If you paste AI-written wording straight into your work, do readers expect you to say so, however hard you worked?

This explores whether pasting AI-generated wording straight into your writing, rather than using AI to brainstorm or polish, changes what readers expect you to disclose. The corpus says yes, and it points to a specific trigger. In a 727-person vignette study, disclosure felt most necessary when AI text was directly incorporated and *irreplaceable*, meaning the piece wouldn't hold up without it. The surprise is that how hard the writer worked had no effect on those judgments. Readers aren't scoring your effort. They're asking whether the words in front of them are yours. The same study found a built-in gap: readers rated disclosure as more necessary than writers did across the board Do readers and writers differ on AI disclosure necessity?.

Direct copying is also the usual case, not an unusual one. One large study found writers edited AI-suggested paragraphs only 23% of the time, and the edited versions stayed about 96% similar to the original Do writers actually edit AI-generated text before publishing?. That matters because the copied text isn't neutral. Across nearly 3,000 writers and 11,000 readers, AI assistance shifted how the writer came across on all 29 measured traits, making them seem more extreme, more confident and more privileged Does AI writing assistance change how readers perceive the writer?. When text goes in nearly verbatim, readers are partly meeting a persona the AI wrote. That gives them a concrete reason to want to know.

The writer's side is just as revealing. People don't feel they *own* AI-generated text they use, yet they still don't credit the AI publicly. They treat it like a ghostwriter nobody mentions Do people feel they own AI-generated text they use?. Feeling of ownership goes up as people gain more control over what the AI produces Does user control over AI text shape feelings of ownership?. Put those findings together and copying text verbatim is the case where writers feel the least ownership, where readers most expect disclosure, and where writers are still inclined to stay quiet. The gap between those expectations is widest exactly where disclosure matters most.

Why stay quiet? Disclosure has a cost, though how much depends on the kind of writing. For a news article, an AI disclosure lowered ratings by less than 0.15 points on a 7-point scale, and that held for both human and LLM raters Does disclosing AI assistance make readers trust articles less?. In personal writing such as messages or interpersonal notes, disclosure sharply lowered perceived trust, caring and likability, because readers don't believe an AI can genuinely empathize How does revealing AI authorship change reader trust?. Readers with higher AI literacy penalize disclosure less, and some react positively Does AI literacy reduce the damage from AI disclosure?. So copying AI text into a heartfelt note to a friend carries much higher stakes than copying it into a report.

A simple 'AI was used' label may also be the wrong tool. Disclosure makes audiences more critical, but 34–62% were still persuaded anyway Does telling people an AI wrote something actually stop them from believing it?. Two alternatives appear in the corpus. Collaborative writers preferred editors that show *when, where and how* AI was prompted, which is a running record of AI use rather than a one-time admission, though some found full visibility intrusive Do writers want to see each other's AI prompts in shared editors?. In newsrooms, linking every claim to its source made AI-assisted output trustworthy enough to adopt Can source traceability make AI writing trustworthy?. Taken together, these suggest disclosure is shifting from a yes/no confession toward showing which words came from where.