Writers are making their prose worse to avoid AI suspicion, according to a new survey that suggests ”too polished” text is now triggering readers. Professionals are deliberately sanding down their own writing so it looks less like it came from a machine, adding rough edges, shorter sentences, and a bit less grammatical shine just to seem human.
In a Use.AI survey of more than 12,600 people across the US, the UK, the EU, and Latin America, 58% said they had seen people publicly criticized for using AI in content creation. That kind of social penalty is turning AI into more than a productivity tool; for many writers, it is becoming a reputational hazard.
Why polished writing now raises eyebrows
According to the report, 46% of respondents worry their own writing could be mistaken for AI-generated text, while 39% say they actively change their style so it reads as less ”perfect” and more human. The stereotype is easy to spot: overly correct grammar, predictable transitions, and a flat emotional tone. Ironically, that means clean writing can now look suspicious in the same way sloppy writing once did.
Creative workers feel the pressure most sharply. Their carefully edited copy can invite more doubt than a casual, slightly messy draft, which creates a perverse incentive to write down to the stereotype. That is a new kind of anti-aspiration: not ”write better,” but ”write less like a model.”
The new etiquette around AI disclosure
The study also shows growing discomfort with hidden AI use. Roughly a third of people said they would think worse of an author if they learned AI had been used without disclosure, and they would be less likely to support that creator. At the same time, 62% said using AI for editing, idea generation, and research is acceptable as part of normal digital literacy.
That split is the real story here: people do not hate AI assistance so much as they hate the feeling of being sold authenticity that was assembled elsewhere. The result is a new workplace ritual, where writers save time with AI and then spend some of that saved time manually reintroducing imperfections.
What writers are changing in practice
- Shorter sentences instead of polished, balanced phrasing
- More stylistic roughness to avoid a ”machine-clean” feel
- Fewer long dashes, which have become an AI tell in some readers’ minds
The paradox is obvious: AI is supposed to reduce effort, but the fear of sounding automated is adding a second editing pass devoted entirely to looking imperfect. That tension will probably get worse before it gets better, especially as more readers learn to associate smooth prose with software and use the same suspicion as a shortcut for judgment.