aipaperdetect
18 agents · nothing stored

Know where your paper stands before peer review finds the problems.

Upload a manuscript and get evidence-backed findings on AI-generated text, fabricated citations, and weak statistics in under two minutes.

Start free3 checks/month, no card required
report · manuscript.pdf7 agents
OVERALL SCORE
68/100
Needs revision
AI Detection32
Citation Consistency87
Reference Integrity91
Source Retrievability52
Reference Relevance41
Reference Relevance41
REFERENCE · [12]

Cited work is off-topic for the paper's stated claim in §3.2.

See full finding
File deleted within 120sAuthor info stripped before AINo training on your data

Built for academic work

Whether you're preparing your first conference paper or reviewing submissions for a top journal, get a structured second opinion in minutes.

Students

Submitting your first paper to a journal or conference? Catch citation errors, writing issues, and AI-detection flags before a reviewer does.

  • Avoid rejections from fixable issues
  • Check if AI-assisted writing is detectable
  • Verify every reference is real and accessible

PhD researchers

A pre-submission quality check for journal manuscripts. Know exactly which sections need work before weeks in review limbo.

  • Structured feedback in under 2 minutes
  • Identify weak citations before peer review
  • Score novelty against recent literature

Peer reviewers & editors

Screen submissions efficiently with a structured first pass on integrity, AI generation, and broken references before you read the full paper.

  • Fast red-flag detection on any submission
  • Fabrication check via Crossref & arXiv
  • AI-generation score with per-section evidence

Universities & labs

Supervise student manuscripts, enforce lab standards, or batch-check thesis drafts with Pro's 3 concurrent analyses.

  • Verify student work before co-signing
  • Ethics & compliance audit for grant submissions
  • Statistical integrity checks for clinical research

18 agents, checked in parallel

Each agent runs independently on only the section of your paper it needs, keeping analysis fast and data exposure low.

FREE · STARTER · PROCore
Writing QualityRegister, clarity, and AI-writing markers.
Citation ConsistencyMatch every in-text citation against the bibliography.
AI DetectionDetect LLM-generated passages with calibrated confidence.
Reference IntegrityVerify DOIs & arXiv IDs via Crossref and arXiv APIs.
Source RetrievabilityValidate references are accessible and URLs resolve.
Reference RelevanceJudge fit to target venue and stated research aims.
Academic IntegrityMethodology, result reporting, and scholarly honesty.
STARTER · PROQualitative
Scientific QualityMethodology soundness, rigor, reproducibility, and statistical validity.
Novelty AssessmentContribution assessment vs. state-of-the-art (web-verified).
PRO ONLYPro
Statistical IntegrityCross-reference numbers, verify statistical validity, flag suspicious results.
Ethics ComplianceCheck IRB approval, informed consent, dual-use risks, funding disclosure.
Literature CoverageIdentify gaps in related work and assess citation diversity.
PRO ONLYIntegrity screeners
Tortured PhrasesDetect papermill fingerprints: mangled terms and template writing.
Retraction CheckCheck cited references against Crossref/Retraction Watch and PubMed.
Duplicate SubmissionSearch for prior versions, duplicate publication, salami-slicing.
GenAI DisclosureAudit AI-use disclosure against venue policy and detection evidence.
PRO ONLYSynthesis
Author FeedbackStructured improvement guidance.
Reviewer ReportSimulated peer review.
See full plan comparison →

Three steps to a stronger submission

1
Upload

Drop a PDF, LaTeX, or plain text file up to 20 MB. We extract text automatically and strip author PII before a single agent reads it.

Accepted: .pdf · .tex · .txt
2
Evaluate

Standard runs 7 core agents. Pro runs all 18 in parallel across 5 pipeline stages, scoring each 0-100.

Typical runtime: 70-110 seconds
3
Improve

Get a structured report with per-module scores, confidence notes, and specific findings. Your original file is deleted after analysis.

File deleted within 120 seconds

Your research stays yours.

Academic papers, especially unpublished manuscripts, contain sensitive intellectual property. We designed this with data minimization as a core principle.

Read the full privacy policy →
File deleted within 120 seconds
Your original manuscript is permanently deleted after analysis completes. Only scores and findings are retained.
Author identity stripped first
Before any text reaches an external AI model, we remove emails, the author/affiliation block, and acknowledgements.
Each agent sees only what it needs
The fabrication checker receives only your bibliography; the style evaluator receives text samples, not the full paper.
No training on your data
Gemini and Claude are accessed via API under terms that prohibit training on inputs, routed through OpenRouter.

Frequently asked

What file formats does it accept?

PDF, LaTeX (.tex), and plain text (.txt) up to 20 MB. Text is extracted automatically from PDFs, no preprocessing needed.

Is my unpublished paper kept confidential?

Yes. Your original file is deleted from our server within 120 seconds of analysis completing. Before content reaches an AI model, we strip author names, emails, and affiliations. AI providers are contractually prohibited from training on API data.

How accurate is the AI detection?

We report a calibrated 0-10 score, not a binary label. On our internal ground-truth benchmark (51 papers: provably pre-LLM human papers vs. fully LLM-generated ones), the detector separated human from AI writing with zero false positives at the default threshold. That sample is small, so we don't advertise a headline accuracy percentage — and scores in the mid-range should always be treated as inconclusive, not as verdicts.

What's the difference between Free and Pro?

Free gets the 7 core agents (3 checks/month). Starter ($9/month) adds the 2 qualitative agents, scientific quality and novelty, with 20 checks/month. Pro ($25/month) adds 3 advanced modules plus 4 integrity screeners, 100 checks/month, and 3 concurrent analyses.

What counts as one 'check'?

One upload of one manuscript, regardless of how many modules run or how long the paper is. Uploading an edited version again counts as a separate check.

Can I use this for thesis review?

Yes. Upload individual chapters (recommended) or the full document. The system handles long-form academic writing and large bibliographies. PDF export is available on Pro.

Screen your next paper in minutes.

Free for 3 papers per month. No credit card. Upgrade to Pro for advanced audits.

Start free