Does labeling content as AI-made hurt trust?

Marketing teams keep asking a version of the same question: if we say a piece of content was made with AI, do we lose the audience's trust? A 2025 study with 3,861 people gives a two-part answer — a cost that is real, and a spillover that is smaller than the fear.

The study sits in the AI content and consumer trust research index. It looked at news content and civic attitudes, so read the marketing implication with that boundary in mind.

What the label did

Wang, Sturgis, and de Kadt ran a survey experiment with 3,861 nationally representative participants, showing some an AI-generated label on a news article and others no label. The label significantly reduced how accurate people rated the article. Attaching "AI-generated" made the same content read as less credible.

That effect is worth taking seriously on its own. A label is not free; it changes how the labeled piece is judged.

What the label did not do

The broader effects were narrow. The label lowered interest in the article's policy topic, but it left policy support and general concern about misinformation largely unchanged. The skepticism attached to the specific piece; it did not spread into wider attitudes the way a reputational-contagion story would predict.

That gap matters. A common assumption is that disclosing AI use taints the brand broadly. In this study, the effect stayed mostly local to the labeled item.

Reading it for marketing, carefully

This is a news study, not an ad study, and the fair move is to hold that difference. News accuracy and ad persuasion are not the same response, and civic topics are not products.

What transfers is the shape of the finding, not a number to quote. An AI label carries a real cost to how the labeled piece is judged, and the evidence here does not support the idea that disclosure detonates broader brand trust.

What actually earns trust

Two other findings in the same research index frame the real lever. Consumers cannot reliably tell AI-written text from human-written text by intuition, so trust does not come from readers detecting authorship. It comes from the brand's substance and reputation carrying the content.

This is why Copper Sun holds AI output to writing standards past brand-voice matching, so a disclosed piece still reads as the brand's own work rather than generic AI. Disclosure is a labeling decision; substance is the trust decision. See how it works.

Frequently Asked Questions

Should we disclose that content was made with AI?

The research supports disclosure as the responsible default, with eyes open about the cost. People cannot reliably detect AI authorship on their own, so a label is the only real transparency mechanism. This study shows the label lowers perceived accuracy of the specific piece, not that it wrecks broader brand trust. Disclose, and put the work into making the content good enough that the label is not the most interesting thing about it.

Does an AI label always reduce trust?

It reduced perceived accuracy of the labeled item in this study of news content. Whether that carries over to an ad or a brand is not settled, since ads and news trigger different responses. The safe reading is that a label carries some cost to how the labeled piece is judged, and the size of that cost depends on context the research has not fully mapped.

If disclosure has a cost, why not skip it?

Because the alternative is worse and less stable. Consumers cannot detect AI content reliably, so skipping disclosure means depending on non-detection rather than on being straight with an audience. Regulatory and platform expectations are also moving toward disclosure, so building the practice now avoids a harder retrofit later. The measured cost here was bounded; the downside of getting caught not disclosing is not.

How is this different from just writing "human-quality" AI content?

Quality and disclosure answer different questions. Quality determines whether the content is worth trusting once someone reads it; disclosure determines whether you told them how it was made. This study suggests the label dents perceived accuracy, and separate research suggests substance and brand reputation are what carry trust over time. You need both: label it plainly, and make the work stand on its own.