of it a machine produced.
EU icons and this grade
This grade does not select an EU icon, and no grade does. The icons are optional tools from the voluntary Code of Practice, and Article 50 requires disclosure in defined situations — a possible deepfake, or qualifying public-interest text — whichever grade the piece happens to carry.
The three icons, and when to reach for oneThe badge
Seen this badge somewhere? It means the publisher declares this level of AI involvement using the voluntary aiflags.eu scale. It is not an official EU label or legal risk class.