The NeurIPS 2025 Citation Scandal: 100+ Fake References That Fooled the World's Top AI Researchers

## The Scandal That Shook AI Research

In January 2026, Fortune broke the story: over 100 AI-hallucinated citations were discovered in NeurIPS 2025 papers. These were completely fabricated references — fictional authors, nonexistent journals, phantom papers.

## The Scale

### What Fake Citations Looked Like

Plausible Author Names: Common names in AI research, combining real first names with different last names.

Convincing Paper Titles: "Efficient Transformer Architectures for Large-Scale Knowledge Distillation" — sounds like it should exist.

Fake Journal Names: "International Journal of Neural Computation and Applications" — close variations of real journals.

Fabricated DOIs: Correct format but resolve to nothing.

## How This Happened

The modern AI-assisted research pipeline: researchers ask ChatGPT for papers, AI generates citations (some hallucinated), researchers integrate without verification, peer reviewers don't check every citation.

### The Journal Editor's Warning

Alison Johnston posted on LinkedIn: "I've rejected 25% of submissions thus far this year, because of fake references." One in four submissions.

## The GEO Connection

If fake citations fool NeurIPS reviewers, they absolutely fool AI search engines. LLMs evaluate authority using the same patterns: academic formatting, statistics, named sources, cross-referencing.

### How Black Hat Operators Are Adapting

1. Fake "industry studies" cited across marketing blogs
2. Phantom whitepapers in product comparison content
3. Fabricated expert endorsements with academic credentials
4. Citation inflation — misrepresenting real studies
5. Preprint abuse — fake studies on preprint servers

### The Training Data Contamination Risk

NeurIPS papers become AI training data. Ghost citations in published research get embedded in future models' knowledge. This creates a feedback loop where models learn to trust fake sources.

## Lessons for GEO Defense

For Content Creators: Always verify citations — check DOIs, author existence, journal legitimacy.

For AI Platforms: Implement citation verification in RAG pipelines, cross-reference against CrossRef and Semantic Scholar.

For Researchers: Verify every AI-generated citation, use established citation managers, advocate for automated verification.

Sources: Fortune — NeurIPS AI-Hallucinated Citations↗; Rolling Stone — AI Inventing Papers↗

Related: Ghost Citations: The Complete Guide | Ghost Citations Detector Tool