A developer reports that software screening of 92 published papers already surfaced five cases of likely data fabrication, prompting two corrigenda and one imminent retraction, and will now be applied to 20,000 papers. Routine, automated pre‑ and post‑publication screening could become a scalable layer of scientific fraud detection.
— If automated tools can reliably flag suspect data at scale, journals, funders, and governments may need to mandate systematic screening, reshaping research oversight and trust.
Scott Alexander
2025.10.13
100% relevant
ACX grant to Markus Englund: 'already scanned 92 published papers' and found irregularities in five, with plans to scan 20,000 more.
2023.07.18
86% relevant
Nature discusses detective work that uncovers suspicious trials; that connects to the idea that automated screening tools can identify fabricated or inconsistent data at scale and should be deployed as a routine layer of fraud detection for published clinical research.
← Back to All Ideas