Zuyao Xu
Back to Research
ResearchIEEE S&P 2026 PosterCCF-NSS 2026 Poster

GhostCite: A Large-Scale Analysis of Citation Validity in the Age of Large Language Models

Zuyao Xu, Yuqi Qiu, Lu Sun, Fasheng Miao et al.

Measuring citation validity in the age of language models.

PDFPoster PDFPublication
The CiteVerifier framework and experimental pipeline (Figure 1).
The CiteVerifier framework and experimental pipeline (Figure 1). Source ↗

The problem

Scientific references are a foundation of verifiable research. GhostCite examines how fabricated or invalid references enter academic workflows as language models become common writing tools.

CiteVerifier and the study

CiteVerifier provides an automated citation-verification pipeline. The study combines model evaluation, analysis of published papers, and a researcher survey to examine the problem from several perspectives.

  • Evaluate citation generation by 13 language models.
  • Analyze roughly 2.2 million citations from 56,381 papers at AI/ML and security venues.
  • Connect automated verification results with researchers’ writing and reviewing practices.

Why it matters

The results motivate citation checks as part of academic writing and reviewing, alongside better tooling and shared responsibility across the research community.

More Research