Proofig AI strengthens the submission and review process by automatically screening scientific images for suspected duplication, manipulation, AI-generated content, and plagiarism before peer review begins. It plugs a gap that text-focused tools cannot address, giving editors a structured, low-noise basis for human review and earlier, more consistent image integrity checks.
AI tools are reshaping editorial workflows, but most focus on text: language editing, plagiarism detection, AI-text screening. Image integrity remains a distinct, underserved gap. Proofig AI’s image integrity platform addresses this gap by screening manuscript figures for duplication, manipulation, AI-generated content, and plagiarism before peer review begins, surfacing suspected issues for human review.
The Image Integrity Gap in Editorial Workflows
When publishers discuss AI-assisted editorial tools, the conversation typically centers on text-based capabilities. Image integrity occupies a fundamentally different problem space, one that requires specialized detection methods and domain-specific training data. The scale of the problem is substantial. According to data from UKRIO and image integrity analyst Jana Christopher, roughly 20 to 35 percent of life-science manuscripts submitted to journals are flagged for image-related issues at the submission stage. Many of these issues go undetected until after publication.
Many of the problems flagged at submission are not caught before publication, and a meaningful share make it through peer review and into the published record. The gap between what is submitted and what is caught beforehand represents a structural vulnerability in the review process, not a failure of individual editors.
This is not a problem that better training alone can solve. It requires purpose-built tools integrated into the editorial workflow.
Why Are Image Issues Hard to See by Eye?
Journal editors and peer reviewers are experts in scientific content, not forensic image analysis. The types of issues that matter, duplication across pages, partial duplication with rotation, subtle cloning within a panel, splicing between figures, or AI-generated microscopy indistinguishable from real data, are often invisible to the untrained eye.
A 2025 peer-reviewed study published in Research Integrity and Peer Review tested medical students and researchers on their ability to detect image duplications. Researchers correctly identified only 32.4% of duplications (median 11 out of 34 possible duplication events), with no significant difference from students. This finding underscores a critical point: even trained scientists miss the majority of image integrity issues when reviewing figures manually.
How Do Rising Submission Volumes Amplify the Problem?
Rising submission volumes compound the challenge. A single manuscript may contain dozens of figures, each with multiple panels. Comprehensive manual review of every panel for every submission is not feasible at the scale most journals now operate. As the volume of submissions grows, the gap between what needs to be checked and what can realistically be checked by human reviewers widens. This is where automated, workflow-integrated screening tools like Proofig AI become essential, not as replacements for editorial judgment, but as a way to ensure that image integrity checks keep pace with submission rates.
What Does Proofig AI Detect, and How Does It Work?
Proofig AI is designed specifically for scientific image integrity. Its core capabilities cover the major categories of image-related issues editors need to screen for:
| Capability | What It Detects |
|---|---|
| Duplication and reuse analysis | Scaling, rotation, flipping, full and partial overlap within a manuscript |
| Manipulation detection | Cloning, splicing, deletion, content-aware edits within a single image |
| AI-generated image identification | Synthetic images across microscopy, histology, Western blots, cell plates, medical scans, and more |
| Image plagiarism (PubMed source) | Reuse of sub-images from published manuscripts, checked against tens of millions of images |
| Self-plagiarism control (My Database) | Comparison against a researcher’s own prior publications |
The workflow is straightforward. An editor or integrity team uploads a manuscript (PDF or image files). Proofig scans all figures automatically, flags suspected issues with forensic detail, and generates a report for review. The tool does not make final determinations about intent or misconduct. It surfaces potential issues so that editors and integrity officers can investigate further.
The threat from AI-generated images is growing. As Nature reported in 2024, AI image-generation tools can now produce convincing fraudulent scientific figures that are difficult to distinguish from real microscopy and other modalities. Proofig’s AI-generated image detection covers the most widely used generative models and is continuously updated as new models emerge.
How Does Proofig Address the False-Alarm Problem?
One legitimate concern editors raise about automated screening is false positives. Flags that do not represent real issues waste editorial time, create friction with authors, and erode confidence in the tool. Proofig addresses this directly.
Proofig’s false-positive rates have been validated as very low across large datasets of real scientific images, covering hundreds of thousands of microscopy and Western blot images. The tool’s type-specific detection approach accounts for the distinct visual characteristics of each image modality, which contributes to its precision.
For editors, this translates to a practical benefit: the tool surfaces genuine concerns without generating noise that burdens editorial staff or damages author relationships.
Who Is Already Using Proofig AI?
Proofig’s adoption by major publishers provides concrete evidence of its editorial value. All six journals in the Science family adopted Proofig AI after a multi-month pilot. In his editorial announcing the adoption, Science Editor-in-Chief H. Holden Thorp described Proofig as analogous to iThenticate for text plagiarism, filling a parallel need for image integrity. At Science, Proofig reports are now reviewed before external peer review, giving editors a standardized way to address image concerns early instead of relying on ad hoc checks.
MDPI signed a multi-year deal to integrate Proofig across its portfolio. Milos Cuculovic, Head of Technology Innovation at MDPI, stated: “From the experts, it seemed Proofig AI was by far the best choice, and here we are: further ensuring research integrity and guaranteeing the high quality of research output published by MDPI.”
These are not pilot programs. They represent integration at scale by publishers processing thousands of submissions.
When Image Integrity Is Part of a Broader Workflow
For publishers seeking a comprehensive manuscript quality assurance layer, Proofig offers PubShield, an institutional submission hub that combines multiple integrity checks in one dashboard. PubShield pairs Proofig AI’s image integrity screening with text similarity checking (powered by iThenticate/Turnitin), AI-generated text detection (Pangram), reference integrity analysis (RefGuard), and data reporting compliance (DataSeer).
To be clear: these partner capabilities belong to PubShield as an integrated hub. They are not standalone Proofig AI features. PubShield exists for institutions and publishers that want a single workflow covering the full spectrum of manuscript integrity checks.
Why Pre-Review Screening Matters for Authors and Editors Alike
Many image integrity issues are unintentional. They are introduced during figure preparation, through collaborator workflows, by students assembling panels, or through accidental reuse across projects. Even researchers who uphold the highest ethical standards may unknowingly submit figures with problems they did not cause. This is precisely why screening before peer review is valuable: it catches potential issues at a stage when they are still correctable, before they become post-publication corrections or retractions.
The consequences of missed issues are well documented. Nature has reported on the significant personal and career stress researchers face when retracting papers, often due to honest mistakes that pre-submission screening could have prevented. For journals, post-publication retractions carry significant financial and reputational costs that compound over time.
Practical Takeaway for Journal Editors
Editors do not need to become image forensics experts. What they need is a reliable, low-friction tool that surfaces suspected issues before peer review begins, reduces the risk of post-publication corrections and retractions, and integrates with existing submission systems. Proofig integrates with Aries Systems for workflow management, fitting into the infrastructure many publishers already use – optimised for automation and scale. Editors interested in learning more can contact their Aries Account Coordinator or visit the Proofig image integrity platform overview.
The goal is not to assume wrongdoing. It is to support human review by catching potential issues early, when they are still correctable, and to do so with a tool that produces very few false alarms. For editors managing rising submission volumes and increasing scrutiny of published figures, automated image integrity screening is becoming an essential part of responsible publishing practice.
