MDPI begins AI integrity screening across 2,000 daily manuscript submissions

Ethicality now checks every manuscript and peer review report for potential paper mill activity, fabricated references, manipulated text and authorship concerns.

MDPI’s Ethicality system now screens about 2,000 manuscript submissions each day for potential research integrity concerns

MDPI has fully deployed its Ethicality research integrity system across all manuscripts submitted to its journals, placing automated AI checks inside an editorial workflow that processes about 2,000 submissions from researchers worldwide each day.

The in-house system examines manuscript text, references, author information and peer review reports for potential integrity issues. These include suspected paper mill activity, fabricated submissions, AI-generated or manipulated writing, fake references, citation irregularities and unusual authorship patterns.

Ethicality also screens peer review activity for suspicious behavior and potential AI-generated content. MDPI says every case flagged by the system is passed to an experienced editor or research integrity professional for human review before any action is taken.

The Basel-based open access publisher announced the completed deployment on June 17, 2026. Ethicality is now applied to submissions across MDPI’s journal portfolio rather than being restricted to selected titles, subject areas or trial groups.

The system remains active throughout the editorial process, monitoring manuscripts from their initial submission through to the publication decision.

Manuscripts and reviews checked within one system

Ethicality divides each submission into components including its title, abstract, author metadata, main text and references before producing an integrity assessment.

The checks are intended to identify patterns that may not be visible through conventional plagiarism screening alone. Fake citations, abnormal referencing behavior, inconsistent author identities and coordinated peer review activity can require comparisons across several parts of a submission or across multiple records.

MDPI says Ethicality has been trained and refined using patterns observed in real submissions. The publisher describes it as an end-to-end integrity layer rather than a single check performed when a manuscript first enters the system.

Dr Milos Cuculovic, head of technology innovation at MDPI, says publishers need to detect potential problems before research enters the scientific record: "What is becoming clear is that traditional, manual processes are no longer sufficient in peer review. The industry needs to shift from reactive approaches, resolving issues after publication, to proactive systems that support editors earlier in the workflow."

He adds: "AI, when used responsibly, acts as a set of guardrails rather than a substitute for human judgment."

MDPI continues to use separate third-party systems alongside Ethicality. Proofig is used to identify potential image manipulation, while submitted manuscripts are checked through iThenticate for duplicated text and possible plagiarism.

Human editors retain responsibility for decisions

The publisher says Ethicality generates risk signals but does not independently reject papers, determine whether misconduct has occurred or replace editorial assessment.

Flagged submissions are reviewed by editors or research integrity specialists, who decide whether further investigation or action is required. This distinction is important because the presence of AI-generated text, an unusual citation pattern or an identity discrepancy does not automatically establish research misconduct.

The system is also being used to reduce the volume of repetitive technical checks handled manually. MDPI says tasks including reference validation, formatting assessment and initial technical triage can be automated so editors and reviewers spend more time examining a paper’s research and scientific quality.

Dr Enric Sayas, product owner of Ethicality, says increasingly sophisticated fraudulent material is making existing detection processes harder to operate at scale: "We are in a technological race. As generative AI makes it easier to produce sophisticated plagiarism and high-quality fake papers, traditional detection methods are no longer sufficient."

He says publishers risk being overwhelmed by fraudulent submissions without systems capable of identifying manipulated content, inconsistent data and AI-generated material before peer review is completed.

AI scrutiny extends into peer review

Ethicality’s inclusion of peer review reports extends the screening process beyond author submissions. MDPI says the system can identify suspicious review patterns and text that may have been generated by AI.

The deployment also comes as academic publishers, universities and researchers confront wider questions about acceptable AI use in manuscript preparation. MDPI’s system is designed to identify potential concerns for editorial examination, rather than treating all AI-supported writing as evidence of misconduct.

Ethicality is now operating across MDPI’s full submission workflow and processing about 2,000 manuscripts each day. Its impact will depend on the quality of its risk signals and the decisions made by the human editors reviewing them.