Why do writers hide AI use from readers yet show it to co-writers, and is that the same kind of openness?

This explores why AI use tends to be hidden from readers but shared with co-writers, and whether those two choices are really about the same thing. The corpus suggests they aren't. Showing AI use to a collaborator lets them see how the work is being made. Disclosing it to a reader puts a label on a finished product, and the reader then judges the product by that label. That difference explains why writers act so differently with each audience.

Take collaborators first. In a study of paired writers using shared editors, people strongly preferred editors that showed more of each other's prompting: when, where and how AI was used Do writers want to see each other's AI prompts in shared editors?. They valued it because it helped them follow a partner's thinking and check text the AI had produced. Visibility gave them something to work with. Even so, some found full sharing intrusive and felt self-conscious, so transparency has costs even among people who trust each other. Readers get none of this process view. They see only the polished result, which is the gap described in Does AI separate intellectual form from the thinking behind it?: AI separates what a piece of writing looks like from the thinking that used to be required to produce it. A collaborator can see both. A reader sees only the surface, so a disclosure label is all they have to go on.

That is where the penalty comes in. Telling readers AI was involved lowers how trustworthy, caring and likable they find the writer, and the drop is steepest in personal writing, where readers expect real empathy How does revealing AI authorship change reader trust?. In news writing the penalty is small but consistent: under 0.15 points on a 7-point scale, from human and LLM raters alike Does disclosing AI assistance make readers trust articles less?. Writers also underestimate how much readers care. Readers rate disclosure as more necessary than writers do, especially when AI text was pasted in directly and couldn't easily be replaced. How much effort the writer put in made no difference to these judgments Do readers and writers differ on AI disclosure necessity?. So writers who hide AI use are guessing wrong about what their audience expects.

The less obvious finding is that the reader penalty isn't fixed. Readers with higher AI literacy react much less to disclosure, and some react positively Does AI literacy reduce the damage from AI disclosure?. Outside writing, people at first avoid partners they know are AI, but that preference reverses once they repeatedly see good results. Disclosure without that feedback changes nothing Does revealing AI identity help or hurt user trust?. In effect, a collaborator gets that feedback loop built in, while a reader usually gets one label and no track record. Disclosure also doesn't switch off persuasion. It makes audiences more critical, but 34–62% were still persuaded Does telling people an AI wrote something actually stop them from believing it?.

Hiding has its own costs, and some fall on people who didn't use AI at all. Writing-process data such as timing and bursts of activity can flag wholesale handoffs to AI, but it can't tell light AI help apart from ordinary drafting Can process data distinguish AI delegation from ordinary collaboration?. One paper claims that heavy rewriting both hides authorship and evades AI detectors, but it never tested the detectors directly Do rewrites that hide authorship also fool AI detectors?. Meanwhile, readers' suspicion spills onto human writers: comments accused of being AI showed no features that actually set AI text apart. The accusations work more like gatekeeping than detection Do unfounded AI accusations harm human writers instead?. Collaborators who can see each other's process don't face this guessing game. Readers do, and when AI use is hidden, they end up suspecting everyone.