FABULEUX: Fighting Fabrication and Misuse of LLMs in Scientific Writing

The Fabuleux project addresses a pressing challenge in science: ensuring the reliability and integrity of research writing in the era of large language models (LLMs). While LLMs are increasingly adopted by researchers to assist in writing and revising scientific texts, their use also introduces serious risks, from fabricated content and distorted citations to the large-scale generation of fraudulent papers. Fabuleux aims to foster the responsible use of LLMs in scientific writing while curbing their misuse, by creating methodologies to detect, analyze, and control problematic behaviors in LLM-assisted scientific writing. It pursues two complementary objectives: 1) to make LLM-assisted revision more reliable and beneficial for researchers, by detecting and mitigating fabricated or misleading content, while providing actionable feedback that helps authors improve their writing skills; and 2) to identify unreliable or fraudulent research, by designing methods and tools that help the scientific community detect questionable publications, reinforce ethical practices and safeguard the integrity of scholarly communication.

Consortium

LS2N Inria and Nantes Université ISIR Sorbonne Université and CNRS Inria Paris ALMAnaCH team LORIA Université de Lorraine, Mosaik team

News