Does telling readers 'AI wrote this' spook them more than saying 'AI just edited this'?
Does disclosing direct AI-generated text trigger stronger skepticism than disclosed editing?
This explores whether readers become more distrustful when told an AI wrote the text outright than when told an AI only helped edit it, and what the collection can and can't say about that difference.
This explores whether readers distrust text more when told an AI wrote it outright than when told an AI only helped edit it. The short answer: the collection doesn't contain a study that directly compares how much skepticism each kind of disclosure produces. What it does have points in a clear direction, and it also raises a less obvious problem with drawing the line between generating and editing at all.
The nearest evidence comes from a 727-person vignette study. When AI text was pasted in directly and couldn't easily be swapped out, people judged disclosure as more necessary than when the AI's role was smaller. Readers also felt this more strongly than writers did Do readers and writers differ on AI disclosure necessity?. How hard the writer worked made no difference. What readers cared about was how much of the AI's output survived into the final text. That is a finding about what readers feel they're owed, not a measure of their skepticism. Still, it suggests readers react to how much AI text they're reading, not just to whether AI was involved. Writers seem to follow a similar logic about their own work: their sense of ownership rises with how much control they had over the text Does user control over AI text shape feelings of ownership?.
Where skepticism has been measured, the label itself does most of the work. In one preregistered experiment, emails written with AI help and left unlabeled were rated as favorably as human-written ones. Strong distrust appeared only once AI involvement was disclosed Do readers trust unlabeled AI-written messages as much as human ones?. Even then, disclosure doesn't erase persuasion. Audiences become more critical, but a third to well over half stay persuaded anyway Does telling people an AI wrote something actually stop them from believing it?. So disclosure acts more like a switch than a dial. Whether a reader's distrust grows with 'wrote it' versus 'edited it' is the open question.
The surprising part is that the line between 'AI-generated' and 'AI-edited' may be thinner in practice than the labels suggest. One study found writers edited AI-drafted paragraphs only 23% of the time, and their edits left the text about 96% the same Do writers actually edit AI-generated text before publishing?. Even light AI assistance shifted how readers saw the writer on all 29 traits measured, making them seem more confident, more extreme and more agreeable Does AI writing assistance change how readers perceive the writer?. So 'I only used AI to edit' may describe text that is mostly AI-shaped. If readers were more lenient toward that label, the leniency could be misplaced.
There's also a cost on the other side. Suspicion doesn't need a disclosure to show up. Readers sometimes accuse human writers of using AI based on features that don't actually tell AI and human text apart, so the distrust ends up landing on people who did nothing Do unfounded AI accusations harm human writers instead?. If you want to look further, the gap to explore is a direct experiment that labels the same text 'AI-written' or 'AI-edited' and measures trust. The corpus suggests the answer would be 'yes, somewhat', but it hasn't run that test.
Sources 7 notes
A 727-person vignette study found readers consistently rated AI disclosure as more necessary than writers did. Disclosure seemed most necessary when AI text was directly incorporated and irreplaceable, while writer effort had no effect on these judgments.
Study 1 found that greater user control over generated text raised sense of ownership, while personalizing the AI model had no impact on the AI Ghostwriter Effect.
In a preregistered experiment (N=647), recipients rated unlabeled AI-assisted emails indistinguishably from human-written ones. Only explicit AI disclosure triggered strong skepticism. Recipients appear to default to trust rather than suspicion when origin is unrevealed.
Audiences aware of AI involvement became more critical and scrutinizing, yet 34–62% across groups remained persuaded. Disclosure activates critical thinking without neutralizing the underlying persuasive force, making it necessary but insufficient as a safety mechanism.
Writers edited AI-generated paragraphs only 23% of the time, with edits averaging 96% similarity to the original. This means AI's opinionated and distorted voice propagates with minimal human filtering before publication.
Show all 7 sources
A study of 2,939 writers and 11,091 readers found AI assistance shifted every tested dimension—29 total—toward extremism, confidence, quality, agreeableness, and perceived privilege. Distortions were statistically significant and directional, not random noise.
Accused comments lack features that distinguish AI text from human writing, suggesting accusations function as gatekeeping rather than detection. This inverts the AI-as-perpetrator framing, placing harm at the receiving side through reader skepticism.
Papers this line draws on 8
The research behind the notes this line reads — ranked by how closely each paper relates.
- Understanding Reader Perception Shifts upon Disclosure of AI Authorship
- Penalizing Transparency? How AI Disclosure and Author Demographics Shape Human and AI Judgments About Writing
- What Influences Readers' and Writers' Perceived Necessity of AI Disclosure?
- Measuring and Mitigating Persona Distortions from AI Writing Assistance
- When Should You Disclose AI Use? The PACED Framework
- "It was 80% me, 20% AI": Seeking Authenticity in Co-Writing with Large Language Models
- Biased AI writing assistants shift users' attitudes on societal issues
- The Assistant Erased You: Measuring Loss of Authorship Signals in AI-Mediated Communication