Skip to content
OctymPowered By Proofig AI
For integrity & investigation teams

Every integrity and quality check, in one review, with evidence that holds up.

When a concern arrives, start from evidence rather than a blank page. A maximum-depth review of any paper, submitted or published, with every finding traced to its source and a record shaped for your committee.

What every review covers

One review, every lens.

A case rarely turns on one signal. The full scan runs across all three families of check, and your focus decides which of them leads.

Internal manuscript integrity

  • AI-generated text signals
  • Image manipulation
  • Image duplication
  • AI-generated images
  • Graph & chart duplication
  • Statistical & analytical consistency
  • Figure–legend consistency
  • Claim & citation consistency
  • Internal consistency
  • Prior publication & reuse disclosure

External checks & requirements

  • Image & graph plagiarism
  • Retracted reference detection
  • Hallucinated / non-existent references
  • Incomplete & broken references
  • Reference analytics
  • Data & code availability
  • Paper-mill signals
  • Authorship & affiliation consistency
  • Funder requirements
  • Submission requirements & completeness

Research quality

  • Methods & rigor
  • Reproducibility
  • Replicability signals
  • Research ethics
  • Novelty & contribution
  • Peer-review-grade analysis
Built for the case

Start from evidence, not a blank page.

A concern is a question that needs a defensible answer. Octym gives your team the full picture: fast, documented, and traced.

Evidence, assembled

A full evidence-backed review instead of a screenshot and a hunch: image forensics, text, references, data, every finding traced to its source.

Any paper

Submitted or already published. The review runs the same forensic depth on both.

Built for your committee

A record shaped for your standards: reviewable, defensible, and never an accusation. A person always decides.

Case file
Published · 2019
doi · 10.xxxx/xxxxxx · traced to source
  • Image duplicationtraced · Fig. 3b

    Matched against a figure in the published literature.

  • Image manipulationtraced · Fig. 1

    Splice boundary detected in the panel.

  • Reference → retractiontraced · Ref. 14

    Cited paper is flagged as retracted.

  • Text overlaptraced · p. 4

    Passage overlaps a prior publication.

  • Data consistencytraced · Table 3

    Reported values reconcile with the figure.

Every finding documented and traced to its source, built for a case file. A person adjudicates.

How a case runs

From concern to committee.

The full forensic scan always runs; your focus decides what leads and how deep the concern is pursued.

1

A concern arrives

A reader's letter, a whistleblower, a routine audit. Any paper, published or not.

2

Maximum-depth review

Every lens, forensic depth, with extra depth where the concern points.

3

A committee-ready record

Every finding documented with its evidence and source, shaped for your process.

Complete control

Your office sets the focus.

What leads, how deep, and how the record is shaped. You decide; your Octym team makes it happen.

You decide

What your reviews lead with (images, text, or data) and the depth each case demands.

Your Octym team builds it

New case types, new evidentiary standards, a different record format: configured for you.

It changes when you do

As your committee's standards evolve, the record evolves with them.

Image & figure integrity is powered by Proofig AI, a trusted leader in scientific image integrity, adopted by leading journals and trusted by the world's top researchers, publishers, and research institutes.

FAQ

Questions, answered.

Yes. The review runs the same forensic depth on a submitted manuscript or an already-published paper, whichever the concern points at.

Every finding is a signal backed by evidence and traced to its source, with a person making the final call. That supports defensible decisions rather than automated accusations.

Yes. The full forensic scan always runs, and your focus decides what leads and how deep the concern is pursued: images, text, or data. You decide; your Octym team makes it happen.

Yes. The record is shaped for your standards: what it documents, how findings are presented, and what your committee needs to review.

No, Octym does not decide misconduct: it assembles the evidence, and your committee decides what that evidence means. Octym is not a fraud detector and accuses no one. Octym runs its deepest review of the paper, with each finding traced to its source and exact location. Investigators confirm, edit or resolve every finding, each decision is logged to an audit trail, and the record is shaped for your committee, so each decision is reviewable and defensible and the risk of a false accusation is lower.

Yes, Octym surfaces the image and reported-number signals investigators check when fabrication is suspected, each with its evidence, never as a finding of fraud. Image forensics powered by Proofig AI flag cloned, spliced or retouched regions, images reused within the paper even when resized or rotated, duplicated graphs and AI-generated figures. In the numbers as reported, Octym flags p-values, test statistics and degrees of freedom that disagree, percentages impossible for the stated sample, and values that differ between text, tables and supplements. Your team weighs every signal.

Yes, Octym compares a paper's images and data visualizations against the published literature to surface figures reused from other papers. The check is powered by Proofig AI, the largest image-plagiarism database in the industry, and covers graphs, charts, diagrams and plots as well as photographs. The Science family of journals uses Proofig to screen images across all six of its titles. Each match is a signal traced to its source; whether it reflects permitted reuse, an earlier version or a problem is for your committee to decide.

Yes, Octym includes an AI agent your investigators can question in plain language, and every answer is grounded in the paper under review. Ask why a figure was flagged, dig into a reference or summarize the reproducibility gaps; each answer links to its source. Every finding also shows its evidence and exact location, so investigators can check it themselves and confirm, edit or resolve it, with an audit trail. A general-purpose chatbot answers without grounding; a single-purpose tool returns its flag without an agent you can question.

Yes, Octym keeps the papers you upload for a case confidential. They are analyzed on secure, private servers and never used to train AI models, by Octym or by its AI providers, whose contracts prohibit it. Processing is automated, staff cannot browse or open documents in the ordinary course, and access to a specific document is limited to a small number of authorized senior staff when you or your organization ask for help with it. Data is encrypted in transit and at rest.

Octym flags signals of undisclosed reuse, checking what a manuscript discloses about preprints, earlier publications and reused datasets. Octym flags results that appear reused from the authors' related work without a clear explanation. Figures are compared against the published literature, powered by Proofig AI, so a reused image can surface even when the text is silent about it. Undisclosed reuse is also one of the paper-mill signals Octym reports, each with its evidence, never a conclusion about the authors.

Yes, Octym checks the references of an already-published paper with the same depth as a submitted manuscript. Reference checks powered by RefGuard, Proofig AI's reference integrity product, flag citations to retracted papers, with the retraction date, nature and reason. They also flag references that cannot be matched to a real publication and citations that do not support their claims. Each finding is traced to its source, in a record shaped for your committee, and a person decides what it means.

See it on a case.

Contact us. We'll configure Octym for your committee's standards.

Analyzed on private servers. Never used to train models.