The paper evaluates how well six popular large language models can imitate the writing style of individual authors using short product reviews. Two prompt styles—simple and stylometric—were tested, and authorship mimicry was measured with cosine distance and embedding visualisations. Results show that while Gemini 3 Pro performed best, it largely copied training text, and overall most models struggled to converge on target author style, highlighting the need to differentiate genuine imitation from direct reuse.
arXiv:2310.00436v2 Announce Type: replace Abstract: Authorship identification uses patterns in writing to infer who wrote a text, but those patterns also reflect topic, genre, and register. This surv...
The paper introduces a new authorship attribution model that represents each author with multiple consistent but distinct writing styles, rather than a single style profile. Experiments across four scenarios—audience size, temporal distance, preceding context, and genre norms—show that authors vary significantly in how idiolectic their writing is, and that certain settings yield text that is more informative about authorship. The authors demonstrate that modeling multiple styles per author leads to significant performance improvements over traditional single-style approaches.
The paper "Evaluating Style-Personalized Text Generation: Challenges and Directions" examines the difficulties of assessing text that is tailored to individual users’ styles. It critiques common metrics such as BLEU, embeddings, and LLM-as-judges, and introduces a style discrimination benchmark covering domain discrimination, authorship attribution, and LLM-generated personalized versus non-personalized discrimination across eight writing tasks. The study finds that ensembles of diverse evaluation metrics outperform single-evaluator approaches and offers guidance for reliable assessment of style-personalized generation.
arXiv:2510.13302v4 Announce Type: replace-cross Abstract: Computational stylometry studies writing style through quantitative textual patterns, enabling applications such as authorship attribution, i...
The paper introduces ATLAS, a likelihood‑based method for model authorship attribution (MAA) that learns how each source family generates text and attributes new text by comparing token‑level likelihoods. It also presents MAABench, a benchmark for evaluating MAA under out‑of‑distribution scenarios such as unseen domains and future model versions. Experiments show ATLAS outperforms existing baselines, achieving 74.27% accuracy on the hardest MAABench setting and correctly identifying teacher models in distillation experiments.
arXiv:2606. 31250v1 Announce Type: cross Abstract: Large language models (LLM) trained on web-scale corpora generate output that may infringe copyright, yet existing technical safeguards focus narrowly on verbatim memorisation.
arXiv:2608.11025v2 Announce Type: replace Abstract: Emergent misalignment (EM) is the phenomenon where fine-tuning a language model on a narrow task leads to harmful behavior in unrelated domains. A...
arXiv:2510.08831v2 Announce Type: replace Abstract: As AI writing tools become widespread, we need to understand how both humans and machines evaluate literary style, a domain where objective standar...
The paper examines how different author representations affect zero‑shot authorship attribution (AA). It compares a label‑only prompting baseline with three strategies: representative writing samples, LLM‑generated descriptions, and style embeddings (LISA). The study finds that incorporating author‑specific representations improves performance, with a two‑stage LISA framework achieving the best results, while LLM‑generated descriptions offer a more compact but slightly less accurate alternative.
Emergent misalignment (EM) is the phenomenon where fine-tuning a language model on a narrow task leads to harmful behavior in unrelated domains. A leading mechanistic account attributes EM to persona features: latent directions acquired during pre-training that misaligned fine-tuning amplifies.
arXiv:2609.22112v1 Announce Type: new Abstract: Large language models (LLMs) have demonstrated the ability to generate user-specific text with high stylistic fidelity. However, the personal data that...
The paper introduces a training‑free method for uncovering prompt‑conditional stylistic axes in large language models (LLMs). By repeatedly sampling completions of a single prompt at high temperature and applying Principal Component Analysis (PCA) to the pooled hidden activations, the authors automatically label the resulting axes using the extreme (pole) generations. Validation against 245 human‑elicited stylistic annotations shows that, for the Qwen‑3.5‑4B‑Instruct model, the top two axes align with human dimensions with 72.8% precision and 43.6% macro‑recall, and 75.6% of validity ratings confirm the axes’ polar generations, while other models exhibit varying degrees of discoverability.