Finding the Fingerprints of Generative Artificial Intelligence in Psychology Publications

Generative artificial intelligence (AI) and large language models (LLMs) in particular have become embedded in the research process such that scientists are increasingly using LLMs to assist with manuscript drafting. However, the extent to which LLMs are being used in psychology-research writing is largely unexplored. In this study, we investigated the presence of AI-associated linguistic markers in psychology publications between 2018 and 2025. Using a corpus of 512,970 abstracts across 975 journals, we analyzed the relative frequencies of 28 words previously identified as overrepresented in AI-assisted writing. Results showed a small gradual increase in these stylistic markers from 2018 to 2022, followed by a marked acceleration in their use after the release of ChatGPT and relative frequencies increasing in 2023, 2024, and 2025. Counterfactual modeling suggested that at least 16% of 2025 abstracts contained linguistic traces consistent with LLM editing. Group-level analyses revealed that increases were not confined to lower-prestige outlets; higher-ranked journals, major publishers, and both English- and non-English-speaking countries all showed significant growth in AI-associated word use. These findings point to large-scale shifts in academic writing that are consistent with the growing influence of LLMs, although such markers provide indirect rather than definitive evidence of their use.