How to Detect Fake Research: A Practical Guide for Scientists Combating Scientific Misconduct

As AI-generated text, deep fakes, and fake news continue to infiltrate the academic world, a group of research integrity experts has created a comprehensive toolkit to help scientists identify suspicious scientific papers and uphold the quality of scholarly work.

This initiative, known as the Collection of Open Science Integrity Guides (COSIG), brings together 27 openly accessible resources offering practical methods for identifying hallmarks of scientific misconduct, including image duplication, citation manipulation, plagiarism, AI-generated papers, and fake research.

“We want to arm scientists with the tools to uphold the integrity of the literature,” says Reese Richardson, a metascientist at Northwestern University and one of the project’s leads. The COSIG guides aim to empower researchers, whether early-career or seasoned, to detect unethical practices and flawed publications.

Tools and Techniques for Research Sleuths

From spotting citation techniques used to inflate impact to recognizing tortured phrases and formulaic language generated by paper mills, the COSIG toolkit covers it all. It includes guides tailored to specific disciplines such as biology, chemistry, computer science, and statistics.

With AI playing a growing role in both content creation and detection, tools like Feet of Clay Detector, Imagetwin, and Proofig are highlighted for their ability to flag problematic citations and identify AI-generated figures or manipulated images.

Other guides explain how to identify:

• Non-existent or contaminated cell lines in cancer research.
• Fabricated tumor sizes that exceed ethical guidelines.
• Misleading X-ray diffraction patterns.
• Confusion between atomic weight and count is a sign of sloppy or fake science.
  

Building a Community of Watchful Reviewers

Richardson emphasizes that “anybody can do it” — sleuthing doesn’t require superhuman attention to detail or endless hours online. The goal is to broaden participation in post-publication review and raise awareness about the threats of AI-generated research and fraudulent studies.

James Heathers, a forensic metascientist at Linnæus University, sees COSIG as a major step forward in the fight against research fraud. “This work has traditionally been done by people who had to essentially invent their own methods,” he notes. Now, the guide provides a standardized approach for ensuring research integrity.

The collection also includes advice for respectfully flagging concerns on PubPeer and how to effectively communicate with journal publishers when misconduct is suspected.

As AI continues to disrupt academia, COSIG represents an important line of defense one that promotes transparency, accountability, and responsible science. The team behind it encourages further contributions and aims to regularly update the guides.
In a time when fake science and manipulated research threaten public trust, COSIG offers hope for a more ethical, verifiable future in science.