Attribution, consent, and synthetic media

Before publishing any AI-generated or AI-modified image, audio, or video, three questions are worth answering on the record. The answers belong in the newsroom's standing policy, not in the head of whoever was on shift.

First: does the synthetic element depict a real, identifiable person? If yes, the bar is consent — explicit, documented, and revocable. There is no acceptable workaround.

Second: is the synthetic element disclosed at the point of consumption? Caption-level disclosure is the minimum. Reader-facing labels and provenance signals are stronger and increasingly expected.

Third: is the synthetic element material to the editorial meaning of the piece? Decorative use is a different conversation from evidentiary use. The disclosure language should reflect that distinction.