Search Result

Background: The academic communication is being attacked by paper mills, altered peer review, fake data, plagiarised content, hijacked journals, citation engineering, and undisclosed machine-generated content. However, the retractation data should be taken with a pinch of salt, as it contains both misconduct and honest error. As a result, it does not measure the prevalence of fake science. The study’s objective was to analyse recent patterns of retraction, and develop a layered preventive framework for detecting unreliable publications while preserving due process and legitimate correction. According to Retraction Watch Database, methods used are retrospective cross-sectional bibliometric analysis done. The retraction records was filtered for 2020-2025, duplicate revisions were removed, and records were de-duplicated based on valid original-paper DOI or record identifier. Different reasons were explained, and a unique family hierarchy was useful for testing associations. We used Chi-square tests, Mann–Whitney U testing, and multivariable logistic regression to study whether the retraction patterns within 2 years of publication. Results: The final dataset included 37,344 records. The top reason for core family publication-manipulation, while peer-review concerns, paper-mill signals, computer-aided or generated-content tags increased sharply in some years. The time to retraction (TTR) was 593 days median. Although the statistical associations were significant, many effect sizes were small, suggesting that statistical significance should not be treated as practical significance. To produce authentic research, coordinating author, journal, institution, database, funder and regulator verification systems. While automated screening play a role in triage, transparent human adjudication, metadata validation, data availability, reviewer authentication and proportionate correction are still essential.