Why do the people writing with AI feel less need to admit it than the people reading their work expect?

This explores why writers who use AI tend to see disclosure as less necessary than their readers do, and what might explain the gap. The gap itself is well documented. A 727-person vignette study found readers consistently rated AI disclosure as more necessary than writers did (Do readers and writers differ on AI disclosure necessity?). The two sides also seem to think about the question differently. When a writer steered the AI less deliberately, readers became *more* convinced that disclosure was needed, while writers became *less* convinced (Why do readers and writers disagree on disclosure necessity?). How hard the writer worked made no difference to either group. What mattered was whether AI text went straight into the piece and couldn't easily be replaced. Readers seem to be asking 'whose words are these?', while writers may be asking 'did I do my part?'

No study here asks writers directly why they hold back, but the evidence on what disclosure costs gives a strong motive. People say they want disclosure, yet they rate AI-attributed text as less trustworthy, less authentic and less useful once it's labeled (Why do people penalize AI disclosure even when they request it?). That is a disclosure paradox: readers ask for the label and then penalize the person who provides it. The penalty depends heavily on the kind of writing. On a news article it is tiny, less than 0.15 points on a 7-point scale (Does disclosing AI assistance make readers trust articles less?). In personal, interpersonal writing it is steep, because readers see AI as unable to genuinely care (How does revealing AI authorship change reader trust?). So writers' reluctance may be strongest exactly where readers feel most entitled to know.

The workplace evidence points the same way. In Anthropic's 1,250-person interview study, nearly all creatives said AI saved them time, yet about 70% described stigma or managing how peers judged their AI use (Why do workers hide productivity gains from AI use?). There is also a less obvious risk: being falsely accused. People accused of using AI often wrote text with nothing that actually marks it as AI-generated. The accusations seem to work as gatekeeping rather than detection, so human writers get wrongly dismissed (Do unfounded AI accusations harm human writers instead?). When suspicion alone carries a cost, admitting AI use can feel like confirming the accusation.

Writers aren't against transparency in general, though. In shared editors, paired co-writers strongly preferred seeing more of each other's AI prompts (Do writers want to see each other's AI prompts in shared editors?). Among collaborators, visibility helps people understand each other. With a general audience, disclosure turns into a judgment. Two findings suggest the gap could shrink. Readers with higher AI literacy penalize disclosure less, and some react positively (Does AI literacy reduce the damage from AI disclosure?). Nielsen Norman Group also argues there is no universal rule and suggests weighing audience, context and how much AI contributed case by case (When should you disclose AI use in written content?).

One caveat: most of this evidence comes from hypothetical vignettes, and the intentionality study's authors warn that their effect may not hold in real situations. Writers' motives are inferred from the penalties they face, not measured directly.