Check My Thesis logoCheckMyThesis
NewScan free, pay to unlock

Don't fail your thesis
because of fake AI citations.

Upload your draft and get an instant integrity check.
See what's broken before your supervisor does.

Cross-checked against Semantic Scholar, OpenAlex, CrossRef, arXiv & PubMed
TODO
Fix broken refs in ch. 3 😬
Fake AI citation
[Chen et al., 2022]
doesn't appear in Nature Methods vol. 19.
AI Detector● Human
2 paragraphs flagged for review
Integrity Score
94
/ 100
Ready to submit ✓
Avoid AI fakes

Don't trust AI citations

AI-generated citations often mix up authors, dates, journals, or even make up sources entirely.

The Problem
ChatGPT generates convincing fake citations
References

Hartman, R. J., & Liang, P. (2019). The role of spaced repetition in undergraduate exam performance: A meta-analysis. Journal of Educational Psychology Review, 41(2), 188-204.

Sweller, J. (1988). Cognitive load during problem solving: Effects on learning. Cognitive Science, 12(2), 257-285.

Okafor, C., & Bennett, S. M. (2021). Dual-coding strategies and multimedia learning in digital classrooms. Journal of Digital Learning Research, 8(4), 332-349.

Roediger, H. L., & Karpicke, J. D. (2006). Test-enhanced learning: Taking memory tests improves long-term retention. Psychological Science, 17(3), 249-255.

Vance, D. R., Whitfield, A., & Romano, K. (2017). Interleaved practice and the illusion of competence in mathematics learning. Cognition and Instruction Quarterly, 29(1), 45-59.

Ask anything
Attach
Voice
ChatGPT can make mistakes. Check important info.

These citations look legitimate, but are they real?

Verify in seconds with Check My Thesis

The Solution
Check My Thesis flags fakes and AI hallucinations

Hartman, R. J., & Liang, P. (2019). The role of spaced repetition in undergraduate exam performance: A meta-analysis. Journal of Educational Psychology Review, 41(2), 188-204.

Fake

Sweller, J. (1988). Cognitive load during problem solving: Effects on learning. Cognitive Science, 12(2), 257-285.

Valid

Okafor, C., & Bennett, S. M. (2021). Dual-coding strategies and multimedia learning in digital classrooms. Journal of Digital Learning Research, 8(4), 332-349.

Fake

Roediger, H. L., & Karpicke, J. D. (2006). Test-enhanced learning: Taking memory tests improves long-term retention. Psychological Science, 17(3), 249-255.

Valid

Vance, D. R., Whitfield, A., & Romano, K. (2017). Interleaved practice and the illusion of competence in mathematics learning. Cognition and Instruction Quarterly, 29(1), 45-59.

Fake

3 out of 5 citations were hallucinated!

Replace fake citations in seconds with AI

Scan your citations now
HOW IT WORKS

Two checks, one upload,
zero surprises at submission.

Most popular

Catch every hallucinated citation

Upload your paper and we cross-check every reference against five academic databases. Hallucinated, outdated, or just misformatted, we flag it all.

  • Cross-checked against Semantic Scholar, OpenAlex, CrossRef, arXiv & PubMed
  • Preprints mapped to published versions
  • Export corrected BibTeX in one click
Try the citation verifier
Citation Report: thesis-draft.pdf
Integrity Score
87/ 100
12
Verified
5
Outdated
3
Not found
1
Efficient Training Methods for Deep Neural Networks
Smith et al. (2023) · Nature Comp. Sci.
Verified
2
A Framework for Multilingual Text Classification
Chen et al. (2022) · ACL
Verified
3
Optimizing Large-Scale Distributed Computing
Johnson et al. (2021) · AAAI
Verified
4
On the Convergence of Stochastic Gradient Methods
Johnson Mueller et al. (2020) · arXiv preprint
Outdated
5
Recent Advances in Natural Language Understanding
Garcia & Lee (2024) · —
Not found
AI Detection

Know exactly which sentences are flagged

Our sentence-level AI detector highlights exactly where AI-generated text appears, no vague percentages, just precise, actionable feedback.

  • Sentence-level detection for ChatGPT, Claude & Gemini patterns
  • Color-coded highlighting: human, mixed, and AI
  • Paragraph-level confidence scoring
Try the AI detector
AI Detection: chapter-3.pdf
78%
Human
Human
Mixed
AI

The transformer architecture has fundamentally reshaped how we approach sequence-to-sequence tasks in natural language processing. Since Smith et al. (2023), self-attention mechanisms have replaced recurrence as the dominant paradigm.

Recent work has shown that scaling model parameters yields consistent improvements in downstream benchmark performance across a wide range of tasks. Furthermore, the implementation of attention mechanisms enables models to capture long-range dependencies with unprecedented efficiency and accuracy.

However, significant challenges remain in addressing bias, hallucination, and factual grounding in model outputs.

BibTeX Formatter
Available
Citation Updater
Available
Plagiarism Checker
Coming Soon
WHAT YOU GET

Two checks, one upload, one number you can trust.

/ 01
Citation verifier
Every [ref] gets pulled, opened, and read. We tell you if the paper exists, if the quote matches, and if the page number is real.
/ 02
AI detector (for real)
Sentence-level scoring that doesn't just flag words like 'delve'. Tuned on 120k academic papers.
/ 03
BibTeX, fixed.
Outdated preprints get mapped to published versions. Export a clean .bib in one click, no more manual cleanup.
/ 04
Your draft stays yours
PDFs are deleted after your report. Nothing is added to a database. Your supervisor will never know.
PRICING

Simple pricing. Powerful thesis check.

Two one-time payments, no subscription: 3 checks for a single submission, or a month of unlimited use.

↻ No subscription🔒 Stripe-secured billing
PAY-AS-YOU-GO
Essential
€29.99
one-time

One-time payment. 3 full checks total, no expiry, no subscription. Buy again any time to top up.

All three tools: Citations, AI, Plagiarism
3 full checks per purchase
Checks never expire
Unlimited results per scan
Core academic databases
Email support
Get 3 checks
✦ BEST VALUE
ONE MONTH UNLIMITED
Complete
€49.99
one-time

One-time payment. Unlimited checks for 30 days, then access ends, no subscription, no surprise renewal.

Pay once. Access expires after 30 days, never auto-renews.

Unlimited checks for 30 days
All three tools: Citations, AI, Plagiarism
Priority processing (2× faster)
Premium academic databases
Advanced analysis engine
Priority email support
Get Complete (1 month)
All prices include VAT. Invoices by email. No setup fees.
FAQ

Frequently asked questions

Grammarly doesn't check citations. Turnitin only tells you about similarity. We actually read every reference in your paper and tell you if it exists, matches your claim, and isn't AI-hallucinated, plus sentence-level AI detection. One upload, less than a coffee.
Nope. Your draft is yours. We don't store PDFs after your report is generated, and nothing gets added to any database. Run it as many times as you want before submitting.
Our model is tuned for academic prose specifically and evaluated against 120k+ student papers. We show you a per-paragraph confidence score, not a scary red percentage meant to make you panic.
It's a proofreading tool, like Grammarly. It doesn't edit your text, it just shows you what needs fixing. You do the fixing.
Upload and preview your results free, a few flagged citations and AI-flagged sentences, no card. To unlock the full report it's €29.99 one-time for Essential (3 checks, never expire) or €49.99 one-time for Complete, 30 days of unlimited checks. Both are one-time payments, no subscription.
ONE LAST THING

Your deadline is close.
Your draft isn't ready.

Upload your PDF. 90 seconds later you know exactly what to fix.

Check my thesis now
No card needed to scan