Evaluating the Prevalence of Artificial Intelligence-Generated Writing in Plastic Surgery Literature

The rise of artificial intelligence (AI) and large language models in academic writing has raised concerns regarding research integrity and authorship transparency. This study evaluated the prevalence of AI-generated content in plastic surgery publications following the release of ChatGPT.

We conducted a cross-sectional study of 1,627 manuscripts published in 10 major plastic surgery journals between January 2024 and May 2025. ZeroGPT was used to quantify AI-generated content. A baseline threshold for substantial AI involvement (22.5%) was established using 300 pre-ChatGPT manuscripts (2010-2011). Outcomes included the proportion of manuscripts exceeding this threshold, average AI content, and associations with publication year, journal, and evidence level.

Overall, 21.5% of 2024-2025 articles exceeded the threshold for substantial AI involvement. The median proportion of AI-generated text rose from 7.4% in 2024 to 12.2% in 2025, while the percentage of manuscripts with substantial involvement increased from 17% to 29%. AI involvement varied widely across journals (0-41%). In multivariable analysis, 2025 publication year (OR 1.86, p<0.001) and certain journals were independently associated with substantial AI involvement. Higher evidence level studies demonstrated greater AI involvement, with Level 4 studies showing significantly lower odds than Level 1 (OR 0.47, p=0.001).

More than one in five recent plastic surgery manuscripts contain substantial AI involvement, with marked variation across journals and evidence levels. These findings highlight the need for standardized editorial guidelines governing AI use to maintain research integrity and transparency in plastic surgery literature.