This repository contains all data and code to reproduce the results of our article "AI Safety for Everyone", published in Nature Machine Intelligence.

The data consists of annotations and RIS citations of papers retrieved during our systematic database search process and our snowballing process.

If you want a list and citation of all papers used in the analysis, click PAPERS.md.

If you would like to access the data itself used in our analyses, you can navigate to the following locations:

• data/export contains all retrieved non-duplicate publications in standard CSL JSON, RIS, and BibTex formats.
• data/annotations contains all selected publications with annotations as described in Section 2.2 of the paper in a JSON format.

To reproduce our analyses in the paper you will need Python 3.11 and pip installed on your computer. You need to run the following commands from the root directory of the repository:

pip install -r requirements.txt
python code/analysis.py

This will recreate all figures in the paper and put them into the output folder.

If you found our work useful and/or used it for your work, we would appreciate it if you cite our paper:

@article{gyevnar2025AISafety
    title     = {AI Safety for Everyone},
    author    = {Gyevnar, Balint and Kasirzadeh, Atoosa},
    journal   = {Nature Machine Intelligence},
    publisher = {Springer},
    address   = {New York, NY, USA},
    year      = {2025},
    doi       = {10.1038/s42256-025-01020-y}
}