What ASCI's AI Labelling Rules Mean For Creator Posts
Most coverage of ASCI's new AI guidance has read it as a problem for brand and agency teams. Read it again from the creator's side and it is sharper there, because the label appears on creator posts, and a post carries someone's name.
Creator work is where generated elements are most common and least documented. A backdrop that was never visited, a voice cleaned up past recognition, a product shot that was never photographed.
What now needs a label on creator posts
A replicated likeness or voice, even with consent. A setting fabricated in a way that changes how the product is understood. A demonstration of something the product does not do. And a paid AI product recommendation, which must say it is sponsored.
What does not need one is reassuring: colour correction, blemish removal, decorative elements and obviously fantastical effects. The ordinary grammar of a creator post is untouched.
The part that cannot be labelled away
Fabricated testimonials and unauthorised likeness are prohibited outright. For creators that matters most in one specific place, which is using another person's face or voice in a paid post without their agreement, including as a joke.
A disclosure fixes a transparency problem. It does nothing for a permission problem.
Who carries the risk
In practice, the creator, because the post is published from their account. Brand contracts rarely allocate this, and most were signed before the guidance existed.
What to change in the brief
One line, added by whichever side is more organised: which AI-assisted elements are expected, who decides if a label is needed, and who writes it. ASCI publishes the full guidance, and it takes effect three months from publication.
More from Social-First.