Generative artificial intelligence in the writing process of accounting research

We examine how accounting researchers use generative AI in their writing. To measure AI use, we construct a list of 13 words that prior studies identify as disproportionately common in AI-generated text. We validate our list using 52 American Accounting Association (AAA) papers that explicitly disclose AI use. These papers contain AI-associated words at more than twice the rate of papers in our post-period, and more than four times the rate of our pre-period sample. We then analyze a sample of 3573 papers across 10 accounting journals (2020-forthcoming). We find that junior researchers, authors at less research-intensive institutions, and authors publishing in lower-ranked journals use more AI-associated words. Within papers, AI-associated words concentrate in abstracts, introductions, and conclusions. Using a broader sample of 4166 abstracts from 22 journals, we document a 75% increase in AI-associated words after 2023. Among papers that disclose AI use, ChatGPT is the most common tool, used primarily for proofreading and improving clarity. Our findings inform the ongoing discussion of the role of AI in accounting research (Graham and Tucker, 2026).