Archive — US — techcrunch
Researchers seek to influence peer review with hidden AI prompts
Researchers are reportedly embedding hidden AI prompts in their papers on arXiv to influence peer review outcomes, with 17 identified papers from institutions across multiple countries encouraging AI to provide positive feedback. This controversial tactic raises ethical questions, particularly as many conferences restrict AI use in reviews.
Key points
- 17 research papers from various academic institutions were found to include hidden AI prompts.
- The authors are affiliated with 14 institutions across 8 countries, including Japan, South Korea, and the U.S.
- Papers typically related to computer science, using brief, hidden instructions for positive AI feedback on their research.
- One defender of this practice argues it counters 'lazy reviewers' who rely on AI for evaluations.
Why it matters
- Concerns over integrity in peer review processes are increasingly relevant as AI tools become more integrated into academic evaluations.
- The use of AI in research and peer review reflects wider trends in 2025's IT sector related to automation and ethical guidelines for AI usage.
- Historically, the introduction of technology in academia has led to debates on ethics and integrity, seen in past controversies around plagiarism detection and automated grading.
The emergence of hidden AI prompts in research papers underscores the ethical dilemmas facing academia today regarding peer review integrity, especially as reliance on technology continues to grow.