The Transparency Dilemma

What it is

Openness is usually treated as a route to trust: tell people how the work was made and they will judge it fairly. The preprint tests that assumption for AI use and finds a cost. Across thirteen experiments, actors who said they had used AI were trusted less than actors who said nothing.

The authors explain the penalty through legitimacy. When a person says a piece of work was made with AI help, observers tend to see the work, and the person, as less legitimate, and trust falls with that judgement. The pattern held across different ways of framing the disclosure.

In effect

Three details shape what the finding means in practice. The penalty appeared even when disclosure was voluntary. It was weaker than the penalty that followed when a third party exposed the AI use, so being found out cost more than saying so first. And it was smaller, though still present, among evaluators who held favourable attitudes to technology.

The Why We Project article on branding in the age of AI uses the finding to explain why cheap, polished output shifts trust towards signals that are harder to fake, such as a record kept over years and work that shows its own effort.

What it does not say

The finding does not say that hiding AI use is wise, since exposure by others carried the larger penalty. It concerns people who disclose, not brands, so applying it to a brand is an extension of the research rather than a direct result. It is reported in a preprint, and it does not measure sales or long-run loyalty.

---