AI isn't just making up answers anymore.

It's fabricating entire academic papers.

A Georgia State University professor was grading student work when he noticed something disturbing. The fake citations weren't just in student papers.

They were in published research.

Real journals. Real academics' names. Completely fabricated sources.

Here's how the contamination spreads:

𝗦𝘁𝗲𝗽 𝟭: ChatGPT invents a citation.

Fake journal.
Fake article.
Real-sounding author name.

𝗦𝘁𝗲𝗽 𝟮: A sloppy paper includes it without verification.

𝗦𝘁𝗲𝗽 𝟯: That paper gets published and cited by others.

𝗦𝘁𝗲𝗽 𝟰: The fake citation now appears in multiple "legitimate" sources.

𝗦𝘁𝗲𝗽 𝟱: Future researchers and LLMs treat it as verified.

The brutal truth: LLMs aren't malfunctioning when they hallucinate. They're doing exactly what they're designed to do, predict plausible text. They don't care about truth.

Research librarians now waste significant time chasing sources that don't exist.

AI-assisted Google searches reinforce the illusion that fake journals are real.

And the "publish or perish" pressure that already created predatory journals? AI just poured gasoline on it.

This isn't a student cheating problem anymore.

It's an information supply chain failure.
Source: LinkedIn