Writers are deliberately adopting an "anti-AI" prose style, favouring quirks, typos and idiosyncratic phrasing to distinguish their work from AI-generated text. The trend, spanning fiction, journalism and criticism, involves consciously avoiding hallmarks of LLM writing, such as em dashes, rule-of-three lists and passive constructions, as authors seek to reassure readers of their human authorship amid growing scrutiny of AI use in publishing.

Novelist Laura Brooke Robson says she now deliberately courts small imperfections and unusual punctuation as "winks to the reader" proving human authorship, and believes AI could push literary fiction further towards the "weird" and "esoteric". The shift follows high-profile controversies, including Hachette pulling Mia Ballard's novel Shy Girl in March over AI-use speculation, and author Steven Rosenbaum admitting his book The Future of Truth contained AI-fabricated quotes. Editors at outlets such as Aeon and Pitchfork say they are now actively encouraging writers to sound more conversational, spiky and idiosyncratic, resisting the smoothing effect of AI tools.