Evidence for the editor, never a verdict.
Abstract
AI Integrity Check screens every manuscript on arrival for text similarity, duplicated or altered images, references that cannot be found or have been retracted, and indicators of AI-generated text. Results are reported with their evidence and their limits, for editors to judge in context.
- Four checks on every manuscript, in one report
- Each flag shows its evidence, its source and its uncertainty
- Editors investigate and decide; a flag is never proof of misconduct
Possible reuse flagged for the handling editor. Ask the authors for the original images before any conclusion.
Text similarity, with context
Similarity is reported by source and by section. Overlap in a methods section with the authors' own cited prior work reads very differently from overlap in results or discussion, and the report makes that distinction visible.
- Breakdown by source and manuscript section
- Own prior work and cited text identified
- Quoted material and references excluded
Images and references
Figure panels are compared within the manuscript and against published figures for duplication, rotation, splicing and local alteration. References are looked up to confirm they exist and checked against retraction records.
When a reference cannot be found, the author is asked to verify it before review begins.
- Panel-level duplication and manipulation detection
- Retracted and corrected reference alerts
- Author verification requests sent from the report
- [12]Zhou et al. 2021, ISME J. 15: 1120Found
- [19]Fierer 2017, Nat. Rev. Microbiol. 15: 579Found
- [31]Patel & Gomes 2019, Soil Biol. 44: 210Not found
- [38]Lindqvist 2018, Appl. Soil Ecol. 9: 88Retracted
- [47]Moreau 2022, preprintNot found
Responsible use of AI checks
AI checks flag possible problems; editors investigate and decide. A flag is never treated as proof of misconduct. AI-text indicators are shown as uncertain signals only and are never the sole basis for rejection.
Unpublished manuscripts are confidential. They are never used to train AI models or shown to anyone outside the editorial process. Plascipub supports, but does not replace, your journal's own ethics policies and guidance from bodies such as COPE.
AI Integrity Check at a glance
- Similarity
- Published literature, preprints, the open web and your own archive
- Images
- Duplication, rotation, flipping, splicing and local alteration
- References
- Existence lookup, retraction and correction notices
- AI-text
- Low, medium or high signal with stated uncertainty
Every journal on Plascipub starts the same way: one letter, fifty free submissions, and an editorial office ready the same week.
- 50
- submissions free, no card needed
- 1 week
- typical setup for a new journal
Dear Editor, we are pleased to accept your journal.
Your submissions, reviews, integrity checks, proofs and DOIs can live in one place from today. No revisions required. Start free, bring your back issues when you are ready, and keep your own branding throughout.
