Drop files here or click to choose
.tex .bib .bbl .pdf .txt — multiple files welcome (e.g. main.tex + refs.bib)
No files handy? — loads a sample paper with 4 chat artifacts and a fabricated citation planted in it.
Checking…
1 · AI-generated-text findings
| Severity | File | Line | Problem | Evidence (sentence) | Fix |
|---|
- High (chat artifacts) — pre-selected for removal
- Low (style tells: em-dashes, semicolons, stock phrases) — flagged only, never auto-removed
- Quoted / table / verbatim matches — treated as deliberate examples, review-only
2 · Citation verification
| Status | Key | Cited as | Checked against | Problem / proposed correction | Action on generate ⓘ |
|---|
- Action on generate — what happens when you press “Generate”: keep · auto-correct fields · delete entry + its \cite. Preset from the status; override freely.
- Not found in all main databases = classic signature of a fabricated citation
- Books / websites / software → unverifiable, never auto-removed
- Unmatched? Use the row’s search ↗ links (Google Scholar, ACM DL, ResearchGate)
3 · Report & fixed files
What was changed
Who made this?
Built by Prof. Zangir Iklassov and Kyriakos Tsourekas (student).
⭐ 2 seconds to support us
If verifAI saved your submission, star the repo — it’s how other researchers find it before their deadline.
★ Star on GitHub📚 Using verifAI in your research?
Cite us — paper coming soon; until then cite the tool:
@misc{verifai2026,
title = {verifAI: Catching AI-generated Text and
Hallucinated Citations in Research Papers},
author = {Iklassov, Zangir and Tsourekas, Kyriakos},
year = {2026},
howpublished = {\url{https://zangir.github.io/verifAI/}},
note = {Paper forthcoming}
}