CiteLM
CiteLM provides an AI-powered citation audit platform that scans academic papers against their sources in seconds, flagging misquoted numbers, wrong attributions, review-vs-original errors, and hallucinated references. The platform automatically finds referenced articles on the web or connects to a user's Zotero library, marks issues inline within the document, and offers an AI chat interface for tailored suggestions. It is benchmarked on independent public datasets, achieving 17% higher macro-F1 than the best published result for citation accuracy.
- Artificial Intelligence
- Software Only
Funding
Founders
Founder details are not available yet.
Product
Problem
Academic peer review is struggling to keep pace with the volume of published research, leaving citation errors—including hallucinations and fabricated references—undetected. Researchers frequently encounter misquoted figures and sources that lead nowhere, eroding trust in scientific literature. A 2026 Nature analysis found that tens of thousands of academic publications may contain fake or faulty references generated by AI.
Solution
CiteLM offers an AI-powered citation audit platform that scans papers against their sources in seconds. After uploading a manuscript, the AI automatically finds referenced articles on the web or connects to a user's Zotero library, then flags misquoted numbers, wrong attributions, review-versus-original errors, and fully hallucinated references. Incorrect citations are marked inline within the document, showing exactly which sentence has a problem and why. Users can chat with the AI for specific suggestions on every flagged citation, helping them understand issues and locate the correct source. The platform is benchmarked on independent public datasets for both hallucinated reference detection and source-support verification.
Target Audience
Primary users are academic researchers, university students preparing theses, and peer reviewers or journal editors seeking rigorous, automated citation verification.
Features
- Automated citation verification across claims and sources, catching misquoted numbers, wrong attributions, review-vs-original errors, and hallucinated references
- Web-based reference discovery without manual searching or downloading, plus Zotero library integration
- Inline document marking of incorrect citations for precise identification of problematic sentences
- AI chat interface providing tailored suggestions and explanations for each flagged citation
- Benchmarked on independent public datasets, achieving 17% higher macro-F1 than the best published result (MultiVerS) for citation accuracy