What Should AI Disclosure In Advertising Say?
AI disclosure is on a well-worn path. A reasonable principle becomes a mandatory label, the label goes on everything because nobody wants to judge the threshold, and a universal label carries no information at all.
Cookie consent went exactly this way. So did allergen warnings that list every possible trace. The failure is not the intention, it is that a signal appearing on all items cannot distinguish between them.
What AI Disclosure Is Supposed To Protect
The reader's ability to judge what they are looking at. That matters when the method changes the meaning: a person who does not exist, an event that did not happen, a demonstration that was rendered rather than filmed.
It does not matter when the method changed only the labour: a background extended, a cut assembled, a version resized. Those have been retouched, composited and automated for thirty years without anyone asking for a label.
Why A Blanket Label Fails Both Ways
It exhausts the reader, who stops seeing it. And it gives cover to the cases that should have been disclosed loudly, because they are now wearing the same badge as a resized banner.
A warning that appears on everything is not a warning. It is wallpaper, and everyone stops reading wallpaper within a week.
A Disclosure Worth Making
Disclose what would change the reader's understanding if they knew it, in the place they would need to know it, in words that say what was done rather than which software was involved.
That is harder than a checkbox because it requires judgement each time, and judgement does not scale the way a policy does. It is also the only version that survives contact with an audience, which is the test any disclosure regime eventually has to pass. It follows directly from how agencies handle transparency internally.
Advertisement
