Ridiculous alert: Hallucinatory citations have been found in research papers released by NeurIPS, the prestigious artificial intelligence conference

AI detection startup GPTZero scanned all 4,841 papers accepted at the prestigious Neural Information Processing Systems (NeurIPS) conference, held last month in San Diego. The company found 100 hallucinogenic citations in 51 research papers that it confirmed were fraudulent, the company told TechCrunch.

Getting a paper accepted by NeurIPS is an accomplishment worthy of a CV in the world of AI. Given that these are the leading minds in AI research, one would assume they would use the MBA for the disastrously tedious task of writing citations.

So caveats abound about this finding: 100 confirmed hallucination citations across 51 papers are not statistically significant. Each paper contains dozens of citations. Out of tens of thousands of citations, that’s statistically zero.

It is also important to note that an inaccurate citation does not negate the paper’s research. As NeurIPS told Fortune, which was the first to report on the GPTZero research, “Even if 1.1% of papers contain one or more incorrect references due to the use of MBA, the content of the papers themselves [is] It is not necessarily invalidated.”

But having said all that, the fake quote isn’t nothing either. NeurIPS prides itself on “rigorous scientific publishing in the field of machine learning and artificial intelligence,” she says. Each paper is peer-reviewed by several people who have been instructed to report hallucinations.

Citations are also a kind of currency for researchers. They are used as a functional measure to show the impact of a researcher’s work among his or her peers. When artificial intelligence creates them, it reduces their value.

No one can blame peer reviewers for not being able to catch a few AI-fabricated citations, given the sheer volume involved. GPTZero is also quick to point this out. The goal of the exercise was to provide concrete data on how AI has seeped through the “submission tsunami” that has “strained these conference review lines to the breaking point,” the startup says in its report. GPTZero also points to a May 2025 paper titled “The Peer Review Crisis in AI Conference” that discussed the issue at early conferences, including NeurIPS.

However, why were the researchers themselves unable to validate LLM’s work for accuracy? They certainly should know the actual list of papers they used in their work.

What it all points to is one big, ironic fact: If the world’s leading AI experts, with their reputations on the line, can’t guarantee the accuracy of their MBA use in detail, what does that mean for the rest of us?