AI
Researchers Launch "Project Rachel," an AI Author Pose to Test Academic Publishing
A deliberate experiment placed AI-authored papers under a fabricated scholarly identity to observe how academia's peer review system responds to artificial intelligence.
Researchers at a leading computer science institute have conducted a controlled experiment to expose the vulnerabilities of the academic publishing ecosystem. Dubbed Project Rachel, the investigation created a complete artificial scholarly identity—a fictional researcher named Rachel So—and systematically published AI-generated research papers to measure institutional responses.
The Experiment#
Between March and October 2025, the Rachel So account published more than 10 papers, leveraging contemporary large language models to generate the research content. The results were striking: the papers garnered citations, and the AI author received at least one invitation to serve as a peer reviewer for an academic journal. This suggests that AI-generated work can pass initial gatekeeping mechanisms in scholarly publishing without immediate detection.
What It Reveals#
The findings underscores critical gaps in how publishers, journals, and the broader academic community vet authorship and research quality. With increasingly capable AI systems now capable of generating plausible academic prose, the traditional assumption that human expertise stands behind submitted manuscripts has become fragile. The experiment provides concrete evidence that current detection and verification protocols remain inadequate to distinguish AI authorship from human scholarship at scale.
Broader Implications#
This action research study contributes empirical grounding to an urgent conversation about the future of science and scholarly communication. As AI systems grow more sophisticated, institutional frameworks for validating authorship, preventing plagiarism, and ensuring research integrity will need substantial revision. The work raises pressing questions for publishers regarding disclosure requirements, for researchers about professional reputation, and for the scientific system as a whole regarding the legitimacy of findings attributed to non-human entities.
The study was submitted to arXiv on November 18, 2025, and has since been revised multiple times, indicating ongoing refinement and community engagement with the findings.
