Generative AI is making scientists in physics and chemistry more secretive about their latest discoveries to avoid being scooped.
April 24, 2026
Original Paper
A Market for Preprints: Knowledge Accessibility, Execution Leakage, and the Economics of Academic Disclosure in the Age of AI
SocArXiv · x24vh_v1
The Takeaway
AI's ability to quickly process and replicate research has triggered a deceleration in the use of preprints in certain scientific fields. While social scientists are sharing work faster, researchers in labs where ideas are easily copied by AI-assisted competitors are pulling back. The fear of execution leakage means that once an idea is public, an AI can help a rival lab finish the experiments first. This creates a structural break in how scientific knowledge is shared, ending the trend toward total transparency. Science is becoming a high-stakes race where the speed of AI forces researchers to hide their work until the very last second.
From the abstract
Research fields differ in knowledge accessibility — the degree to which a field’s questions and execution methods are legible to trained non-specialists. We develop a sequential game-theoretic model of preprint disclosure in which researcher A chooses whether to declare an idea; if she declares, researcher B learns the idea and executes with probability λ, the execution leakage rate. The declarer retains residual priority share α; the rival incurs execution cost ce. A single cutoff λc separates