ChatLarge language models generate convincing prose that sounds profoundly authoritative; over one-third of their academic citations are complete fabrications. By conducting a massive forensic audit of scholarly and legal references, computer scientists proved that generative AI routinely invents phantom research papers with realistic authors and fake digital identifiers.
In law firms, university classrooms, and research laboratories, professionals rely on generative AI to draft research memos and academic literature reviews. The writing sounds impeccably professional, but lawyers and scientists are getting caught submitting papers referencing books and court cases that do not exist.
A planetary-scale audit of over one hundred thousand AI citations revealed why this happens: language models operate on statistical plausibility, not factual reality. The AI weaves together famous real-world professors, plausible journal names, and synthetic digital numbers to invent phantom footnotes that sound entirely real.
This study established the urgent need for automated citation verification software. By integrating real-time DOI cross-checking tools, by educating courts and universities on AI hallucinations, and by preventing fake scholarship from polluting legal and scientific records, citation auditing protects academic integrity.
LLM hallucinations in the wild: Large-scale evidence from non-existent citations
Large language models (LLMs) are known to generate plausible but false information across a wide range of contexts, yet the real-world magnitude and consequences of this hallucination problem remain poorly understood. Here we leverage a uniquely verifiable object - scientific citations - to audit 111 million references across 2.5 million papers in arXiv, bioRxiv, SSRN, and PubMed Central. We find a sharp rise in non-existent references following widespread LLM adoption, with a conservative estimate of 146,932 hallucinated citations in 2025 alone. These errors are diffusely embedded across many papers but especially pronounced in fields with rapid AI uptake, in manuscripts with linguistic signatures of AI-assisted writing, and among small and early-career author teams. At the same time, hallucinated references disproportionately assign credit to already prominent and male scholars, suggesting that LLM-generated errors may reinforce existing inequities in scientific recognition. Preprint moderation and journal publication processes capture only a fraction of these errors, suggesting that the spread of hallucinated content has outpaced existing safeguards. Together, these findings demonstrate that LLM hallucinations are infiltrating knowledge production at scale, threatening both the reliability and equity of future scientific discovery as human and AI systems draw on the existing literature.
Neoclassical economists spent two centuries assuming that humans calculate money like rational, emotionless computing robots; Daniel Kahneman and Amos Tversky proved that the human brain feels the pain of losing 100. Awarded the 2002 Nobel Prize in Economics, Prospect Theory shattered rational-choice theory and founded Behavioral Economics, revolutionizing financial markets, consumer design, and government policy worldwide.
Philosophy taught that human reason was an immaculate engine of pure logic and objective truth; Amos Tversky and Daniel Kahneman proved that our daily decisions are run by quick mental shortcuts that systematically trick our brains into predictable cognitive illusions. Published in Science in 1974, "Judgment under Uncertainty" mapped the universal cognitive biases governing the human mind, revolutionizing behavioral psychology, courtroom legal jury decisions, and medical diagnostics.
Detecting a single viral gene or defective hereditary mutation in a blood sample was like finding a single needle in an ocean of biological noise; the Polymerase Chain Reaction amplifies a specific target DNA sequence into billions of identical copies in just a few hours. This thermal cycling discovery became the bedrock of modern medical diagnostics, forensic crime analysis, and pandemic response.
Building peptides in liquid solution required painstaking, months-long crystallization after every single amino acid addition; Bruce Merrifield anchored growing peptide chains to insoluble plastic beads, allowing unreacted chemicals to be washed away in seconds. This ingenious solid-phase paradigm automated the chemical assembly of proteins and laid the foundation for modern peptide therapeutics like GLP-1 weight-loss drugs.
Economists once believed that human competition was a pure zero-sum fight where one person’s win required another’s total loss; John Nash proved in a single page of mathematics that all competing systems reach a stable state of mutual balance. Published in 1950 when Nash was just twenty-one, the Nash Equilibrium earned him the Nobel Prize and created the strategic grammar governing global economics, geopolitics, and evolutionary biology.
91816-6/v1.webp)
Doctors spent a century insisting that painful stomach ulcers were caused by spicy food, nervous stress, and excess stomach acid; Barry Marshall and Robin Warren proved that ulcers are caused by a resilient spiral bacterium living in the stomach lining. Awarded the 2005 Nobel Prize in Medicine after Marshall drank a beaker of bacterial culture to prove the medical establishment wrong, this discovery turned stomach ulcers from a chronic lifelong surgery into a simple two-week antibiotic cure.
Factory engineers once tuned automatic industrial valves through stressful trial-and-error that frequently blew out boilers and ruined chemical batches; John Ziegler and Nathaniel Nichols created simple mathematical formulas to tune PID controllers for optimal stability. Published in 1942, the Ziegler-Nichols tuning rules established the operational foundation for modern industrial automation, controlling over ninety percent of all automated factory loops, chemical plants, and thermostats in the world.
Niels Bohr taught that observing a quantum particle mysteriously collapses all alternate possibilities out of existence; Hugh Everett III proved that the mathematical equations of quantum physics never collapse at all, meaning every possible outcome actually happens in a branching parallel universe. Published in 1957 to career-ending ridicule, Everett’s "Many-Worlds Interpretation" is now one of the dominant frameworks of theoretical physics, driving modern quantum cosmology and quantum computing theory.
Ask this paper your own questions, or keep browsing the verified research catalogue.
