This page was translated automatically from Turkish. Read the original

This page is an AI-assisted Turkish summary of the study presented at the 26th World Congress of Psychiatry. The study belongs to the presenting authors; please refer to the original source when citing it.

SO245

Underreporting of artificial intelligence use in psychiatric research: a cross-sectional analysis of three leading journals

UNDERREPORTING OF ARTIFICIAL INTELLIGENCE USE IN PSYCHIATRIC RESEARCH: CROSS SECTIONAL ANALYSIS OF THREE LEADING JOURNALS

Only 1.4% of 429 research articles published in three leading psychiatry journals disclosed the use of artificial intelligence.

The aim was to assess how frequently artificial intelligence (AI) use is reported in psychiatric publications despite journals' disclosure policies. A cross-sectional review was conducted of research articles with abstracts published in Lancet Psychiatry, JAMA Psychiatry, and American Journal of Psychiatry between January 1, 2024, and March 7, 2026; letters, editorials, and commentaries were excluded, and articles were manually screened for the terms 'AI' and 'artificial intelligence'. The primary outcome was the proportion of research articles reporting AI use in manuscript preparation. Only 6 of the 429 articles (1.4%) disclosed AI use, and all disclosures appeared in Lancet Psychiatry (6 of 174 articles, 3.4%); no disclosures were identified in JAMA Psychiatry (99 articles) or American Journal of Psychiatry (156 articles) (0%). Disclosures were limited to improving language and clarity, proofreading, and restructuring text; no use was reported for data analysis, interpretation, or discussion. The observed rate of 1.4% is below the rates of 3.3% to 5.7% reported for submissions to major medical journals and suggests underreporting; the abstract noted that, in a 2023 Nature survey, 30% of researchers reported using AI for writing, editing, or summarizing the literature. Possible reasons include uncertainty about which uses require disclosure, inconsistent implementation of policies, and concerns about negative perceptions of AI-assisted work. In conclusion, clearer definitions, structured disclosure requirements, and consistent editorial oversight may strengthen transparency and research integrity.

All WCP 2026 summaries