New AI-Powered Tool To Detect Fake Research Papers with 94 per cent Accuracy

The team behind the tool called Xfaxy analyzed 4,000 authentic research papers from PubMed, an open-access database managed by the US National Institutes of Health, and compared them with 300 fake papers generated by ChatGPT.

Hyderabad (Telangana): In a groundbreaking development, researchers from the State University of New York and Hefei University of Technology, China have created an advanced AI tool, ‘Xfaxy’, which can identify fake research papers with an accuracy of 94 per cent. This tool promises to revolutionize the way academic integrity is maintained in research.

The team behind Xfaxy analyzed 4,000 authentic research papers from PubMed, an open-access database managed by the US National Institutes of Health, and compared them with 300 fake papers generated by ChatGPT.

Previously, traditional data mining methods had lower accuracy rates, detecting fake documents with only 38 to 52 per cent precision. However, Xfaxy achieved a significant improvement in distinguishing real research from AI-generated content with an accuracy ranging from 80 to 94 per cent.

Xfaxy works by scanning writing patterns and content usage. It found certain phrases ‘climate change’ and ‘clinical trials’, commonly used in legitimate research, were rarely used in papers written by AI chatbots. This distinction highlights a fundamental difference in writing style as AI-generated papers often try to mimic genuine research more convincingly in contrast to human researchers who focus on presenting their experimental results and methodologies.

The introduction of Xfaxy is expected to be a major step forward in combating the rise of fake research and ensuring the reliability of academic publications.