AI disclosure means telling the audience, plainly, that AI was involved in producing a piece of content: an ad, an image, a written summary, a customer-service reply. It’s increasingly a legal requirement rather than a courtesy. The EU AI Act’s Article 50 transparency obligations, several US state disclosure laws, and platform-level labeling rules on social networks are converging on the same baseline: audiences get told when AI made or materially shaped what they’re looking at.
The complication is that disclosure doesn’t reliably do what it’s meant to. Controlled studies comparing identical content, one version labeled AI-generated and one not, find the labeled version rated as less natural, less appealing, and less trustworthy, purely from the label. A bare “AI-generated” tag doesn’t read as transparency to most audiences. It reads as confirmation of something they already suspected, which makes disclosure a compliance requirement that can actively work against the trust it’s supposed to build.
The fix isn’t skipping disclosure, since the legal and platform requirement to make it usually isn’t optional. It’s disclosing the right thing: what a person did, directed the brief, selected the final version, rejected the alternatives, rather than only the fact that AI touched the process somewhere upstream.