Awesome GUI Agent Safety
A curated collection of research papers on GUI agent safety, organized by environment and keyword, covering benchmarks, datasets, attacks, defenses and evaluation frameworks for computer-use and mobile or desktop agents.
This repository collects research papers on the safety of GUI agents: systems that act inside web, mobile or desktop environments rather than only generating text. It groups papers by environment, such as web, mobile and desktop, and by keyword, such as evaluation, benchmark, risk, attack, injection, privacy, jailbreaking and multi-agent, alongside a full chronological list from most recent to oldest.
Each paper entry gives its title, authors, institutions, publication date and publisher, its environment and keyword tags, and a short summary of what it studies, for example a benchmark for evaluating harmful behavior in computer-use agents.
Contribution instructions describe how to add a new paper to the collection, and a keyword visualization section summarizes how many papers fall under each topic.