Study finds wide journal variation in AI-generated text, up to 83.5% in 2026

[medRxiv] A preprint study of abstracts from 12 general medical journals estimated that the prevalence of AI-generated text rose after 2022, with marked differences across journals in timing and magnitude. Researchers randomly sampled up to 200 abstracts per journal-year from publications dated Jan 1, 2020, through Aug 8, 2026, totaling 15,268 abstracts, and scored them with GPTZero using journal-specific thresholds calibrated to a 1% false-positive rate in the 2020-2021 pre-AI baseline. Through Aug 8, 2026, estimates ranged from 5.7% (95% CI, 2.1%-10.0%) in the New England Journal of Medicine to 83.5% (95% CI, 77.1%-87.5%) in Frontiers in Medicine; the publication-volume-weighted estimate across sampled journals was 66.7% (95% CI, 64.0%-69.2%). The authors said the proportion was lower in journals with higher impact factors and higher in fully open-access journals. These are unreviewed preprint findings.